Some options, such as --show-touches or --stay-awake, modify Android
settings and must be restored upon exit.
If scrcpy terminates (e.g. due to an early error) in the middle of the
clean up configuration, the device may be left in an inconsistent state
(some settings might be changed but not restored).
This issue can be reproduced with high probability by forcing scrcpy to
fail:
scrcpy --show-touches --video-encoder=fail
To prevent this problem, ensure that the clean up thread is not
interrupted until the clean up process is started.
Refs #5601 <https://github.com/Genymobile/scrcpy/issues/5601>
PR #5613 <https://github.com/Genymobile/scrcpy/pull/5613>
Add an option to disable the following flag for virtual displays:
DisplayManager.VIRTUAL_DISPLAY_FLAG_DESTROY_CONTENT_ON_REMOVAL
With this option, when the virtual display is closed, the running apps
are moved to the main display rather than being destroyed.
PR #5615 <https://github.com/Genymobile/scrcpy/pull/5615>
On --no-vd-system-decoration, the boolean option must be set to false.
It was wrongly assigned from optarg (this worked because optarg is NULL
at this point, so it was converted to false).
PR #5615 <https://github.com/Genymobile/scrcpy/pull/5615>
When running scrcpy with --tcpip=xx.xx.xx.xx, to make sure a new working
connection is established, it was first disconnected by a call to:
adb disconnect <addr>
However, this caused all running instances connected to that address to
be killed. Running several instances of scrcpy on the same device is now
useful with virtual displays, so change the default behavior to NOT
disconnect.
To force a reconnection, a '+' prefix can be added:
scrcpy --tcpip=+192.168.0.x
Fixes#5562 <https://github.com/Genymobile/scrcpy/issues/5562>
For non-Windows portable builds, use the absolute path to the adb
executable located in the same directory as scrcpy.
On Windows, just use "adb", which is sufficient to use the local one.
PR #5560 <https://github.com/Genymobile/scrcpy/pull/5560>
Contrary to getenv(), the result of sc_get_env() is encoded in UTF-8 on
all platforms. Since it is allocated, it requires an explicit init() and
destroy() functions.
PR #5560 <https://github.com/Genymobile/scrcpy/pull/5560>
Contrary to getenv(), sc_get_env() returns an allocated string that is
guaranteed to be encoded in UTF-8 on all platforms (it uses _wgetenv()
internally on Windows and converts the strings).
PR #5560 <https://github.com/Genymobile/scrcpy/pull/5560>
The position of touch and scroll must normally be "resolved" with a
"position mapper" associated to the display.
But to support the injection of such events with scrcpy-server alone
without video, handle the case where there is no display.
Fixes#5542 <https://github.com/Genymobile/scrcpy/issues/5542>
On Windows, interrupting a socket with shutdown() does not wake up
accept() or read() calls, the socket must be closed.
Introduce a new macro constant SC_SOCKET_CLOSE_ON_INTERRUPT, distinct of
_WIN32, because Windows will not be the only platform exhibiting this
behavior.
Refs #5536 <https://github.com/Genymobile/scrcpy/issues/5536>
The command sha256sum does not exist on macOS, but `shasum -a256` works
both on Linux and macOS.
PR #5515 <https://github.com/Genymobile/scrcpy/pull/5515>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Make dependencies build scripts more flexible, to accept a build type
(native or cross) and a link type (static or shared).
This lays the groundwork for building binaries for Linux and macOS.
PR #5515 <https://github.com/Genymobile/scrcpy/pull/5515>
Extract the code that processes arguments into a function.
This will make it optional, so the script that only downloads the
official ADB binaries will not use arguments.
PR #5515 <https://github.com/Genymobile/scrcpy/pull/5515>
Since commit 2687d202809dfaafe8f40f613aec131ad9501433, the Makefile
named release.mk stopped handling dependencies between recipes, because
they have to be executed separately (from different Github Actions
jobs).
Using a Makefile no longer provides any real benefit. Replace it by
several individual release scripts for simplicity and readability.
Refs #5306 <https://github.com/Genymobile/scrcpy/pull/5306>
PR #5515 <https://github.com/Genymobile/scrcpy/pull/5515>
Allow capturing virtual displays at a lower resolution using
-m/--max-size.
In the original implementation in #5370, the virtual display size was
necessarily the same as the capture size. The --max-size value was only
allowed to determine the virtual display size when no explicit size was
provided.
Since the dpi was scaled down accordingly, it is often better to create
a virtual display at the target capture size directly. However, not
everything is rendered according to the virtual display DPI. For
example, a page in Firefox is rendered too big on small virtual
displays. Thus, it makes sense to be able create a virtual display at a
given size, and capture it at a lower resolution with --max-size. This
is now possible using OpenGL filters.
Therefore, change the behavior of --max-size for virtual displays:
- --max-size does not impact --new-display without size argument
anymore (the virtual display size is the main display size);
- it is used to limit the capture size (whether an explicit size is
provided or not).
This new behavior is consistent with main display capture.
Refs #5370 comment <https://github.com/Genymobile/scrcpy/pull/5370#issuecomment-2438944401>
Refs #5370 <https://github.com/Genymobile/scrcpy/pull/5370>
PR #5506 <https://github.com/Genymobile/scrcpy/pull/5506>
Add an option to disable the following flag for virtual displays:
DisplayManager.VIRTUAL_DISPLAY_FLAG_SHOULD_SHOW_SYSTEM_DECORATIONS
Some devices render a broken UI when this flag is enabled.
Fixes#5494 <https://github.com/Genymobile/scrcpy/issues/5494>
If no size is provided with --new-display, the main display size is
used. But the actual size depended on the current device orientation.
To make it deterministic, use the size of the natural device orientation
(portrait for phones, landscape for tablets).
PR #5455 <https://github.com/Genymobile/scrcpy/pull/5455>
On rotation, it is expected that many successive events are ignored due
to size mismatch, when an event was generated from the mirroring window
having the old size, but was received on the device with the new size
(especially since mouse hover events are forwarded).
Do not flood the console with warnings.
PR #5455 <https://github.com/Genymobile/scrcpy/pull/5455>
Deprecate --lock-video-orientation in favor of a more general option
--capture-orientation, which supports all possible orientations
(0, 90, 180, 270, flip0, flip90, flip180, flip270), and a "locked" flag
via a '@' prefix.
All the old "locked video orientations" are supported:
- --lock-video-orientation -> --capture-orientation=@
- --lock-video-orientation=0 -> --capture-orientation=@0
- --lock-video-orientation=90 -> --capture-orientation=@90
- --lock-video-orientation=180 -> --capture-orientation=@180
- --lock-video-orientation=270 -> --capture-orientation=@270
In addition, --capture-orientation can rotate/flip the display without
locking, so that it follows the physical device rotation.
For example:
scrcpy --capture-orientation=flip90
always flips and rotates the capture by 90° clockwise.
The arguments are consistent with --display-orientation and
--record-orientation and --orientation (which provide separate
client-side orientation settings).
Refs #4011 <https://github.com/Genymobile/scrcpy/issues/4011>
PR #5455 <https://github.com/Genymobile/scrcpy/pull/5455>
On Android 14, DisplayListener may be broken (it never sends events).
This is fixed in recent Android 14 upgrades, but we can't really detect
it directly.
As a workaround, a RotationWatcher and DisplayFoldListener were
registered as a fallback, until a first "display changed" event was
triggered.
To simplify, on Android 14, register a DisplayWindowListener (introduced
in Android 11) to listen to configuration changes instead.
Refs #5455 comment <https://github.com/Genymobile/scrcpy/pull/5455#issuecomment-2481302084>
PR #5455 <https://github.com/Genymobile/scrcpy/pull/5455>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Detecting display size changes is not straightforward:
- from a DisplayListener, "display changed" events are received, but
this does not imply that the size has changed (it must be checked);
- on Android 14 (see e26bdb07a21493d096ea5c8cfd870fc5a3f015dc),
"display changed" events are not received on some versions, so as a
fallback, a RotationWatcher and a DisplayFoldListener are registered,
but unregistered as soon as a "display changed" event is actually
received, which means that the problem is fixed.
Extract a "display size monitor" to share the code between screen
capture and virtual display capture.
PR #5455 <https://github.com/Genymobile/scrcpy/pull/5455>
Expose two methods on Size directly:
- limit() to downscale a size;
- round8() to round both dimensions to multiples of 8.
This will allow removing ScreenInfo completely.
PR #5455 <https://github.com/Genymobile/scrcpy/pull/5455>
Introduce several key components to perform OpenGL filters:
- OpenGLRunner: a tool for running a filter to be rendered to a Surface
from an OpenGL-dedicated thread
- OpenGLFilter: a simple OpenGL filter API
- AffineOpenGLFilter: a generic OpenGL implementation to apply any 2D
affine transform
- AffineMatrix: an affine transform matrix, with helpers to build
matrices from semantic transformations (rotate, scale, translate…)
PR #5455 <https://github.com/Genymobile/scrcpy/pull/5455>
If a new display is set, force options.getDisplayId() to return
Device.DISPLAY_ID_NONE, to avoid any confusion between a local displayId
and options.getDisplayId().
All options were configured dynamically by sending a single byte to an
output stream. But in practice, only the power mode must be changed
dynamically, the others are configured once on start.
For simplicity, pass the value of static options as command line
arguments, and handle dynamic options in a loop only from a separate
thread once the clean up process is started.
This will allow to easily add cleanup options with values which do not
fit in 1 byte.
Also handle the clean up thread (and the loading of initial settings
values) from the CleanUp class, to expose a simpler clean up API.
Refs 9efa162949c2a3e3e42564862ff390700270394d
PR #5447 <https://github.com/Genymobile/scrcpy/pull/5447>
A call to swr_set_compensation() configures the resampler to drop or
duplicate "diff" samples over an interval of "distance" samples.
If the function is not called again, then after "distance" samples, no
more compensation will be applied. So it must always be called, even if
the new computed diff value happens to be the same as the previous one.
In practice, it is unlikely that the diff value is exactly the same
every second, except when it is actively clamped (to 2% of the sample
rate).
Reset video capture/encoding on MOD+Shift+r.
Like on device rotation, this starts a new encoding session which
produces a video stream starting by a key frame.
PR #5432 <https://github.com/Genymobile/scrcpy/pull/5432>
When the MediaCodec input is a Surface, no EOS (end-of-stream) will
never occur automatically: it may only be triggered manually by
MediaCodec.signalEndOfInputStream().
Use this signal to interrupt the blocking call to dequeueOutputBuffer()
immediately on reset, without waiting for the next frame to be dequeued.
PR #5432 <https://github.com/Genymobile/scrcpy/pull/5432>
When the capture source becomes "invalid" (because the display size
changes for example), a reset request is performed to restart the
encoder.
The reset state was stored in SurfaceCapture. The capture implementation
set the flag, and the encoder consumed it.
However, this mechanism did not allow a reset request to _interrupt_ the
encoder, which may be waiting on a blocking call (until a new frame is
produced).
To be able to interrupt the encoder, a reset request must not only set a
flag, but run a callback provided by the encoder. For that purpose,
introduce the CaptureListener interface, which is notified by the
SurfaceCapture implementation whenever the capture is invalidated.
For now, the listener implementation just set a flag as before, so the
behavior is unchanged. It lays the groundwork for the next commits.
PR #5432 <https://github.com/Genymobile/scrcpy/pull/5432>
While moving code, commit 874eaec487369f7fcaa9ed8c5f85569659565d4f added
a condition `if (displayId == 0)` to register a rotation watcher,
without good reasons.
This condition was kept when the rotation watcher was moved to a
fallback in e26bdb07a21493d096ea5c8cfd870fc5a3f015dc.
Note: use `git show -b` to show this commit ignoring whitespace changes.
Refs #5428 <https://github.com/Genymobile/scrcpy/issues/5428>
Accept a single boolean "on" rather than a "mode" (which, in practice,
could only take 2 values: NORMAL and OFF).
Also rename "screen power mode" to "display power".
PR #5418 <https://github.com/Genymobile/scrcpy/pull/5418>
If video playback is disabled, then SDK mouse (which uses absolute
positions) could not be used, so the default mouse mode was
automatically switched to UHID.
But UHID does not work on all devices, so it could make the whole scrcpy
session fail.
Instead, disable the mouse by default. It is still possible to pass -M
or --mouse=uhid to enable it explicitly.
Fixes#5410 <https://github.com/Genymobile/scrcpy/issues/5410>
By adding the '?' prefix, the app is searched by its name instead of its
package name (retrieving app names on the device may take some time):
scrcpy --start-app=?firefox
An app matches if its label starts with the given name,
case-insensitive.
If '+' is also passed to force-stop the app before starting, then the
prefixes must be in that order:
scrcpy --start-app=+?firefox
PR #5370 <https://github.com/Genymobile/scrcpy/pull/5370>
The previous commit introduced:
scrcpy --start-app=name
By adding a '+' prefix, the app is stopped beforehand:
scrcpy --start-app=+name
This may be useful to start a fresh app on a new virtual display:
scrcpy --new-display --start-app=+org.mozilla.firefox
PR #5370 <https://github.com/Genymobile/scrcpy/pull/5370>
Add a command line option --start-app=name to start an Android app by
its package name. For example:
scrcpy --start-app=org.mozilla.firefox
The app will be started on the correct target display:
scrcpy --new-display=1920x1080 --start-app=org.videolan.vlc
PR #5370 <https://github.com/Genymobile/scrcpy/pull/5370>
Co-authored-by: Simon Chan <1330321+yume-chan@users.noreply.github.com>
Add a feature to create a new (separate) virtual display instead of
mirroring the device screen:
scrcpy --new-display=1920x1080
scrcpy --new-display=1920x1080/420 # force 420 dpi
scrcpy --new-display # use the main display size and density
scrcpy --new-display -m1920 # scaled to fit a max size of 1920
scrcpy --new-display=/240 # use the main display size and 240 dpi
Fixes#1887 <https://github.com/Genymobile/scrcpy/issues/1887>
PR #5370 <https://github.com/Genymobile/scrcpy/pull/5370>
Co-authored-by: Simon Chan <1330321+yume-chan@users.noreply.github.com>
Co-authored-by: anirudhb <anirudhb@users.noreply.github.com>
When a new capture starts, send a new PositionMapper to the Controller
without using the global Device as an intermediate.
Now all Device methods are static.
PR #5370 <https://github.com/Genymobile/scrcpy/pull/5370>
Extract the function that converts coordinates from video space to
display space into a separate component.
It only requires the specific data it uses and does not need a full
ScreenInfo object (although it can be created from a ScreenInfo
instance).
PR #5370 <https://github.com/Genymobile/scrcpy/pull/5370>
Remove from Device the functions using an implicit displayId.
Move them to Controller, which knows best which displayId it must use.
This will allow to properly dispatch events either to the origin display
or to the virtual display created for mirroring.
PR #5370 <https://github.com/Genymobile/scrcpy/pull/5370>
Add a function called before each capture starts (before getSize() is
called).
This allows to compute the ScreenInfo instance once exactly when needed.
PR #5370 <https://github.com/Genymobile/scrcpy/pull/5370>
Move the code related to screen size and rotation/fold to ScreenCapture.
For now, keep the ScreenInfo instance in the Device class to communicate
with the Controller, but it will be removed by further commits.
PR #5370 <https://github.com/Genymobile/scrcpy/pull/5370>
Make it possible to build scrcpy-server and Windows binaries in
parallel from different GitHub Actions workflows, and to package
everything as a final step.
PR #5306 <https://github.com/Genymobile/scrcpy/pull/5306>
Scrcpy requires Android 5.0+, so there is no point testing for older
versions.
Btw, there were two mistakes:
- the constant name in AndroidVersions should have been
API_20_ANDROID_4_4W (Android 4.4 without 'W' is API 19)
- the method isInteractive() was introduced in Android 5.0, not 4.4W:
<037c33eae7%5E%21/>
Also adapt the call to build-tools/35.0.0/aidl, which now requires an
import path (-I. for the current directory).
Otherwise, it fails with:
ERROR: android/view/IRotationWatcher.aidl:23.1-10: directory ./ is not found in any of the import paths:
-
Also upgrade AGP (8.7.1) and Gradle (8.9), required for SDK 35.
<https://developer.android.com/build/releases/gradle-plugin#updating-gradle>
The server params were passed from the main thread to the server thread,
so a deep copy was performed in case the caller instance was destroyed.
But in practice, it only contains memory that lives until the end of the
program (command line arguments), so simply reference it. Several copies
of string fields were missing anyway.
With the old macros definitions, the type of the result depended on the
type of `sec`.
In particular, if sec is a 32-bit type, sec * 1000000 was likely to
overflow (even if the result was assigned to a sc_tick by the caller of
the macro).
This was the case on Windows, where the long type is a 32-bit signed
integer: the --time-limit argument, expressed in seconds, was first
parsed to a long value, then multiplied by 1000000 by the
SC_TICK_FROM_SEC() macro, causing an overflow when the value was greater
than 2147 (2^31 / 1000000).
Fixes#5355 <https://github.com/Genymobile/scrcpy/issues/5355>
Os.write() did not update the ByteBuffer position before Android 6.
A workaround was added by commit
b882322f7371b16acd53677c4a3adbaaed0aef77, which fixed part of the
problem, but the position was still not updated across calls, causing
the wrong chunk to be written.
Refs <d9f7e57f5d%5E%21/>
Replace argument suggestion:
--video-encoder='c2.android.avc.encoder'
by:
--video-encoder=c2.android.avc.encoder
On Linux, the quotes are interpreted by the shell, but on Windows they
are passed as is.
This was harmless, because even transmitted as is, they were interpreted
by the shell on the device. However, special characters are now
validated since commit bec3321fff4c6dc3b3dbc61fdc6fd98913988a78, making
the command fail.
Fixes#5329 <https://github.com/Genymobile/scrcpy/issues/5329>
The audio player had 2 roles:
- handle the SDL audio output device;
- resample input samples to maintain a target latency.
Extract the latter to a separate component (an "audio regulator"),
independent of SDL.
Moving classes into subpackages changed the expected imports order.
Reorganize them all at once automatically to avoid spurious changes in
future commits.
Capture the gamepads even when the window is not focused.
Note: In theory, with this flag set, we could capture gamepad events
even without a window (--no-window). In practice, scrcpy still requires
a window, because --no-window implies --no-control, and the input
manager is owned by the sc_screen instance, which does not exist if
there is no window. Supporting this use case would require a lot of
refactors.
Refs <https://github.com/Genymobile/scrcpy/pull/5270#issuecomment-2339360460>
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
Suggested-by: Luiz Henrique Laurini <luizhenriquelaurini@gmail.com>
The sc_uhid_devices instance is initialized only when there is a UHID
keyboard.
The device message receiver assumed that it could not receive HID output
reports without a sc_uhid_devices instance (i.e. without a UHID
keyboard), but in practice, a UHID driver implementation on the device
may decide to send UHID output reports for mouse or for gamepads (and we
must just ignore them).
So remove the assert().
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
There was a registration mechanism to listen to HID outputs with a
specific HID id.
However, the UHID gamepad processor handles several ids, so it cannot
work. We could complexify the registration mechanism, but instead,
directly dispatch to the expected processor based on the UHID id.
Concretely, instead of passing a sc_uhid_devices instance to construct a
sc_keyboard_uhid, so that it can register itself, construct the
sc_uhid_devices with all the UHID instances (currently only
sc_keyboard_uhid) so that it can dispatch HID outputs directly.
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
Initialize UHID devices with a custom name:
- "scrcpy: $GAMEPAD_NAME" for gamepads
- "scrcpy" for keyboard and mouse (or if no gamepad name is available)
The name may appear in Android apps.
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
For convenience, short options were added to select UHID input modes:
- -K for --keyboard=uhid
- -M for --mouse=uhid
- -G for --gamepad=uhid
In OTG mode, UHID is not available, so the short options should select
AOA instead.
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
Similar to UHID keyboard and mouse, but for gamepads.
Can be enabled with --gamepad=uhid or -G.
It is not enabled by default because not all devices support UHID
(there is a permission error on old Android versions).
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
This message will be sent on gamepad disconnection.
Contrary to keyboard and mouse devices, which are registered once and
unregistered when scrcpy exists, each physical gamepad is mapped with
its own HID id, and they can be plugged and unplugged dynamically.
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
Trigger SDL_CONTROLLERDEVICEADDED for all gamepads already connected
when scrcpy starts. We want to handle both the gamepads initially
connected and the gamepads connected while scrcpy is running.
This is not racy, because this event may not be trigged automatically
until SDL events are "pumped" (SDL_PumpEvents/SDL_WaitEvent).
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
Introduce a gamepad processor trait, similar to the keyboard processor
and mouse processor traits.
Handle gamepad events received from SDL, convert them to scrcpy-specific
gamepad events, and forward them to the gamepad processor.
Further commits will provide AOA and UHID implementations of the gamepad
processor trait.
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
Co-authored-by: Luiz Henrique Laurini <luizhenriquelaurini@gmail.com>
Mouse and keyboard events with unknown button/keycode/scancode cannot be
handled properly. Discard them without forwarding them to the
keyboard or mouse processors.
This can happen for example if a more recent version of SDL introduces
new enum values.
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
Now that the AOA open/close are asynchronous, an open error did not make
scrcpy exit anymore.
Add a mechanism to exit if the AOA device could not be opened
asynchronously.
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
Pushing a close event from the keyboard_aoa or mouse_aoa implementation
was racy, because the AOA thread might be stopped before these events
were processed.
Instead, keep the list of open AOA devices to close them automatically
from the AOA thread before exiting.
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
This allows to handle HID open/close reports at the same place as HID
input reports (in the HID layer).
This will be especially useful to manage HID gamepads, to avoid
implementing one part in the HID layer and another part in the gamepad
processor implementation.
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
For AOA keyboard and mouse, only input reports were asynchronous.
Register/unregister were called from the main thread.
This had the benefit to fail immediately if the AOA registration failed,
but we want to open/close AOA devices dynamically in order to add
gamepad support.
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
The HID ids (accessory ids or UHID ids) were defined by the keyboard and
mouse implementations.
Instead, define them in the common HID part, and make that id part of
the sc_hid_event.
This prepares the introduction of gamepad support, which will handle
several gamepads (and ids) in the common HID gamepad code.
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
Control messages are queued from the main thread and sent to the device
from a separate thread.
When the queue is full, messages are just dropped. This avoids to
accumulate too much delay between the client and the device in case of
network issue.
However, some messages should not be dropped: for example, dropping a
UHID_CREATE message would make all further UHID_INPUT messages invalid.
Therefore, mark these messages as non-droppable.
A non-droppable event is queued anyway (resizing the queue if
necessary, unless the allocation fails).
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
This allows to schedule a runnable to be executed on the main thread,
until the event loop is explicitly terminated.
It is guaranteed that all accepted runnables will be executed (this
avoids possible memory leaks if a runnable owns resources).
PR #5270 <https://github.com/Genymobile/scrcpy/pull/5270>
It is better to disable Nagle's algorithm to avoid unnecessary latency
for control messages. (I'm not sure this has any impact for a local TCP
socket though.)
Many parsing and formatting C functions like strtof() and asprintf() are
locale-dependent. Forcing a C locale just for the conversions in a way
that works on all platforms is a mess.
In practice, this is not a problem, scrcpy always uses the C locale,
because it never calls:
setlocale(LC_ALL, "");
But the max-fps option should not depend on the locale configuration
anyway.
Since the value is parsed by the client in Java anyway, just forward the
string value as is.
Android accepts a float value, there is no reason to limit the option
to be an integer.
In particular, it allows to capture at a rate lower than 1 fps. For
example, to capture 1 frame every 5 seconds:
scrcpy --video-source=camera --max-fps=0.2
It was already possible to pass a float manually:
scrcpy --video-source=camera \
--video-codec-options=max-fps-to-encoder:float=0.2
But accepting a float directly for --max-fps is more convenient.
Refs <https://developer.android.com/reference/android/media/MediaFormat#KEY_MAX_FPS_TO_ENCODER>
A video width or height of 0 triggered an assert.
Fail explicitly instead: the server may actually send this size in
practice (for example on cropping with small dimensions, even if the
requested crop size is not 0).
The delay buffer clock estimates the clock offset between the PTS and
the frame decoded date using an "Exponentially Weighted Moving Average"
(EWMA).
But for the first frames, the clock have less than SC_CLOCK_RANGE
points to average. Since the timing for the first frames are typically
the worst ones, give more weight to the last point for the estimation.
Once SC_CLOCK_RANGE points are available (i.e. when SC_CLOCK_RANGE ==
clock->range), the new estimation is equivalent to the previous version.
The delay buffer `stopped` field was not initialized.
Since it practice the unique instance of sc_delay_buffer is initialized
in static memory, the flag was initialized to false as a side effect.
But with commit fd0f432e877153d83ed435474fb7b04e41de4269, in debug mode
only, the delay buffer was broken.
"Could not" implies that the system tried to disable the option but
encountered an issue or failure.
"Cannot" indicates a rule or restriction, meaning it's not possible to
perform the action at all.
In Java, control messages were parsed using manual buffering, which was
convoluted and error-prone.
Instead, read the socket directly through a DataInputStream and a
BufferedInputStream. Symmetrically, use a DataOutputStream and a
BufferedOutputStream to write messages.
Even if the pointer is a mouse, inject it as a finger unless it is
required to be a mouse, that is:
- when it is a HOVER_MOUSE event, or
- when a secondary button is pressed.
Some apps/games only accept events from a finger/touchscreen, so using a
mouse by default does not work for them.
For simplicity, make this change on the server side just before
event injection (so that the client does not need to know about this
hacky behavior).
Refs 6808288823239b0f3a76f9be377e4de82e91b35a
Refs c7b1d0ea9af8bb9603ec8f713f1a5fbf3f628b6a
Fixes#5162 <https://github.com/Genymobile/scrcpy/issues/5162>
Fixes#5163 <https://github.com/Genymobile/scrcpy/issues/5163>
By default, the audio source is initialized to SC_AUDIO_SOURCE_AUTO, and
is "resolved" only if audio is enabled.
But the server arguments were built assuming that the audio source was
never SC_AUDIO_SOURCE_AUTO (even with audio disabled), causing a crash.
Regression introduced by a10f8cd798023f858796b023cb846fa2184ad2c7.
Automatically switch implicit audio source to "playback" if --audio-dup
is passed.
This allows to run:
scrcpy --audio-dup
without specifying explicitly:
scrcpy --audio-source=playback --audio-dup
PR #5102 <https://github.com/Genymobile/scrcpy/pull/5102>
Add a new method to capture audio playback.
It requires Android 13 (where the Shell app has MODIFY_AUDIO_ROUTING
permission).
The main benefit is that it supports keeping audio playing on the device
(implemented in a further commit).
Fixes#4380 <https://github.com/Genymobile/scrcpy/issues/4380>
PR #5102 <https://github.com/Genymobile/scrcpy/pull/5102>
Co-authored-by: Simon Chan <1330321+yume-chan@users.noreply.github.com>
The MediaRecorder constant should not belong to the AudioSource enum.
This will allow to add a new AudioSource which has no meaningful
MediaRecorder audio source value.
PR #5102 <https://github.com/Genymobile/scrcpy/pull/5102>
Move the implementation to AudioDirectCapture and extract an
AudioCapture interface.
This will allow to provide another AudioCapture implementation.
PR #5102 <https://github.com/Genymobile/scrcpy/pull/5102>
The compatibility depends on the capture constraints, not the encoding.
This will allow to add a new capture implementation with different
constraints.
PR #5102 <https://github.com/Genymobile/scrcpy/pull/5102>
The install_release.sh script is updated one commit after the release
tag, which may be confusing.
For convenience, new lightweight tags have been added (for example
v2.5-install-release) to point to the commit where install_release.sh is
updated.
But these tags interfere with "git describe" to generate pretty
filenames when executing ./release.sh on a development branch, so ignore
them.
Before:
release-v2.5-install-release-17-gc57a0512b
After:
release-v2.5-18-gc57a0512b
Refs #4098 comment <https://github.com/Genymobile/scrcpy/issues/4098#issuecomment-1600332180>
Passing an unknown enum value to convert them to string would return
NULL without any error, possibly causing undefined behavior later.
Add assertions to catch such programming errors early.
Add secondary bindings (Shift+click) for mouse buttons.
In addition to:
--mouse-bind=xxxx
It is now possible to pass a sequence of secondary bindings:
--mouse-bind=xxxx:xxxx
<--> <-->
primary secondary
bindings bindings
If the second sequence is omitted, then it is the same as the first one.
By default, for SDK mouse, primary bindings trigger shortcuts and
secondary bindings forward all clicks.
For AOA and UHID, the default bindings are reversed: all clicks are
forwarded by default, whereas pressing Shift+click trigger shortcuts.
--mouse-bind=bhsn:++++ # default for SDK
--mouse-bind=++++:bhsn # default for AOA and UHID
Refs 035d60cf5d3f4c83d48735b4cb4cd108a5b5f413
Refs f5e6b8092afd82bab402e7c2c3d00b1719f9bb57
Fixes#5055 <https://github.com/Genymobile/scrcpy/issues/5055>
PR #5076 <https://github.com/Genymobile/scrcpy/pull/5076>
The buttons state was tracked by SDL_GetMouseState(), and scrcpy applied
a mask to ignore buttons used for shortcuts.
Instead, track the buttons actually pressed (ignoring shortcuts)
manually, to prepare the introduction of more dynamic mouse shortcuts.
PR #5076 <https://github.com/Genymobile/scrcpy/pull/5076>
For pinch-to-zoom, rotation and tilt simulation, always use a finger
source (instead of a mouse) for both pointers (the real one and the
simulated one).
A "virtual" mouse does not work on all devices (e.g. on Pixel 8).
PR #5076 <https://github.com/Genymobile/scrcpy/pull/5076>
A device disconnection (when the adb connection is closed) makes the
read() on the "receiver" socket fail.
Since commit 063a8339ed27b94a8fe1e53a284507eb2d044e15, this is reported
as an error. As a consequence, scrcpy fails with:
ERROR: Controller error
instead of:
WARN: Device disconnected
To fix the issue, report a device disconnection in that case.
PR #5044 <https://github.com/Genymobile/scrcpy/pull/5044>
This also avoids a warning with some compilers which do not understand
that the condition to initialize the variable is the same as the
condition to use it:
../app/src/scrcpy.c: In function ‘scrcpy’:
../app/src/scrcpy.c:750:13: warning: ‘src’ may be used uninitialized in this function [-Wmaybe-uninitialized]
750 | sc_frame_source_add_sink(src, &s->screen.frame_sink);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Refs 45fe6b602b4c050c5b1fba87cec7160093052af3
Refs <https://github.com/Genymobile/scrcpy/issues/5045#issuecomment-2201589757>
The documentation of avcodec_close() says:
> Do not use this function. Use avcodec_free_context() to destroy a
> codec context (either open or closed).
It was deprecated in FFmpeg 7 by commit
1cc24d749569a42510399a29b034f7a77bdec34e:
<1cc24d7495>
> Its use has been discouraged since 2016, but now is no longer used in
> avformat, so there is no reason to keep it public.
By default, only the left click is forwarded to the device, and
secondary clicks trigger shortcuts (the behavior can be configured by
--mouse-bind=xxxx).
But when the mouse mode is relative (AOA and UHID modes), forward all
clicks by default. This makes more sense since the cursor is handled on
the device side, the user expects all mouse buttons to be forwarded.
Refs <https://github.com/Genymobile/scrcpy/issues/4727#issuecomment-2069869750>
PR #5022 <https://github.com/Genymobile/scrcpy/pull/5022>
Add a new option --mouse-bind=xxxx.
The argument must be exactly 4 characters, one for each secondary click:
--mouse-bind=xxxx
^^^^
||||
||| `- 5th click
|| `-- 4th click
| `--- middle click
`---- right click
Each character must be one of the following:
- `+`: forward the click to the device
- `-`: ignore the click
- `b`: trigger shortcut BACK (or turn screen on if off)
- `h`: trigger shortcut HOME
- `s`: trigger shortcut APP_SWITCH
- `n`: trigger shortcut "expand notification panel"
This deprecates --forward-all-clicks (use --mouse-bind=++++ instead).
Refs <https://github.com/Genymobile/scrcpy/pull/2258#issuecomment-2182394460>
PR #5022 <https://github.com/Genymobile/scrcpy/pull/5022>
To resize the window to fit the device screen, it is possible to
double-click in the "black bars".
This feature was mistakenly disabled when --forward-all-clicks was set.
Instead, disable it only if mouse relative mode is enabled (AOA or
UHID), because in that case the mouse cursor is on the device.
Restrict shortcut modifiers to be composed of only one item each.
Before, it was possible to select a list of multiple combinations of
modifier keys, like --shortcut-mod='lctrl+lalt,rctrl+rsuper', meaning
that shortcuts would be triggered either by LCtrl+LAlt+key or
RCtrl+RSuper+key.
This was overly generic, probably not used very much, and it prevents to
solve inconsistencies between UP and DOWN events of modifier keys sent
to the device.
Refs #4732 <https://github.com/Genymobile/scrcpy/issues/4732>
PR #4741 <https://github.com/Genymobile/scrcpy/pull/4741>
Old Android versions may not be able to register a rotation watcher for
a secondary display. In that case, report the error instead of
registering a rotation watcher for the default display.
Refs <https://github.com/Genymobile/scrcpy/pull/4740#issuecomment-2051245633>
Suggested by: Kaiming Hu <huxxx1234@gmail.com>
The version of the methods with the display id parameter must be tried
first, otherwise they will never be used (since the old versions without
the display id are still present).
Regression introduced by ee6620d123e87d4af8e51cd272de5eafb677122a.
Refs #4740 <https://github.com/Genymobile/scrcpy/pull/4740>
PR #4841 <https://github.com/Genymobile/scrcpy/pull/4841>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Changed since AOSP/framework_base commit
670fb7f5c0d23cf51ead25538bcb017e03ed73ac, included in tag
android-14.0.0_r29.
Refs <670fb7f5c0%5E%21/>
PR #4740 <https://github.com/Genymobile/scrcpy/pull/4740>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Do not call sc_screen_hide_window() if screen is not initialized.
To reproduce:
scrcpy --no-video --record=file.mp4
This only segfaults in debug mode since commit
fd0f432e877153d83ed435474fb7b04e41de4269.
The project has 3 build dependencies:
- SDL
- FFmpeg
- libusb
For Windows, the release script downloaded pre-built build dependencies
(either from upstream, or from the scrcpy-deps repository).
Instead, download the source releases and build locally. This offers
more flexibility.
The official adb release is still downloaded and included as is in the
release archive (it is not a build dependency).
Also upgrade FFmpeg to 6.1.1 and libusb to 1.0.27.
PR #4713 <https://github.com/Genymobile/scrcpy/pull/4713>
Initially, if AOA initialization failed, default injection method was
used, in order to use the same command/shortcut when the device is
connected via USB or via TCP/IP, without changing the arguments.
Now that there are 3 keyboard modes, it seems unexpected to switch to
another specific mode if AOA fails (and it is inconsistent). If the user
explicitly requests AOA, then use AOA or fail.
Refs #2632 comment <https://github.com/Genymobile/scrcpy/pull/2632#issuecomment-945190859>
PR #4473 <https://github.com/Genymobile/scrcpy/pull/4473>
The options were deprecated, but for convenience, reassign them to
aliases for --keyboard=uhid and --mouse=uhid respectively.
Their long version (--hid-keyboard and --hid-mouse) remain deprecated.
PR #4473 <https://github.com/Genymobile/scrcpy/pull/4473>
The keyboard settings can be opened by:
adb shell am start -a android.settings.HARD_KEYBOARD_SETTINGS
Add a shortcut (MOD+k) for convenience if the current keyboard is HID.
PR #4473 <https://github.com/Genymobile/scrcpy/pull/4473>
On close, the controller is expected to throw an IOException because the
socket is closed, so the exception was ignored.
However, message handling actions may also throw IOException, and they
must not be silently ignored.
PR #4473 <https://github.com/Genymobile/scrcpy/pull/4473>
There is a dependency cycle in the initialization order:
- keyboard depends on controller
- controller depends on acksync
- acksync depends on keyboard initialization
To break this cycle, bind the async instance to the controller in a
second step.
PR #4473 <https://github.com/Genymobile/scrcpy/pull/4473>
Rename {keyboard,mouse}_inject to {keyboard,mouse}_sdk.
All implementations "inject" key events and mouse events, what differs
is the mechanism. For these implementations, the Android SDK API is used
to inject events.
Note that the input mode enum variants were already renamed
(SC_KEYBOARD_INPUT_MODE_SDK and SC_MOUSE_INPUT_MODE_SDK).
PR #4473 <https://github.com/Genymobile/scrcpy/pull/4473>
An event contained several fields:
- the accessory id
- the HID event data
- a field ack_to_wait specific to the AOA implementation.
Extract the HID event part to prepare the factorization of HID event
creation.
PR #4473 <https://github.com/Genymobile/scrcpy/pull/4473>
Until now, there was two modes for keyboard and mouse:
- event injection using the Android system API (default)
- HID/AOA over USB
For this reason, the options were exposed as simple flags:
- -K or --hid-keyboard to enable physical keyboard simulation (AOA)
- -M or --hid-mouse to enable physical mouse simulation (AOA)
Replace them by explicit --keyboard and --mouse options, with 3 possible
values:
- disabled
- sdk (default)
- aoa
This will allow to add a new mode (uhid).
PR #4473 <https://github.com/Genymobile/scrcpy/pull/4473>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The input manager assumed that if a controller was present, then both a
key processor and a mouse processor were present.
Remove this assumption, to support disabling keyboard and mouse
separately. This prepares the introduction of new command line options
--keyboard and --mouse.
PR #4473 <https://github.com/Genymobile/scrcpy/pull/4473>
Some functions in input_manager.c only have access to a sub-object (for
example the controller). For consistency, always pass the whole
input manager instance.
This will allow to add assertions when keyboard and mouse could be
disabled separately.
PR #4473 <https://github.com/Genymobile/scrcpy/pull/4473>
The assumption that underflow and overbuffering are caused by jitter
(and that the delay between the producer and consumer will be caught up)
does not always hold.
For example, if the consumer does not consume at the expected rate (the
SDL callback is not called often enough, which is an audio output
issue), many samples will be dropped due to overbuffering, decreasing
the average buffering indefinitely.
Prevent the average buffering to become negative to limit the
consequences of an unexpected behavior.
PR #4572 <https://github.com/Genymobile/scrcpy/pull/4572>
The buffering level does not change continuously: it increases abruptly
when a packet is received, and decreases abruptly when an audio block is
consumed.
To estimate the buffering level, a rolling average is used.
To make the buffering more stable, increase the smoothness of this
rolling average. This decreases the risk of enabling audio compensation
due to an estimation error.
PR #4572 <https://github.com/Genymobile/scrcpy/pull/4572>
Use different thresholds for enabling and disabling compensation.
Concretely, enable compensation if the difference between the average
and the target buffering levels exceeds 4 ms (instead of 1 ms). This
avoids unnecessary compensation due to small noise in buffering level
estimation.
But keep a smaller threshold (1 ms) for disabling compensation, so that
the buffering level is restored closer to the target value. This avoids
to keep the actual level close to the compensation threshold.
PR #4572 <https://github.com/Genymobile/scrcpy/pull/4572>
The audio output thread only reads samples from the buffer, and most of
the time, the audio receiver thread only writes samples to the buffer.
In these cases, using atomics avoids lock contention.
There are still corner cases where the audio receiver thread needs to
"read" samples (and drop them), so lock only in these cases.
PR #4572 <https://github.com/Genymobile/scrcpy/pull/4572>
Some actions may be performed when scrcpy exits, currently:
- disable "show touches"
- restore "stay on while plugged in"
- power off screen
- restore "power mode" (to disable "turn screen off")
They are performed from a separate process so that they can be executed
even when scrcpy-server is killed (e.g. if the device is unplugged).
The clean up actions to perform were configured when scrcpy started.
Given that there is no method to read the current "power mode" in
Android, and that "turn screen off" can be applied at any time using an
scrcpy shortcut, there was no way to determine if "power mode" had to be
restored on exit. Therefore, it was always restored to "normal", even
when not necessary.
However, setting the "power mode" is quite fragile on some devices, and
may cause some issues, so it is preferable to call it only when
necessary (when "turn screen off" has actually been called).
For that purpose, make the scrcpy-server main process and the clean up
process communicate the actions to perform over a pipe (stdin/stdout),
so that they can be changed dynamically. In particular, when the power
mode is changed at runtime, notify the clean up process.
Refs 1beec99f8283713b1fbf0b3704eb4dceecc9a590
Refs #4456 <https://github.com/Genymobile/scrcpy/issues/4456>
Refs #4624 <https://github.com/Genymobile/scrcpy/issues/4624>
PR #4649 <https://github.com/Genymobile/scrcpy/pull/4649>
This exception is a super-type of:
- ClassNotFoundException
- IllegalAccessException
- InstantiationException
- InvocationTargetException
- NoSuchFieldException
- NoSuchMethodException
Use it to simplify.
Create the service managers from each manager wrapper class rather than
from their getter in ServiceManager.
The way a wrapper retrieve the underlying service is an implementation
detail, and it must be consistent with the way it accesses it, so it is
better to write the creation in the wrapper.
Power on the device on start only if video capture is enabled.
Note that it only impacts display mirroring, since control is completely
disabled if video source is camera.
Refs 110b3a16f6d02124a4567d2ab79fcb74d78f949f
If the initial timestamp could not be retrieved, use the current time as
returned by System.nanoTime(). In practice, it is the same time base as
AudioRecord timestamps.
Fixes#4536 <https://github.com/Genymobile/scrcpy/issues/4536>
The result is assigned to a long (64-bit signed integer), but the
intermediate multiplication was stored in an int (32-bit signed
integer).
This value is only used as a fallback when no timestamp could be
retrieved, that's why it did not cause too much harm so far.
Fixes#4536 <https://github.com/Genymobile/scrcpy/issues/4536>
Reported by checkstyle:
> [ant:checkstyle] [INFO]
> scrcpy/server/src/main/java/com/genymobile/scrcpy/wrappers/ClipboardManager.java:48:
> Line is longer than 150 characters (found 167). [LineLength]
For AV1, the config packet must not be merged with the next non-config
packet.
This fixes the following error when passing --video-codec=av1:
> INFO: [FFmpeg] libdav1d 1.3.0
> ERROR: [FFmpeg] Unknown OBU type 0 of size 29393
> ERROR: [FFmpeg] Error parsing OBU data
> ERROR: Decoder 'video': could not send video packet: -1094995529
PR #4487 <https://github.com/Genymobile/scrcpy/pull/4487>
The SDL video subsystem may be initialized so that clipboard
synchronization works even without video playback.
But if the video subsystem initialization fails (e.g. because no video
device is available), consider it as an error only if video playback is
enabled.
Refs 5e59ed31352251791679e5931d7e5abf0c2d18f6
Fixes#4477 <https://github.com/Genymobile/scrcpy/issues/4477>
On some Samsung devices, DisplayManagerGlobal.getDisplayInfoLocked()
calls ActivityThread.currentActivityThread().getConfiguration(), which
requires a non-null ConfigurationController.
Fixes <https://github.com/Genymobile/scrcpy/issues/4467>
Fix the following warning/error:
../app/src/cli.c:2158:17: warning: a label can only be part of a
statement and a declaration is not a statement [-Wpedantic]
With some compilers, this is an error rather than a pedantic warning.
Refs <https://github.com/Genymobile/scrcpy/issues/2256#issuecomment-1467008307>
Install all the prebuilt dependencies for Windows to a specific folder,
and use meson command line options to specify their location.
This removes crossbuild-specific code from the meson scripts and will
simplify dependency upgrades.
PR #4460 <https://github.com/Genymobile/scrcpy/pull/4460>
On Android 14 (Pixel 8), a device rotation while the camera app was
running resulted in an incorrect capture.
Destroying and recreating the display fixes the issue.
Audio PTS are retrieved by AudioRecord.getTimestamp(), so they do not
necessarily exactly match the number of samples (this allows to take
drift and lag into account).
As a consequence, two consecutive timestamps in microseconds may
sometimes end up within the same millisecond, causing the warning. This
is particularly true for the Matroska muxer which uses a timebase of
1/1000 (1 ms precision).
Since this is "expected", lower the log level from warning to debug.
If the difference of PTS between two consecutive blocks of audio is less
than 1 sample, then it will be considered as non-increasing by FFmpeg
muxers having a time_base of 1/sample_rate.
Increase the PTS by 1 sample instead.
If several reads are performed for a single captured audio block (e.g.
if the read size is smaller than the captured block), then the provided
timestamp was the same for all packets.
Recompute the timestamp for each of them.
In practice, the system captures audio samples by blocks of 1024
samples.
Remplace the hardcoded value of 5 milliseconds (240 samples), and let
AudioRecord fill the input buffer provided by MediaCodec (or by
AudioRawRecorder), with a maximum size of 1024 samples (just in case).
FLAC is not low latency: the default encoder produces blocks of 4096
samples, which represent ~85.333ms.
Increase the audio buffer by default so that audio playback works.
Clipboard synchronization requires SDL_INIT_VIDEO, so always initialize
the video subsystem, even if --no-video or --no-video-playback is
passed.
Refs caf594c90ef1b71ed844b2a9b42c3b3371215d6f
Fixes#4418 <https://github.com/Genymobile/scrcpy/issues/4418>
It is incorret to ever call:
streamer.writeDisableStream(...);
after:
streamer.writeAudioHeader();
Move the try-catch block so that it can never happen.
Some devices may provide invalid ranges, causing an
IllegalArgumentException "lower must be less than or equal to upper".
Catch the exception to list the cameras anyway.
Refs #4403 <https://github.com/Genymobile/scrcpy/issues/4403>
Interrupting async processors may require to shutdown the connection to
wake up blocking calls.
Therefore, shutdown the connection first, then join the threads, then
close the connection.
Refs commit 9c08eb79cb7941848882cb908cefee9933450de5
Add --camera-high-speed to enable high frame rate camera capture. If
the option is enabled, then --camera-fps is mandatory.
PR #4213 <https://github.com/Genymobile/scrcpy/pull/4213>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Andrew Gunnerson <accounts+github@chiller3.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Add a new option for specifying the camera frame rate.
By default, Android's default frame rate (30 fps) is used.
PR #4213 <https://github.com/Genymobile/scrcpy/pull/4213>
Signed-off-by: Andrew Gunnerson <accounts+github@chiller3.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
In addition to --camera-size to specify an explicit size, make it
possible to select the camera size automatically, respecting the maximum
size (already used for display mirroring) and an aspect ratio.
For example, "scrcpy --video-source=camera" followed by:
- (no additional arguments)
: mirrors at the maximum size, any a-r
- -m1920
: only consider valid sizes having both dimensions not above 1920
- --camera-ar=4:3
: only consider valid sizes having an aspect ratio of 4:3 (+/- 10%)
- -m2048 --camera-ar=1.6
: only consider valid sizes having both dimensions not above 2048
and an aspect ratio of 1.6 (+/- 10%)
PR #4213 <https://github.com/Genymobile/scrcpy/pull/4213>
Co-authored-by: Simon Chan <1330321+yume-chan@users.noreply.github.com>
Add an option to select the camera by its lens facing (front, back or
external).
PR #4213 <https://github.com/Genymobile/scrcpy/pull/4213>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
If --audio-source is not specified, select the default value
according to the video source:
- for display mirroring, use device audio by default;
- for camera mirroring, use microphone by default.
PR #4213 <https://github.com/Genymobile/scrcpy/pull/4213>
Extract an interface SurfaceCapture from ScreenEncoder, representing a
video source which can be rendered to a Surface for encoding.
Split ScreenEncoder into:
- ScreenCapture, implementing SurfaceCapture to capture the device
screen,
- SurfaceEncoder, to encode any SurfaceCapture.
This separation prepares the introduction of another SurfaceCapture
implementation to capture the camera instead of the device screen.
PR #4213 <https://github.com/Genymobile/scrcpy/pull/4213>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
By default, the Java process exits when all non-daemon threads are
terminated.
The Android SDK might start some non-daemon threads internally,
preventing the scrcpy server to exit in some cases.
So force the process to exit explicitly.
PR #4213 <https://github.com/Genymobile/scrcpy/pull/4213>
The option is named "display id" everywhere.
This will be consistent with --camera-id (there will be many camera
options, so an option --camera would be confusing).
PR #4213 <https://github.com/Genymobile/scrcpy/pull/4213>
The script install_release.sh downloads a file named scrcpy-server to
the repo root directory. Add it to .gitignore so that it is ignored.
PR #4364 <https://github.com/Genymobile/scrcpy/pull/4364>
Signed-off-by: Romain Vimont <rom@rom1v.com>
If --no-video was set, then device disconnection was not reported. To
avoid the problem, report device disconnection also on audio
end-of-stream (EOS).
If both video and audio are enabled, then a device disconnection event
will be sent twice, but only the first one will be handled (since it
makes scrcpy exit).
Fixes#4207 <https://github.com/Genymobile/scrcpy/issues/4207>
Some control messages can still be used even when video playback is
disabled (i.e. there is no window), for example to turn the screen off.
This reverts commit 92483fe11b6fd6bae5ef775ccaff78fefa92aad4
(semantically).
Fixes#4175 <https://github.com/Genymobile/scrcpy/issues/4175>
Some devices (Meizu) assume that the video encoding thread has a
Looper. By moving video encoding to a separate thread, commit
feab87053abcceded41342d9d856763dedc09187 broke this assumption.
Call Looper.prepare() from this thread to fix the problem.
Fixes#4143 <https://github.com/Genymobile/scrcpy/issues/4143>
On sway (a window manager), SDL_WINDOWEVENT_EXPOSED and
SDL_WINDOWEVENT_SIZE_CHANGED might not be called before a mouse event is
triggered. As a consequence, the "content rectangle" might not be
initialized when the mouse event is processed, causing a division by
zero.
To avoid the problem, initialize the content rect immediately when the
window is shown.
Fixes#4115 <https://github.com/Genymobile/scrcpy/issues/4115>
In forward mode, the dummy byte must be written immediately after the
first accept(), otherwise the client will wait indefinitely, causing a
deadlock (or a timeout).
Regression introduced by 8c650e53cd37a53d5c3aa746c30a71c6b742a4e2.
Audio did not work on Honor devices.
To make it work, a system context must be set as a base context of
FakeContext (so that a PackageManager is available), and a current
Application and ActivityThread must be set.
These workarounds must not be applied for all devices, because they
might cause other issues.
Fixes#4015 <https://github.com/Genymobile/scrcpy/issues/4015>
Refs #3085 <https://github.com/Genymobile/scrcpy/issues/3805>
Co-authored-by: Simon Chan <1330321+yume-chan@users.noreply.github.com>
In the recorder, if the video or audio stream is disabled, do not wait
for its initialization (it will never happen) to process the header.
In that case (scrcpy --no-audio --record=file.mp4), this caused the
whole content to be buffered in memory, and written only on exit.
The sc_cond_wait() in sc_recorder_process_header() needs to be notified
of changes to video_init/audio_init (protected by stream_cond) and
video_queue/audio_queue (protected by queue_cond).
Use only one condition variable to simplify.
The flag is used to reset the capture (restart the encoding) on rotation
change. It will also be used for other events (on folding change), so
rename it.
PR #3979 <https://github.com/Genymobile/scrcpy/pull/3979>
The comment was outdated:
- the "meta" header is now always present (not only when recording is
enabled);
- it is not only used for the video stream, but also for the audio
stream.
This option impacts video and audio _playback_. For example, if we use
V4L2, the device is still "mirrored" (via V4L2), even if playback is
disabled. Therefore, "playback" is more approriate than "mirror".
The initial option --no-display option was renamed to --no-mirror by
commit 6928acdeac29eee404a7c7014654965ef5128b88, but this has never been
released, so it is ok to rename it one more time.
Refs #3978 <https://github.com/Genymobile/scrcpy/pull/3978#issuecomment-1549420103>
PR #4033 <https://github.com/Genymobile/scrcpy/pull/4033>
On some window managers (e.g. on Windows), performing a resize while the
window is minimized does nothing (the restored window keeps its old
size).
Therefore, like for maximized and fullscreen states, wait for the window
to be restored to apply a resize.
Refs #3947 <https://github.com/Genymobile/scrcpy/issues/3947>
When the scrcpy window is minimized on Windows with D3D9, texture
creation and update fail.
In that case, do not terminate scrcpy. Instead, store the pending size
or frame to update, to attempt again during the next update or
rendering.
Fixes#3947 <https://github.com/Genymobile/scrcpy/issues/3947>
Some server parameters may depend on one another. For example,
audio_bit_rate is meaningless if audio is false.
But it is inconsistent to disable some parameters based on these
dependencies checks, but not others. Handling all dependencies between
parameters would add too much complexity for no benefit.
So just pass individual parameters independently.
PR #3978 <https://github.com/Genymobile/scrcpy/pull/3978>
Contrary to the other tasks (controller and audio capture/encoding), the
screen encoder was executed synchronously. As a consequence,
scrcpy-server could not terminate until the screen encoder returned.
Convert it to an async processor. This allows to terminate on controller
error, and this paves the way to disable video mirroring.
PR #3978 <https://github.com/Genymobile/scrcpy/pull/3978>
The async processors use the socket file descriptors from the
connection. Therefore, the connection must not be closed before all
async processor threads are joined.
PR #3978 <https://github.com/Genymobile/scrcpy/pull/3978>
Reword the supported API versions for audio forwarding sentence to
clarify that it supports API >= 30
PR #3949 <https://github.com/Genymobile/scrcpy/pull/3949>
Signed-off-by: Romain Vimont <rom@rom1v.com>
By default, SDL creates an OpenGL 2.1 context on macOS for an OpenGL
renderer. As a consequence, mipmapping is not supported.
Force to use a core profile context, to get a higher version.
Before:
INFO: Renderer: opengl
INFO: OpenGL version: 2.1 NVIDIA-14.0.32 355.11.11.10.10.143
WARN: Trilinear filtering disabled (OpenGL 3.0+ or ES 2.0+ required)
After:
INFO: Renderer: opengl
DEBUG: Creating OpenGL Core Profile context
INFO: OpenGL version: 4.1 NVIDIA-14.0.32 355.11.11.10.10.143
INFO: Trilinear filtering enabled
when running with:
scrcpy --verbosity=debug --render-driver=opengl
Note: Since SDL_CreateRenderer() causes a fallback to OpenGL 2.1, the
profile and version attributes have to be set and the context created
_after_.
PR #3895 <https://github.com/Genymobile/scrcpy/pull/3895>
Signed-off-by: Romain Vimont <rom@rom1v.com>
If a line did not end with '\r', then the final `\n' was replaced by
'\0' for parsing the current line. This `\0` was then mistakenly
considered as the end of the whole "ip route" output, so the remaining
lines were not parsed, causing "scrcpy --tcpip" to fail in some cases.
To fix the issue, read the final character of the current line before it
is (possibly) overwritten by '\0'.
The slope encodes the drift between the device clock and the computer
clock. Its real value is expected very close to 1.
To estimate it, just assume it is exactly 1.
Since the clock is used to estimate very close points in the future, the
error caused by clock drift is totally negligible, and in practice it is
way lower than the slope estimation error.
Therefore, only estimate the offset.
The code to start audio capture is more complicated for Android 11
(launch a fake popup, wait, make several attempts, close the shell
package).
Use a distinct code path specific to Android 11.
On Android 11, a fake popup must be briefly opened to make the system
think that the shell app is in the foreground so that audio may be
recorded.
Making the shell app foreground may take some time depending on the
device, so make 3 attempts, waiting 100ms before each.
Fixes#3796 <https://github.com/Genymobile/scrcpy/issues/3796>
On some systems, the SDL audio callback is not called frequently enough
(for example it requests 5ms of samples every 10ms), because the output
buffer is too small.
By default, we want to use a small value (5ms) to minimize latency and
buffer underrun, but if it does not work well, users need a way to
increase it.
Refs #3793 <https://github.com/Genymobile/scrcpy/issues/3793>
There were several workarounds applied in a single method. Some of them
are specific to Meizu phones, but cause issues on other devices.
Split the method to be able to only fill the app context for audio
capture without applying the Meizu workarounds.
Fixes#3801 <https://github.com/Genymobile/scrcpy/issues/3801>
scrcpy is available directly in the distro, drop link to the overlay
(which only contains older versions).
PR #3816 <https://github.com/Genymobile/scrcpy/pull/3816>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The codec id to write as codec parameters is the one from the v4l2
encoder, not from the decoder.
Regression introduced by be985b8242e0626288674b84c5039725170f8f0c.
Fixes#3795 <https://github.com/Genymobile/scrcpy/issues/3795>
An int was compared with an unsigned:
../app/src/audio_player.c:290:27: warning: comparison of integers of
different signs: 'int' and 'unsigned int' [-Wsign-compare]
if (abs(diff) < ap->sample_rate / 1000) {
~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~
In C, a label can only be followed by a statement, not a declaration.
An error in `app/src/screen.c` violated this, and led to a build error
with an error message similar to the one below:
../app/src/screen.c:821:13: error: expected expression
bool ok = sc_screen_init_size(screen);
^
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/lib/clang/13.0.0/include/stdbool.h:15:14: note: expanded from macro 'bool'
#define bool _Bool
^
../app/src/screen.c:822:18: error: use of undeclared identifier 'ok'
if (!ok) {
^
2 errors generated.
This could be fixed by introducing a new block (or compound statement;
as is already being done in the next `case`). That is a statement.
Fixes#3785 <https://github.com/Genymobile/scrcpy/issues/3785>
PR #3787 <https://github.com/Genymobile/scrcpy/pull/3787>
Signed-off-by: Ruoyu Zhong <zhongruoyu@outlook.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The README.md page is HUGE. Split it up.
Also document audio forwarding and improve installation instructions for
each platform and user documentation.
PR #3774 <https://github.com/Genymobile/scrcpy/pull/3774>
On buffer underflow, the average buffering must be updated, but it is
intended to be accessed only from the receiver thread.
Make the player and the receiver thread communicate the underflow via a
new field (ap->underflow).
On initial connection, scrcpy sent some device metadata:
- the device name (to be used as window title)
- the initial video size (before any frame or even SPS/PPS)
But it is better to provide the initial video size as part as the video
stream, so that it can be demuxed and exposed via AVCodecContext to
sinks.
This avoids to pass an explicit "initial frame size" for the screen, the
recorder and the v4l2 sink.
Previously, the packet sink push() implementation just set the codec and
notified a wait condition. Then the recorder thread read the codec and
created the AVStream.
But this was racy: an AVFrame could be pushed before the creation of the
AVStream, causing its video_stream_index or audio_stream_index to be
initialized to -1.
Also, in the future, the AVStream initialization might need data
provided by the packet sink open(), so initialize it there (with a
mutex).
The sc_file_pusher is lazy-initialized, but it was stopped and joined in
all cases (accessing uninitialized values).
Detected by poisoning the struct scrcpy instance with ASAN enabled.
All server logs were printed to stdout, while all client logs were
printed to stderr.
Instead, use stderr for warnings and errors, stdout for the others:
- stdout: verbose, debug, info
- stderr: warn, error
System.out.println() first prints the message, then the new line.
Between these two calls, the client might print a message, breaking
formatting.
Instead, call System.out.print() with '\n' appended to the message.
On the server side, several components are started, stopped and joined.
Extract an interface to handle them generically.
This will help to support both encoded and raw audio stream, because
they will be two different concrete components, but implementing the
same interface.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
Expose an option to add a buffering delay (in milliseconds) before
playing audio.
This is similar to the options --display-buffer and --v4l2-buffer for
video frames.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
Play the decoded audio using SDL.
The audio player frame sink receives the audio frames, resample them
and write them to a byte buffer (introduced by this commit).
On SDL audio callback (from an internal SDL thread), copy samples from
this byte buffer to the SDL audio buffer.
The byte buffer is protected by the SDL_AudioDeviceLock(), but it has
been designed so that the producer and the consumer may write and read
in parallel, provided that they don't access the same slices of the
ring-buffer buffer.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
Co-authored-by: Simon Chan <1330321+yume-chan@users.noreply.github.com>
A delay buffer delayed all the frames except the first one, to open the
scrcpy window immediately and get a picture.
Make this feature optional, so that the delay buffer might also be used
for audio (especially for simulating a high delay for debugging).
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
For clarity, the fields used only when a delay was set were wrapped in
an anonymous structure.
Now that the delay buffer has been extracted to a separate component,
the delay is necessarily set (it may not be 0), so the fields are always
used.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
The components needing delayed frames (sc_screen and sc_v4l2_sink)
managed a sc_video_buffer instance, which itself embedded a
sc_frame_buffer instance (to keep only the most recent frame).
In theory, these components should not be aware of delaying: they should
just receive AVFrames later, and only handle a sc_frame_buffer.
Therefore, refactor sc_delay_buffer as a frame source (it consumes)
frames) and a frame sink (it produces frames, after some delay), and
plug an instance in the pipeline only when a delay is requested.
This also removes the need for a specific sc_video_buffer.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
There was a frame sink trait, implemented by components able to receive
AVFrames, but each frame source had to manually send frame to sinks.
In order to mutualise sink management, add a frame sink trait.
There was a packet sink trait, implemented by components able to
receive AVPackets, but each packet source had to manually send packets
to sinks.
In order to mutualise sink management, add a packet source trait.
A video buffer had 2 responsibilities:
- handle the frame delaying mechanism (queuing packets and pushing them
after the expected delay);
- keep only the most recent frame (using a sc_frame_buffer).
In order to be able to reuse only the frame delaying mechanism, extract
it to a separate component, sc_delay_buffer.
The video_buffer thread clears the queue once it is stopped, but new
frames might still be pushed asynchronously.
To avoid the problem, do not push any frame once the video_buffer is
stopped.
The packets queued for buffering were wrapped in a dynamically allocated
structure with a "next" field.
To avoid this additional layer of allocation and indirection, use a
VecDeque.
The packets queued for recording were wrapped in a dynamically allocated
structure with a "next" field.
To avoid this additional layer of allocation and indirection, use a
VecDeque.
Since in scrcpy a video packet passed to avcodec_send_packet() is always
a complete video frame, it is sufficient to call avcodec_receive_frame()
exactly once.
In practice, it also works for audio packets: the decoder produces
exactly 1 frame for 1 input packet.
In theory, it is an implementation detail though, so
avcodec_receive_frame() should be called in a loop.
By default, scrcpy mirrors only the video when audio capture fails on
the device. Add an option to force scrcpy to fail if audio is enabled
but does not work.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
On Android 11, it is possible to start the capture only when the running
app is in foreground. But scrcpy is not an app, it's a Java application
started from shell.
As a workaround, start an existing Android shell existing activity just
to start the capture, then close it immediately.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
If there is exactly one producer, then it can assume that the remaining
space in the buffer will only increase until it writes something.
This assumption may allow the producer to write to the buffer (up to a
known safe size) without any synchronization mechanism, thus allowing
to read and write different parts of the buffer in parallel.
The producer can then commit the write with a lock held, and update its
knowledge of the safe empty remaining space.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
When audio capture fails on the device, scrcpy continues mirroring the
video stream. This allows to enable audio by default only when
supported.
However, if an audio configuration occurs (for example the user
explicitly selected an unknown audio encoder), this must be treated as
an error and scrcpy must exit.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
If no bit-rate is passed, let the server use the default value (8Mbps).
This avoids to define a default value on both sides, and to pass the
default bit-rate as an argument when starting the server.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
The permission "android.permission.RECORD_AUDIO" has been added for
shell in Android 11.
Moreover, on lower versions, it may make the server segfault on the
device (happened on a Nexus 5 with Android 6.0.1).
Refs <4feeee8891%5E%21/>
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
By default, audio is enabled (--no-audio must be explicitly passed to
disable it).
However, some devices may not support audio capture (typically devices
below Android 11, or Android 11 when the shell application is not
foreground on start).
In that case, make the server notify the client to dynamically disable
audio forwarding so that it does not wait indefinitely for an audio
stream.
Also disable audio on unknown codec or missing decoder on the
client-side, for the same reasons.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
For video streams (at least H.264 and H.265), the config packet
containing SPS/PPS must be prepended to the next packet (the following
keyframe).
For audio streams (at least OPUS), they must not be merged.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
The client does not use the audio stream if there is no display and no
recording (i.e. only V4L2), so disable audio so that the device does not
attempt to capture it.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
For OPUS codec, FFmpeg expects the raw extradata, but MediaCodec wraps
it in some structure.
Fix the config packet to send only the raw extradata.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
Create an AudioRecorder to capture the audio source REMOTE_SUBMIX.
For now, the captured packets are just logged into the console.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
When audio is enabled, open a new socket to send the audio stream from
the device to the client.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Audio will be enabled by default (when supported). Add an option to
disable it.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Remove USER_ID from ServiceManager, and replace it by a constant in
FakeContext.
This is the same as android.os.Process.ROOT_UID, but this constant has
been introduced in API 29.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
Since scrcpy-server is not an Android application (it's a java
executable), it has no Context.
Some features will require a Context instance to get the package name
and the UID. Add a FakeContext for this purpose.
PR #3757 <https://github.com/Genymobile/scrcpy/pull/3757>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The Callbacks interface notifies new packets. But in addition, the
screen encoder will need to write headers on start.
We could add a function onStart(), but for simplicity, just remove the
interface, which brings no value, and call the streamer directly.
Refs 87972e2022686b1176bfaf0c678e703856c2b027
On any error, all previously opened sockets must be closed.
Handle these errors in a single catch-block. Currently, there are only 2
sockets, but this will simplify even more with more sockets.
Note: this commit is better displayed with --ignore-space-change (-b).
The recorder opened the target file from the packet sink open()
callback, called by the demuxer. Only then the recorder thread was
started.
One golden rule for the recorder is to never block the demuxer for I/O,
because it would impact mirroring. This rule is respected on recording
packets, but not for the initial recorder opening.
Therefore, start the recorder thread from sc_recorder_init(), open the
file immediately from the recorder thread, then make it wait for the
stream to start (on packet sink open()).
Now that the recorder can report errors directly (rather than making the
demuxer call fail), it is possible to report file opening error even
before the packet sink is open.
The recorder has two initialization phases: one to initialize the
concrete recorder object, and one to open its packet_sink trait.
Initialize mutex and condvar as part of the object initialization.
If there were several packet_sink traits (spoiler: one for video, one
for audio), then the mutex and condvar would still be initialized only
once.
Stop scrcpy on recorder errors.
It was previously indirectly stopped by the demuxer, which failed to
push packets to a recorder in error. Report it directly instead:
- it avoids to wait for the next demuxer call;
- it will allow to open the target file from a separate thread and stop
immediately on any I/O error.
User-friendly error messages are printed on specific configuration
exceptions. In that case, do not print the stacktrace.
Also handle the user-friendly error message directly where the error
occurs, and print multiline messages in a single log call, to avoid
confusing interleaving.
On the scrcpy-deps repo, I built FFmpeg 5.1.2 binaries for Windows with
only the features used by scrcpy.
For comparison, here are the sizes of the dll for FFmpeg 5.1.2:
- before: 89M
- after: 4.7M
It also allows to upgrade the old FFmpeg version (4.3.1) used for win32.
Refs <https://github.com/rom1v/scrcpy-deps>
Refs <https://github.com/Genymobile/scrcpy/issues/1753>
As reported by gradle:
> Setting the namespace via a source AndroidManifest.xml's package
> attribute is deprecated.
>
> Please instead set the namespace (or testNamespace) in the module's
> build.gradle file, as described here:
> https://developer.android.com/studio/build/configure-app-module#set-namespace
Running scrcpy --tcpip on a device already connected via TCP/IP did not
initialize server->serial.
As a consequence, in debug mode, an assertion failed:
scrcpy: ../app/src/server.c:770: run_server: Assertion
`server->serial' failed.
In release mode, scrcpy failed with this error:
adb: -s requires an argument
Scrcpy does not use FFmpeg network features. Initialize network locally
instead (useful only for Windows).
The include block has been moved to fix the following warning:
Please include winsock2.h before windows.h
DesktopConnection implements Closeable, so it is implicitly closed after
its try-with-resources block. Closing the DesktopConnection shutdowns
the sockets, so it is necessary in particular to wake up blocking read()
calls from the controller.
But the controller thread was joined before the DesktopConnection was
closed, causing a deadlock. To fix the problem, join the controller
thread only after the DesktopConnection is closed.
Refs 400a1c69b198103feff9682895529529890c1acc
When a call to a packet or frame sink fails, do not log the error on the
caller side: either the "failure" is expected (explicitly stopped) or it
must be logged by the packet or frame sink implementation.
On close, the client closes the socket. This wakes up socket blocking
calls on the server-side, by throwing an exception. Since this exception
is expected, it was not logged.
However, other IOExceptions might occur, which must not be ignored. For
that purpose, log only IOException when they are not caused by an EPIPE
error.
A random identifier is generated to differentiate multiple running
scrcpy instances. Rename it from 'uid' to 'scid' (scrcpy id) not to
confuse it with Linux UID.
Fixes#3729 <https://github.com/Genymobile/scrcpy/issues/3729>
Refs 4315be164823d2c8fc44b475b52af79bfee98ff1
The PTS received from MediaCodec are expressed relative to an arbitrary
clock origin. We consider the PTS of the first frame to be 0, and the
PTS of every other frame is relative to this first PTS (note that the
PTS is only used for recording, it is ignored for mirroring).
For simplicity, this relative PTS was computed on the server-side.
To prepare support for multiple stream (video and audio), send the
packet with its original PTS, and handle the PTS offset on the
client-side (by the recorder).
Since we can't know in advance which stream will produce the first
packet with the lowest PTS (a packet received later on one stream may
have a PTS lower than a packet received earlier on another stream),
computing the PTS on the server-side would require unnecessary waiting.
On mouse click events:
- the first button pressed must first generate ACTION_DOWN;
- all button pressed (including the first one) must generate
ACTION_BUTTON_PRESS;
- all button released (including the last one) must generate
ACTION_BUTTON_RELEASE;
- the last button released must in addition generate ACTION_UP.
Otherwise, Chrome does not work properly.
Fixes#3635 <https://github.com/Genymobile/scrcpy/issues/3635>
Signed-off-by: Romain Vimont <rom@rom1v.com>
If the pointer is a mouse, the pointer is UP only when no buttons are
pressed (not when a button is released, because there might be other
buttons still pressed).
Refs #3635 <https://github.com/Genymobile/scrcpy/issues/3635>
Signed-off-by: Romain Vimont <rom@rom1v.com>
On click event, only the whole buttons state was passed to the device.
In addition, on ACTION_DOWN and ACTION_UP, pass the button associated to
the action.
Refs #3635 <https://github.com/Genymobile/scrcpy/issues/3635>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
MediaCodec may fail spuriously, typically when stopping an encoding and
starting a new one immediately (for example on device rotation).
In that case, retry a few times, in many cases it should work.
Refs #3693 <https://github.com/Genymobile/scrcpy/issues/3693>
For the initial connection between the device and the computer, an adb
tunnel is established (with "adb reverse" or "adb forward").
The device-side of the tunnel is a local socket having the hard-coded
name "scrcpy". This may cause issues when several scrcpy instances are
started in a few seconds for the same device, since they will try to
bind the same name.
To avoid conflicts, make the client generate a random UID, and append
this UID to the local socket name ("scrcpy_01234567").
Use av_packet_ref() to reference the packet without copy.
This also simplifies the logic, by making the "offset" variable and the
memcpy() call local to the if-block.
Right click and middle click require the source device to be a mouse,
not a touchscreen. Therefore, the source device was changed only when a
button other than the primary button was pressed (see
adc547fa6e8e6167cd9633a97d98de6665b8c23a).
However, this led to inconsistencies between the ACTION_DOWN when a
secondary button is pressed (with a mouse as source device) and the
matching ACTION_UP when the secondary button is released (with a
touchscreen as source device, because then there is no button pressed).
To avoid the problem in all cases, force a mouse as source device when
--forward-all-clicks is set.
Concretely, for mouse events in --forward-all-clicks mode:
- device source is set to InputDevice.SOURCE_MOUSE;
- motion event toolType is set to MotionEvent.TOOL_TYPE_MOUSE;
Otherwise (when --forward-all-clicks is unset, or for real touch
events), finger events are injected:
- device source is set to InputDevice.SOURCE_TOUCHSCREEN;
- motion event toolType is set to MotionEvent.TOOL_TYPE_FINGER.
Fixes#3568 <https://github.com/Genymobile/scrcpy/issues/3568>
PR #3579 <https://github.com/Genymobile/scrcpy/pull/3579>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
This fixes the following warning:
> WARNING: Running the setup command as `meson [options]` instead of
> `meson setup [options]` is ambiguous and deprecated.
If the current adb port is not 5555 (typically 0 because it is not in
TCP/IP mode), --tcpip automatically executes (among other commands):
adb tcpip 5555
In case adb was already listening on another port, this command forced
to listen on 5555, and the connection should still succeed.
But this reconfiguration might be inconvenient for the user. If adb is
already in TCP/IP mode, use the current enabled port without
reconfiguration.
Fixes#3591 <https://github.com/Genymobile/scrcpy/issues/3591>
Mandatory arguments may be passed in either of these two forms:
1. --key value
2. --key=value
Optional argument may only be passed in the second form.
For consistency, always document using --key=value.
Refs f76fe2c0d4847d0e40d8708f97591abf3fa22ea5
There were exactly one instance of ServiceManager and Settings, stored
in Device.
Since a Device instance is not created by the CleanUp executable, it was
not straightforward to call wrapper methods on cleanup.
Remove this artificial restriction and expose them publicly via static
methods (this is equivalent to expose a singleton, but less verbose).
Add a launcher which opens a terminal, and keep it open in case of
errors (so that the user has time to read error messages).
The behavior is the same as scrcpy-console.bat on Windows.
PR #3351 <https://github.com/Genymobile/scrcpy/pull/3351>
This method is deprecated since Meson 0.56.0:
<https://mesonbuild.com/Release-notes-for-0-56-0.html#mesonbuild_root-and-mesonsource_root-are-deprecated>
We could replace it with meson.project_source_root(), but this would
make Meson 0.56 or above mandatory. Since the path in always computed
from the server/ directory, just add '..' to reference the root project
directory.
Refs c456e3826470753bdaefcffc93e9ae1bf25b12f7
This lowers the barrier to contribute to translations, and frees up my
maintenance time used to review and merge translations and their updates
in many languages.
On Windows and macOS, resizing blocks the event loop. Handling it
properly would require the same workaround as done in screen.c.
This reverts commit 436b368f9df4469274502f29626586b48db44a0c.
The previous commit replaced the IInterface instance (the "input"
service) by the InputManager instance (retrieved by
InputManager.getInstance()).
Both define an "injectInputEvent" method, but the alternate version
(probably) does not concern the InputManager.
This reverts commit b7a06278fee0dbfbb18b651ad563d9fb0797c8bc.
PR #3190 <https://github.com/Genymobile/scrcpy/pull/3190>
We must distinguish 3 cases for await_for_server():
- an error occurred
- no error occurred, the device is connected
- no error occurred, the device is not connected (user requested to
quit)
For this purpose, use an additional output parameter to indicate if the
device is connected (only set when no error occurs).
Refs #3085 <https://github.com/Genymobile/scrcpy/pull/3085>
Add a paragraph about toggling an option to bypass having to physically
connect the device to the user's computer.
PR #1303 <https://github.com/Genymobile/scrcpy/pull/1303>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Add a small troubleshooting section since wireless might add some
complexity, and to lessen incoming relevant issue posts.
PR #1303 <https://github.com/Genymobile/scrcpy/pull/1303>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Mention that the device must be plugged via USB before configuring
TCP/IP connections.
It wasn't obvious that the device should be first plugged before running
scrcpy wirelessly, especially to those who aren't very familiar with
adb.
Note from committer: add this new step with index 0 to make the diff
readable, the next commit will renumber all the steps.
PR #1303 <https://github.com/Genymobile/scrcpy/pull/1303>
Signed-off-by: Romain Vimont <rom@rom1v.com>
For simplicity, the parsing of `adb devices -l` output is performed in a
single pass on the whole output.
This output was limited to 4096 bytes. Since there are about 100 chars
per device line, this limited the number of connected devices to ~40.
Increase to 65536 bytes to avoid a limitation in practice.
PR #3035 <https://github.com/Genymobile/scrcpy/pull/3035>
The files in data/ are specific to the client app (not the server).
This also avoids to reference the parent directory (../) from
app/meson.build.
Refs 8d583d36e259ba7f5f21d7a703cca73184200aa9
All adb commands are executed with an "interruptor", so that they can be
interrupted on Ctrl+C.
Make this interruptor optional, so that we could call "adb kill-server"
in OTG mode. This command always returns almost immediately anyway.
Ideally, we should make all blocking calls interruptible (including
libusb calls, by using the asynchronous API), but it's a lot of work,
and in practice it works well enough.
PR #3011 <https://github.com/Genymobile/scrcpy/pull/3011>
USB device disconnection is detected via a hotplug callback when it is
supported.
In addition, report disconnection on libusb calls returning
LIBUSB_ERROR_NO_DEVICE or LIBUSB_ERROR_NOT_FOUND. This allows to detect
disconnection after a libusb call when hotplug is not available.
PR #3011 <https://github.com/Genymobile/scrcpy/pull/3011>
Using the screen instance or not in these functions is an implementation
detail. Further changes will require the screen instance.
Refs 7848a387c8281cb156f0e0da6bbbb05cda31db22
PR #3031 <https://github.com/Genymobile/scrcpy/pull/3031>
There were three links that weren't displayed correctly due to incorrect
references:
- the Windows release link to `README.md#windows`
- 2 links that expected a German reference but got an English reference
PR #3026 <https://github.com/Genymobile/scrcpy/pull/3026>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The value of sc_tick_now() has microsecond precision, but
sc_cond_timedwait() has only millisecond precision.
To guarantee that sc_tick_now() >= deadline when sc_cond_timedwait()
returns due to timeout, round up to the next millisecond.
This avoids to call a non-blocking sc_cond_timedwait() in a loop for no
reason until a target deadline during up to 1 millisecond.
Refs 682a6911735cb8f6dccd9653ce30b72f267235c6
The type uint32_t is not sufficient to store the result of
sc_tick_now().
As a consequence, the FPS counter entered a live loop and caused a lock
starvation (deadlock in practice).
Refs ec871dd3f596a8183e37982821645ac5a5791fe0
Refs 682a6911735cb8f6dccd9653ce30b72f267235c6
A special PTS value was used to encode a config packet.
To prepare for adding more flags, use the most significant bits of the
PTS field to store flags.
The relative mouse mode is tracked by SDL, and accessible via
SDL_GetRelativeMouseMode().
This is more robust in case SDL changes the relative mouse mode on its
own.
If sc_usb_connect() failed, then the sc_usb_device was never destroyed.
The assignment was mistakenly removed by commit
61969aeb80093d0777c7716a61698cbdaf9ddd71.
The prebuilt dependencies are specific to the client app (not the
server).
This also avoids to reference the parent directory (../) from
app/meson.build.
Without this log, the user would have no way to know that a USB device
is rejected because it could not be opened (typically due to
insufficient permissions).
If several devices are connected (as listed by `adb devices`), it was
necessary to provide the explicit serial via -s/--serial.
If only one device is connected via USB (respectively, via TCP/IP), it
might be convenient to select it automatically. For this purpose, two
new options are introduced:
- -d/--select-usb: select the single device connected over USB
- -e/--select-tcpip: select the single device connected over TCP/IP
PR #3005 <https://github.com/Genymobile/scrcpy/pull/3005>
Currently, a device is selected either from a specific serial, or if it
is the only one connected.
In order to support selecting the only device connected via USB or via
TCP/IP separately, introduce a new selection structure.
PR #3005 <https://github.com/Genymobile/scrcpy/pull/3005>
This does nothing if the adb daemon is already started, but allows to
print any output/errors to the console.
Otherwise, the daemon starting would occur during `adb devices`, which
does not output to the console because the result is parsed.
PR #3005 <https://github.com/Genymobile/scrcpy/pull/3005>
Since the previous commit, if a serial is given via -s/--serial (either
a real USB serial or an IP:port), a device is selected if its serial
matches exactly.
In addition, if the user pass an IP without a port, then select any
device with this IP, regardless of the port (so that "192.168.1.1"
matches any "192.168.1.1:port"). This is also the default behavior of
adb.
PR #3005 <https://github.com/Genymobile/scrcpy/pull/3005>
List all USB devices in a first step, then select the matching one(s).
This allows to report a user-friendly log message containing the list of
devices, with the matching one(s) highlighted.
PR #3005 <https://github.com/Genymobile/scrcpy/pull/3005>
One log macro was provided for each log level (LOGV(), LOGD(), LOGI(),
LOGW(), LOGE()).
Add a generic macro LOG(LEVEL, ...) accepting a log level as parameter,
so that it is possible to write logging wrappers.
PR #3005 <https://github.com/Genymobile/scrcpy/pull/3005>
This util function was error-prone:
- it accepted a buffer as parameter (not necessarily a NUL-terminated
string) and its length (including the NUL char, if any);
- it wrote '\0' over the last character of the buffer, so the last
character was lost if the buffer was not a NUL-terminated string, and
even worse, it caused undefined behavior if the length was empty;
- it returned the length of the resulting NUL-terminated string,
which was inconsistent with the input buffer length.
In addition, it was not necessarily optimal:
- it wrote '\0' twice;
- it required to know the buffer length, that is the input string
length + 1, in advance.
Remove this function, and let the client use strcspn() manually.
The function assumed that the raw output of "adb connect" was a
NUL-terminated string, but it is not the case.
It this output did not end with a space or a new line character, then
sc_str_truncate() would write '\0' over the last character. Even worse,
if the output was empty, then sc_str_truncate() would write
out-of-bounds.
Avoid the error-prone sc_str_truncate() util function.
The function assumed that the raw output of "adb get-serialno" was a
NUL-terminated string, but it is not the case.
It this output did not end with a space or a new line character, then
sc_str_truncate() would write '\0' over the last character. Even worse,
if the output was empty, then sc_str_truncate() would write
out-of-bounds.
Avoid the error-prone sc_str_truncate() util function.
The function assumed that the raw output of "adb getprop" was a
NUL-terminated string, but it is not the case.
It this output did not end with a space or a new line character, then
sc_str_truncate() would write '\0' over the last character. Even worse,
if the output was empty, then sc_str_truncate() would write
out-of-bounds.
Avoid the error-prone sc_str_truncate() util function.
The parser assumed that its input was a NUL-terminated string, but it
was not the case: it is just the raw output of "adb devices ip route".
In practice, it was harmless, since the output always ended with '\n'
(which was replaced by '\0' on truncation), but it was incorrect
nonetheless.
Always write a '\0' at the end of the buffer, and explicitly parse as a
NUL-terminated string. For that purpose, avoid the error-prone
sc_str_truncate() util function.
Before starting the server, the actual device serial (possibly its
ip:port if it's over TCP/IP) must be known.
A serial might be requested via -s/--serial (stored in the
sc_server_params), but the actual serial may change afterwards:
- if none is provided, then it is retrieved with "adb get-serialno";
- if --tcpip is requested, then the final serial will be the target
ip:port.
The requested serial was overwritten by the actual serial in the
sc_server_params struct, which was a bit hacky.
Instead, store a separate serial field in sc_server (and rename the one
from sc_server_params to "req_serial" to avoid confusion).
Now that providing a serial is mandatory for adb commands where it is
relevant, the whole argv array may be built statically, without
allocations at runtime.
If no serial is passed, then the command would work if there is exactly
one device connected, but will fail with multiple devices.
To avoid such cases, ensure that a serial is always provided.
On Windows, adb is provided in the release archive. Most missing adb
issues come from users setting the ADB environment variable to an
incorrect value (on all platforms).
Suggesting to install platform-tools to solve the problem will just make
things worse (there will be one more adb in yet another location).
The new retry mechanism with a lower definition only worked if the error
occurred during encode(). For example:
java.lang.IllegalStateException
at android.media.MediaCodec.native_dequeueOutputBuffer(Native Method)
at android.media.MediaCodec.dequeueOutputBuffer(MediaCodec.java:3452)
at com.genymobile.scrcpy.ScreenEncoder.encode(ScreenEncoder.java:114)
at com.genymobile.scrcpy.ScreenEncoder.internalStreamScreen(ScreenEncoder.java:95)
at com.genymobile.scrcpy.ScreenEncoder.streamScreen(ScreenEncoder.java:61)
at com.genymobile.scrcpy.Server.scrcpy(Server.java:80)
at com.genymobile.scrcpy.Server.main(Server.java:255)
However, MediaCodec may also fail before encoding, during configure() or
start(). For example:
android.media.MediaCodec$CodecException: Error 0xfffffc0e
at android.media.MediaCodec.native_configure(Native Method)
at android.media.MediaCodec.configure(MediaCodec.java:1956)
at android.media.MediaCodec.configure(MediaCodec.java:1885)
at com.genymobile.scrcpy.ScreenEncoder.configure(ScreenEncoder.java:158)
at com.genymobile.scrcpy.ScreenEncoder.streamScreen(ScreenEncoder.java:68)
at com.genymobile.scrcpy.Server.scrcpy(Server.java:28)
at com.genymobile.scrcpy.Server.main(Server.java:110)
Also downscale and retry in these cases.
Refs #2947 <https://github.com/Genymobile/scrcpy/pull/2947>
Refs #2988 <https://github.com/Genymobile/scrcpy/issues/2988>
PR #2990 <https://github.com/Genymobile/scrcpy/pull/2990>
The function was initially implemented to truncate lines, but was later
generalized to accept custom delimiters. The whole documentation has not
been updated accordingly.
Refs 9619ade706824a92ea387bdc9d0d27816bf79da5
This aims to fix two issues with the previous implementation:
1. the whole content of downloaded archives were extracted, while only
few files are necessary;
2. the archives were extracted in the prebuild-deps/ directory as is.
As a consequence of (2), the actual directory name relied on the root
directory of the archive. For adb, this root directory was always
"platform-tools", so when bumping the adb version, the target directory
already existed and the dependency was not upgraded (the old one had to
be removed manually).
Expose common function to download a file and check its checksum, but
let the custom script for each dependency extract only the needed files
and reorganize the content if necessary.
Add an option --otg to run scrcpy with only physical keyboard and mouse
simulation (HID over AOA), without mirroring and without requiring adb.
To avoid adding complexity into the scrcpy initialization and screen
implementation, OTG mode is implemented totally separately, with a
separate window.
PR #2974 <https://github.com/Genymobile/scrcpy/pull/2974>
Input events helpers to convert from SDL events to scrcpy events were
implemented in input_manager. To reuse them for OTG mode, move them to
input_events.h.
PR #2974 <https://github.com/Genymobile/scrcpy/pull/2974>
The device disconnection is detected when the video socket closes.
In order to introduce an OTG mode (HID events) without mirroring (and
without server), we must be able to detect USB device disconnection.
This feature will only be used in OTG mode.
PR #2974 <https://github.com/Genymobile/scrcpy/pull/2974>
Acksync is used to delay HID events until some request (in practice,
device clipboard synchronization) is acknowledged by the device.
This mechanism will not be necessary for OTG mode.
PR #2974 <https://github.com/Genymobile/scrcpy/pull/2974>
The device was automatically found by sc_usb_connect(). Instead, expose
a function to find a device from a serial, and let the caller connect to
the device found (if any).
This will allow to list all devices first, then select one device to
connect to.
PR #2974 <https://github.com/Genymobile/scrcpy/pull/2974>
Use it from accept_device() to simplify (at the cost an additional
allocation for each serial, but it is not important).
It will also be useful in other functions in further commits.
PR #2974 <https://github.com/Genymobile/scrcpy/pull/2974>
The server needs to interrupt the sockets on stop, but it must not close
them while other threads may attempt to read from or write to them.
In particular, the video_socket is read by the stream thread, and the
control_socket is written by the controller and read by receiver.
Therefore, close the socket only on sc_server_destroy(), which is called
after all other threads are joined.
Reported by TSAN on close:
WARNING: ThreadSanitizer: data race (pid=3287612)
Write of size 8 at 0x7ba000000080 by thread T1:
#0 close ../../../../src/libsanitizer/tsan/tsan_interceptors_posix.cpp:1690 (libtsan.so.0+0x359d8)
#1 net_close ../app/src/util/net.c:280 (scrcpy+0x23643)
#2 run_server ../app/src/server.c:772 (scrcpy+0x20047)
#3 <null> <null> (libSDL2-2.0.so.0+0x905a0)
Previous read of size 8 at 0x7ba000000080 by thread T16:
#0 recv ../../../../src/libsanitizer/sanitizer_common/sanitizer_common_interceptors.inc:6603 (libtsan.so.0+0x4f4a6)
#1 net_recv_all ../app/src/util/net.c:228 (scrcpy+0x234a9)
#2 stream_recv_packet ../app/src/stream.c:33 (scrcpy+0x2045c)
#3 run_stream ../app/src/stream.c:228 (scrcpy+0x21169)
#4 <null> <null> (libSDL2-2.0.so.0+0x905a0)
Refs ddb9396743072f97628fab168ef7fcd45a597b03
On Windows, the file explorer does not provide any "open a terminal
here" shortcut.
Add a bat file which just runs `cmd` to easily open a terminal directly
in the scrcpy directory (so there is no need to `cd C:\path\…`).
PR #2970 <https://github.com/Genymobile/scrcpy/pull/2970>
For convenience, this new option forces the 3 following options:
- send_device_meta=false
- send_frame_meta=false
- send_dummy_byte=false
This allows to send a raw H.264 stream on the video socket.
Concretely:
adb push scrcpy-server /data/local/tmp/scrcpy-server.jar
adb forward tcp:1234 localabstract:scrcpy
adb shell CLASSPATH=/data/local/tmp/scrcpy-server.jar \
app_process / com.genymobile.scrcpy.Server 1.21 \
raw_video_stream=true tunnel_forward=true control=false
As soon as a client connects via TCP to localhost:1234, it will receive
the raw H.264 stream.
Refs #1419 comment <https://github.com/Genymobile/scrcpy/pull/1419#issuecomment-1013964650>
PR #2971 <https://github.com/Genymobile/scrcpy/pull/2971>
Similar to send_device_frame, this option allows to disable sending the
device name and size on start.
This is only useful when using the scrcpy-server alone to get a raw
H.264 stream, without using the scrcpy client.
PR #2971 <https://github.com/Genymobile/scrcpy/pull/2971>
Move the options unused by the scrcpy client at the end.
These options may be useful to use scrcpy-server directly (to get a raw
H.264 stream for example).
PR #2971 <https://github.com/Genymobile/scrcpy/pull/2971>
Retry with a lower definition if MediaCodec fails before the first
frame, not the first packet.
In practice, the first packet is a config packet without any frame, and
MediaCodec might fail just after.
Refs 2eb6fe7d81c15912a2252f229d317ae03f618a33
Refs #2963 <https://github.com/Genymobile/scrcpy/issues/2963>
If --no-control is requested, then the controller instance is not
initialized. However, its reference was still passed to screen and
input_manager.
Instead, pass NULL if no controller is available.
MediaCodec errors always trigger IllegalStateException or a subtype
(like MediaCodec.CodecException).
In practice, this avoids to retry if the error is caused by an
IOException when writing the video packet to the socket.
The purpose of automatic downscaling on error is to make mirroring work
by just starting scrcpy without an explicit -m value, even if the
encoder could not encode at the screen definition.
It is only useful when we detect an encoding failure before the first
frame. Downsizing later could be surprising, so disable it.
PR #2947 <https://github.com/Genymobile/scrcpy/pull/2947>
The optimal initial size was computed from the expected dimensions, sent
immediately by the server before encoding any video frame.
However, the actual frame size may be different, for example when the
device encoder does not support the requested size.
To always handle this case properly, position and size the window only
once the first frame size is known.
PR #2947 <https://github.com/Genymobile/scrcpy/pull/2947>
Show the window only after the actual frame size is known (and if no
error has occurred).
This will allow to properly position and size the window when the size
of the first frame is different from the size initially announced by the
server.
PR #2947 <https://github.com/Genymobile/scrcpy/pull/2947>
Now that scrcpy attempts with a lower definition on any MediaCodec
error (or the user explicitly requests to disable auto-downsizing), the
suggestion is unnecessary.
PR #2947 <https://github.com/Genymobile/scrcpy/pull/2947>
Some devices are not able to encode at the device screen definition.
Instead of just failing, try with a lower definition on any MediaCodec
error.
PR #2947 <https://github.com/Genymobile/scrcpy/pull/2947>
The FFmpeg dependency is downloaded from two separate zipfiles.
Symlink include/ to expose everything from a single directory, to
simplify the meson script.
PR #2952 <https://github.com/Genymobile/scrcpy/pull/2952>
The fact that the current prebuilt FFmpeg is split into two separate
zipfiles is an implementation detail.
Use a single Makefile recipe for both files.
PR #2952 <https://github.com/Genymobile/scrcpy/pull/2952>
The components should be configurable independently of the global
scrcpy_options instance: their configuration could be provided
separately, like it is the case for example for some screen parameters.
For consistency, keyboard injection should not depend on scrcpy_options.
In relative mouse mode, the mouse pointer must be "captured" from the
computer.
Toggle (disable/enable) relative mouse mode using any of the hardcoded
capture keys:
- left-Alt
- left-Super
- right-Super
These capture keys do not conflict with shortcuts, since a shortcut is
always a combination of the MOD key and some other key, while the
capture key triggers an action only if it is pressed and released alone.
The relative mouse mode is also automatically enabled on any click in
the window, and automatically disabled on focus lost (it is possible to
lose focus even without the mouse).
The decision to not send motion events when no click is pressed is
specific to Android mouse injection. Other mouse processors (e.g. for
HID mouse) will need to receive all events.
The default mouse injection works in absolute mode: it forwards clicks
at a specific position on screen.
To support HID mouse, add a flag to indicate that the mouse processor
works in relative mode: it forwards mouse motion vectors, without any
absolute reference to the screen.
A scroll event might be produced when a mouse button is pressed (for
example when scrolling while selecting a text). For consistency, pass
the actual buttons state (instead of 0).
In practice, it seems that this use case does not work properly with
Android event injection, but it will work with HID mouse.
The input_manager is strongly tied to the screen, it could not work
independently of the specific screen implementation.
To implement a user-friendly HID mouse behavior, some SDL events
will need to be handled both by the screen and by the input manager. For
example, a click must typically be handled by the input_manager so that
it is forwarded to the device, but in HID mouse mode, the first click
should be handled by the screen to capture the mouse (enable relative
mouse mode).
Make the input_manager a descendant of the screen, so that the screen
decides what to do on SDL events.
Concretely, replace this structure hierarchy:
+- struct scrcpy
+- struct input_manager
+- struct screen
by this one:
+- struct scrcpy
+- struct screen
+- struct input_manager
This avoids to directly pass the options instance (which contains more
data than strictly necessary), and limit the number of parameters for
the init function.
Not all key processors support text injection (HID keyboard does not
support it).
Instead of providing a dummy op function, set it to NULL and check on
the caller side before calling it.
Pass scrcpy input events instead of SDL input events to mouse
processors.
These events represent exactly what mouse processors need, abstracted
from any visual orientation and scaling applied on the SDL window.
This makes the mouse processors independent of the "screen" instance,
and the implementation source code independent of the SDL API.
This aims to make the key/mouse processors independent of the "screen",
by processing scrcpy-specific input events instead of SDL events.
In particular, these scrcpy events are not impacted by any UI window
scaling or rotation (contrary to SDL events).
The input manager exposed functions taking an "actions" parameter,
containing a bitmask-OR of ACTION_UP and ACTION_DOWN.
But they are never called with both actions simultaneously anymore, so
simplify.
Refs 964b6d2243fa1921543e48810f3064b9bd2d50d1
Refs d0739911a3e413b70275ded3eef839f9dc57ba7a
This allows to report a meaningful error message if an unsupported
feature is used on an incompatible platform. This is consistent with the
behavior of -K/--hid-keyboard.
The "resize to fit" feature (MOD+w or double-click on black borders)
computed the "optimal size" using the same function computing the
initial window size on start.
However, on "resize to fit", only the black borders must be removed (the
content size must be preserved), so the display bounds must not be
considered.
Since commit 04267085441d6fcd05eff7df0118708f7622e237, the server is run
in a dedicated thread. For SDL, many signals, including SIGINT and
SIGTERM, are masked for new threads. As a result, if the adb server is
not already running, adb commands invoked by scrcpy will start an adb
server that ignores those signals and cannot be terminated at system
shutdown.
Fixes#2873 <https://github.com/Genymobile/scrcpy/issues/2873>
PR #2870 <https://github.com/Genymobile/scrcpy/pull/2870>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The icon will be associated to scrcpy.exe in the Windows explorer.
The .ico was created using imagemagick:
convert icon.png icon.ico
It is included as a binary for simplicity.
Refs #2815 <https://github.com/Genymobile/scrcpy/issues/2815>
Prefix the name of threads by "scrcpy-". This improves readability in
the output of `top -H` for example.
Limit the thread names to 16 bytes, because it is limited on some
platforms.
The sockets were never interrupted or closed by the client since recent
changes to run the server from a dedicated thread (see commit
04267085441d6fcd05eff7df0118708f7622e237).
As a side effect, the server could never terminate properly (it was
waiting on socket blocking calls), so it was always killed by the client
after the WATCHDOG_DELAY.
Interrupt the sockets on stop to give the servera chance to terminate
property, then close them.
If --no-control is enabled, then it is not necessary to create a second
communication socket between the client and the server.
This also facilitates the use of the server alone (without the client)
to receive only the raw video stream.
In ffmpeg/doc/APIchanges:
> 2016-04-21 - 7fc329e - lavc 57.37.100 - avcodec.h
> Add a new audio/video encoding and decoding API with decoupled input
> and output -- avcodec_send_packet(), avcodec_receive_frame(),
> avcodec_send_frame() and avcodec_receive_packet().
Refs de9b79ec2dd73e5442c6bf0161669bcd8ca7d5be
Refs #2862 <https://github.com/Genymobile/scrcpy/issues/2862>
Thank you clang:
../app/src/control_msg.c:45:5: warning: suspicious concatenation of
string literals in an array initialization; did you mean to separate
the elements with a comma? [-Wstring-concatenation]
"hover-exit",
^
In function ‘memcpy’,
inlined from ‘control_msg_serialize.constprop’ at ../app/src/control_msg.c:77:5,
inlined from ‘run_controller’ at ../app/src/controller.c:69:12:
/usr/include/x86_64-linux-gnu/bits/string_fortified.h:34:10:
warning: ‘__builtin___memcpy_chk’ writing 262138 bytes into a region
of size 262130 overflows the destination [-Wstringop-overflow=]
return __builtin___memcpy_chk (__dest, __src, __len, __bos0 (__dest));
Refs 901d8371655582b432d5d92430177a59df8058b9
PR #2859 <https://github.com/Genymobile/scrcpy/pull/2859>
Signed-off-by: Yu-Chen Lin <npes87184@gmail.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The PLATFORM variable is assigned either from $ANDROID_PLATFORM or gets
a default value (currently $PLATFORM_VERSION).
The check to use either dx (SDK < 31) or d8 (SDK >= 31) must be based on
the actual $PLATFORM, not the default $PLATFORM_VERSION.
Refs 52138fd9213216a21fbdcda4d430394d7c8f0979
Refs <57d30780dd/debian/patches/0002-Workaround-broken-script.patch>
PR #2850 <https://github.com/Genymobile/scrcpy/pull/2850>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Expose the inject input event mode so that it is possible to wait for
the events to be "finished". This will be necessary to read the
clipboard content only after the COPY or CUT key event is handled.
PR #2834 <https://github.com/Genymobile/scrcpy/pull/2834>
There was only two key injection modes:
- the default one
- the mode with --prefer-text enabled
To prepare the addition of another mode (--raw-key-events), use an enum
instead of a bool.
PR #2831 <https://github.com/Genymobile/scrcpy/pull/2831>
Expose an option to automatically configure and reconnect the device
over TCP/IP, to simplify wireless connection without using adb
explicitly.
There are two variants:
- If a destination address is provided, then scrcpy connects to this
address before starting. The device must listen on the given TCP port
(default is 5555).
- If no destination address is provided, then scrcpy attempts to find
the IP address of the current device (typically connected over USB),
enables TCP/IP mode, then connects to this address before starting.
PR #2827 <https://github.com/Genymobile/scrcpy/pull/2827>
This will allow to read the property "service.adb.tcp.port" to know if
the TCP/IP mode is enabled on the device, and which listening port is
used.
PR #2827 <https://github.com/Genymobile/scrcpy/pull/2827>
Depending on the platform and adb versions, the lines output by adb
could end with "\r\r\n". This util function helps to remove all trailing
'\r'.
PR #2827 <https://github.com/Genymobile/scrcpy/pull/2827>
"adb connect" always returns successfully (with exit code 0), even in
case of failure.
As a workaround, capture its output and check if it starts with
"connected".
PR #2827 <https://github.com/Genymobile/scrcpy/pull/2827>
In addition to disable stdout and stderr of the child process, add a
flag to disable the error log printed by scrcpy if the command failed.
This will we useful for commands which are expected to fail in some
cases (like "adb disconnect" if the device is not connected).
PR #2827 <https://github.com/Genymobile/scrcpy/pull/2827>
Let the caller decide if stdout and stderr must be inherited on process
creation, i.e. if stdout and stderr of the child process should be
printed in the scrcpy console.
This allows to get output and errors for specific adb commands depending
on the context.
PR #2827 <https://github.com/Genymobile/scrcpy/pull/2827>
If SOCK_CLOEXEC exists, then set the flag on socket creation.
Otherwise, use fcntl() (or SetHandleInformation() on Windows) to set the
flag afterwards.
This avoids the sockets to be inherited in child processes.
Refs #2783 <https://github.com/Genymobile/scrcpy/pull/2783>
The options values to configure the server were identified by their
command-line argument index. Now that there are a lot of arguments, many
of them being booleans, it became unreadable and error-prone.
Identify the arguments by a key string instead, and make them optional.
This will also simplify running the server manually for debugging.
To allow seamless copy-paste, on Ctrl+v, a SET_CLIPBOARD request is
performed before injecting Ctrl+v.
But when HID keyboard is enabled, the Ctrl+v injection is not sent on
the same channel as the clipboard request, so they are not serialized,
and may occur in any order. If Ctrl+v happens to be injected before the
new clipboard content is set, then the old content is pasted instead,
which is incorrect.
To minimize the probability of occurrence of the wrong order, a delay of
2 milliseconds was added before injecting Ctrl+v. Then 5ms. But even
with 5ms, the wrong behavior sometimes happens.
To handle it properly, add an acknowledgement mechanism, so that Ctrl+v
is injected over AOA only after the SET_CLIPBOARD request has been
performed and acknowledged by the server.
Refs e4163321f00bb3830c6049bdb6c1515e7cc668a0
Refs 45b0f8123a52f5c73a5860d616f4ceba2766ca6a
PR #2814 <https://github.com/Genymobile/scrcpy/pull/2814>
Pass the information that device clipboard has been set to the key
processor. This avoids the keyprocessor to "guess", and paves the way to
implement a proper acknowledgement mechanism.
PR #2814 <https://github.com/Genymobile/scrcpy/pull/2814>
Tunnel host and port are only meaningful in "adb forward" mode.
They indicate where the client must connect to communicate with the
server. In "adb reverse" mode, the client _listens_, it does not
_connect_.
Refs #2807 <https://github.com/Genymobile/scrcpy/pull/2807>
In "adb forward" mode, by default, scrcpy connects to localhost:PORT,
where PORT is the local port passed to "adb forward". This assumes that
the tunnel is established on the local host with a local adb server
(which is the common case).
For advanced usage, add --tunnel-host and --tunnel-port to force the
connection to a different destination.
Fixes#2801 <https://github.com/Genymobile/scrcpy/issues/2801>
PR #2807 <https://github.com/Genymobile/scrcpy/pull/2807>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Inconditionnally print the scrcpy version first, without using LOGI().
The log level is configured only after parsing the command line
parameters (it may be changed via -V), and command line parsing logs
should not appear before the scrcpy version.
The output of -h/--help was printed on stderr, although it is not an
error.
Printing on stdout allows to pipe the result directly:
scrcpy --help | less
Instead of (in bash):
scrcpy --help |& less
This allows to execute all adb commands with the specific -s parameter,
even if it is not provided by the user.
In practice, calling adb without -s works if there is exactly one device
connected. But some adb commands (for example "adb push" on drag & drop)
could be executed after another device is connected, so the actual
device serial must be known.
The interruptible version of the function to check process success
(sc_process_check_success_intr()) did not accept a close parameter to
avoid a race condition. But as the result, the processes were not closed
at all.
Add a close parameter, and close the process separately to avoid the
race condition.
Cleanup is used for some options like --show-touches to restore the
state on exit.
If the configuration fails, do not crash the whole process. Just log an
error.
PR #2802 <https://github.com/Genymobile/scrcpy/pull/2802>
Before Android 8, executing the "settings" command from a shell was
very slow (~1 second), because it spawned a new app_process to execute
Java code. Therefore, to access settings without performance issues,
scrcpy used private APIs to read from and write to settings.
However, since Android 12, this is not possible anymore, due to
permissions changes.
To make it work again, execute the "settings" command on Android 12 (or
on previous version if the other method failed). This method is faster
than before Android 8 (~100ms).
Fixes#2671 <https://github.com/Genymobile/scrcpy/issues/2671>
Fixes#2788 <https://github.com/Genymobile/scrcpy/issues/2788>
PR #2802 <https://github.com/Genymobile/scrcpy/pull/2802>
Settings read/write errors were silently ignored. Report them via a
SettingsException so that the caller can handle them.
This allows to log a proper error message, and will also allow to
fallback to a different settings method in case of failure.
PR #2802 <https://github.com/Genymobile/scrcpy/pull/2802>
Until now, the code that needed to read/write the Android settings had
to explicitly open and close a ContentProvider.
Wrap these details into a Settings class.
This paves the way to provide an alternative implementation of settings
read/write for Android >= 12.
PR #2802 <https://github.com/Genymobile/scrcpy/pull/2802>
Interrupt any blocking call on process terminated, like on
server_stop().
This allows to interrupt any blocking accept() with correct
synchronization without additional complexity.
Define server callbacks, start the server asynchronously and listen to
connection events to initialize scrcpy properly.
It will help to simplify the server code, and allows to run the UI event
loop while the server is connecting. In particular, this will allow to
receive SIGINT/Ctrl+C events during connection to interrupt immediately.
Currently, server_stop() is called from the same thread as
server_connect_to(), so interruption may never happen.
This is a step to prepare executing the server from a dedicated thread.
Use the option descriptions to generate the optstring and longopts
parameters for the getopt_long() command.
That way, the options are completely described in a single place.
On Linux, socket functions are unblocked by shutdown(), but on Windows
they are unblocked by closesocket().
Expose net_interrupt() and net_close() to abstract these differences:
- net_interrupt() calls shutdown() on Linux and closesocket() on
Windows (if not already called);
- net_close() calls close() on Linux and closesocket() on Windows (if
not already called).
This simplifies the server code, and prevents a data race on close
(reported by TSAN) on Linux (but does not fix it on Windows):
WARNING: ThreadSanitizer: data race (pid=836124)
Write of size 8 at 0x7ba0000000d0 by main thread:
#0 close ../../../../src/libsanitizer/tsan/tsan_interceptors_posix.cpp:1690 (libtsan.so.0+0x359d8)
#1 net_close ../app/src/util/net.c:211 (scrcpy+0x1c76b)
#2 close_socket ../app/src/server.c:330 (scrcpy+0x19442)
#3 server_stop ../app/src/server.c:522 (scrcpy+0x19e33)
#4 scrcpy ../app/src/scrcpy.c:532 (scrcpy+0x156fc)
#5 main ../app/src/main.c:92 (scrcpy+0x622a)
Previous read of size 8 at 0x7ba0000000d0 by thread T6:
#0 recv ../../../../src/libsanitizer/sanitizer_common/sanitizer_common_interceptors.inc:6603 (libtsan.so.0+0x4f4a6)
#1 net_recv ../app/src/util/net.c:167 (scrcpy+0x1c5a7)
#2 run_receiver ../app/src/receiver.c:76 (scrcpy+0x12819)
#3 <null> <null> (libSDL2-2.0.so.0+0x84f40)
When Ctrl+v is pressed, a control is sent to the device to set the
device clipboard before injecting Ctrl+v.
With the InputManager method, it is guaranteed that the device
synchronization is executed before handling Ctrl+v, since the commands
are executed on the device in sequence.
However, HID are injected from the computer, so there is no such
guarantee. As a consequence, on Android, Ctrl+v triggers a paste with
the old clipboard content.
To workaround the issue, wait a bit (2 milliseconds) from the AOA
thread before injecting the event, to leave enough time for the
clipboard to be set before injecting Ctrl+v.
When an AOA HID keyboard is registered, CAPSLOCK and NUMLOCK are both
disabled, regardless of the state of the computer keyboard.
To synchronize the state, on first key event, inject CAPSLOCK and/or
NUMLOCK if necessary.
The serial is necessary to find the correct Android device for AOA.
If it is not explicitly provided by the user via -s, then execute "adb
getserialno" to retrieve it.
The AVOutputFormat name is a comma-separated list. In theory, possible
names for V4L2 are:
- "video4linux2,v4l2"
- "v4l2,video4linux2"
- "v4l2"
- "video4linux2"
To find the muxer in all cases, we must request exactly one muxer name
at a time.
PR #2718 <https://github.com/Genymobile/scrcpy/pull/2718>
Co-authored-by: Romain Vimont <rom@rom1v.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The first frames are typically received and decoded with more delay than
the others, causing a wrong slope estimation on start.
To compensate, assume an initial slope of 1, then progressively use the
estimated slope.
To minimize latency (at the cost of jitter), scrcpy always displays a
frame as soon as it available, without waiting.
However, when recording (--record), it still writes the captured
timestamps to the output file, so that the recorded file can be played
correctly without jitter.
Some real-time use cases might benefit from adding a small latency to
compensate for jitter too. For example, few tens of seconds of latency
for live-streaming are not important, but jitter is noticeable.
Therefore, implement a buffering mechanism (disabled by default) to add
a configurable latency delay.
PR #2417 <https://github.com/Genymobile/scrcpy/issues/2417>
Currently, a frame is available to the consumer as soon as it is pushed
by the producer (which can detect if the previous frame is skipped).
Notify the new frames (and frame skipped) via callbacks instead.
This paves the way to add (optional) buffering, which will introduce a
delay between the time when the frame is produced and the time it is
available to be consumed.
The current video buffer only stores one pending frame.
In order to add a new buffering feature, move this part to a separate
"frame buffer". Keep the video_buffer, which currently delegates all its
calls to the frame_buffer.
To fix a data race, commit 5caeab5f6d12fe0fc69e708a829cfdbec3a401b4
called video_buffer_push() and video_buffer_consume() under the
v4l2_sink lock.
Instead, use the previous_skipped indication (initialized with video
buffer locked) to lock only for protecting the has_frame flag.
This enables the possibility for the video_buffer to notify new frames
via callbacks without lock inversion issues.
The function sc_cond_timedwait() accepted a parameter representing the
max duration to wait, because it internally uses SDL_CondWaitTimeout().
Instead, accept a deadline, to be consistent with
pthread_cond_timedwait().
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 21d206f360535047e16cc4ea7de889a55bd9444d 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>
Make the versionCode a decimal representation of the scrcpy version.
This will for example allow to correctly number the versionCode of
v1.17.1 after a v1.18 is released:
- v1.18 -> 11800
- v1.17.1 -> 11701
- v1.18.1 -> 11801
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 5af9d0ee0faddfca689f468c7d463c86233e8d93.
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.
Remove useless intermediate method with a "mode" parameter (it's always
called with the same mode).
This also avoids the need for a specific injectEventOnDisplay() method,
since it does not conflict with another injectEvent() method anymore.
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
ebccb9f6cc111e8acfbe10d656cac5c1f1b744a0.
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().
The (non-daemon) threads were not interrupted on video stream stopped,
leaving the server process alive.
Interrupt them to wake up their blocking call so that they terminate
properly.
Refs #1992 <https://github.com/Genymobile/scrcpy/issues/1992>
Before this change, release.sh built some native stuff, and
Makefile.CrossWindows built the Windows releases.
Instead, use a single release.make to build the whole release. It also
avoids to build the server one more time.
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);
This should increase the portability of bash scripts across various *nix
systems such as BSD-like distributions.
PR #1716 <https://github.com/Genymobile/scrcpy/pull/1716>
Signed-off-by: Luís Ferreira <contact@lsferreira.net>
Signed-off-by: Romain Vimont <rom@rom1v.com>
This commit in AOSP framework_base added a parameter "attributionTag" to
the call() method:
12ac3f406f%5E%21/#F17
As a consequence, the method did not exist, so scrcpy used the legacy
call() method (using the authority "unknown") as a fallback, which fails
for security reasons.
Fixes#1468 <https://github.com/Genymobile/scrcpy/issues/1468>
The expression port + 1 is promoted to int, but printed as uint16_t.
This is the same mistake fixed for a different log by
7eb16ce36458011d87972715f08bd9f03ff39807.
Refs #1726 <https://github.com/Genymobile/scrcpy/issues/1726>
The behavior of some methods do not depend on the user-provided options.
These methods can be static. This will allow to call them directly from
the cleanup process.
To avoid possible copy-paste loops between the computer and the device,
the device clipboard is not set if it already contains the expected
content.
But the condition was wrong: it was not set also if it was empty.
Refs 1223a72eb83ebafb878a25356250896c45b5cefa
Fixes#1658 <https://github.com/Genymobile/scrcpy/issues/1658>
A new "repeat" field has been added by
3c1ed5d86c38bcbd5353b0a7e6ef5653d6c44d0d, 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>
Do not explicitly set the clipboard text if it already contains the
expected content. This avoids possible copy-paste loops between the
computer and the device.
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.
When the client requests to set the clipboard, it may request to press
the PASTE key in addition. To be a bit generic, it was stored as a flag
in ControlMessage.java.
But flags suggest that it represents a bitwise union. Use a simple
boolean instead.
To power the device on, inject KEYCODE_WAKEUP to avoid a possible
race condition (the device might become off between the test
isScreenOn() and the POWER keycode injection).
In theory, this was expected to only happen when a touch event is sent
just before the device is rotated, but some devices do not respect the
encoding size, causing an unexpected mismatch.
Refs #1518 <https://github.com/Genymobile/scrcpy/issues/1518>
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 e40532a3761da8b3aef484f1d435ef5f50d94574
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.
The function must return a SDL_LogPriority, but returned an enum
sc_log_level.
(It was harmless because this specific return should never happen, as
asserted.)
This reverts commit 8c8649cfcd710859ce18eab557ed2af8cedb9a42.
I cannot reproduce the issue with Ctrl+Shift+o on any device, so in
practice it works, it's too bad to remove the feature for a random bug
on some Android versions on some devices.
In practice, it does not change anything (it just avoids a spurious
wake-up), but semantically, it makes no sense to call
pushClipboardText() with a null value.
Clean up has been broken by 3df63c579da7a35048458c6884f5f386154d2353.
The verbosity was correctly initialized for the Server process, but not
for the CleanUp process.
To avoid the problem, initialize a default log level.
Only the main injection method was exposed on Device, the convenience
methods were implemented in Controller.
For consistency, move them all to the Device class.
The Android device listens for clipboard changes to synchronize with the
computer clipboard.
However, if the change comes from scrcpy (for example via Ctrl+Shift+v),
do not notify the change.
Add a command-line option to force "adb forward", without attempting
"adb reverse" first.
This is especially useful for using SSH tunnels without enabling remote
port forwarding.
The verbosity was set either to info (in release mode) or debug (in
debug mode).
Add a command-line argument to change it, so that users can enable debug
logs using the release:
scrcpy -Vdebug
The field lock_video_orientation may only take values between -1 and 3
(included). But the compiler may trigger a warning on the sprintf()
call, because its type could represent values which could overflow the
string (like "-128"):
> warning: ‘%i’ directive writing between 1 and 4 bytes into a region of
> size 3 [-Wformat-overflow=]
Increase the buffer size to remove the warning.
Trilinear filtering can currently only be enabled for OpenGL renderers.
Do not print a warning if the renderer is not OpenGL, as it can confuses
users, while nothing is wrong.
On macOS with renderer "metal", HiDPI scaling may be incorrect on
initialization when several displays are connected.
Resetting the window size fixes the problem.
Refs #15 <https://github.com/Genymobile/scrcpy/issues/15>
Position and scale the content "manually" instead of relying on the
renderer "logical size".
This avoids possible rounding differences between the computed window
size and the content size, causing one row or column of black pixels on
the bottom or on the right.
This also avoids HiDPI scale issues, by computing the scaling manually.
This will also enable to draw items at their expected size on the screen
(unscaled).
Fixes#15 <https://github.com/Genymobile/scrcpy/issues/15>
In maximized state (but not fullscreen), it was possible to resize to
fit the device screen (with Ctrl+x or double-clicking on black borders).
This caused problems on macOS with the "expand to fullscreen" feature,
which behaves like a fullscreen mode but is seen as maximized by SDL.
In that state, resizing to fit causes unexpected results.
To keep the behavior consistent on all platforms, just disable "resize
to fit" when the window is maximized.
On Windows, in maximized+fullscreen state, disabling fullscreen mode
unexpectedly triggers the "restored" then "maximized" events, leaving
the window in a weird state (maximized according to the events, but not
maximized visually).
Moreover, apply_pending_resize() asserts that fullscreen is disabled.
To avoid the issue, if fullscreen is set, just ignore the "restored"
event.
If the content size changes (due to rotation for example) while the
window is maximized or fullscreen, the resize must be applied once
fullscreen and maximized are disabled.
The previous strategy consisted in storing the windowed size, computing
the target size on rotation, and applying it on window restoration. But
tracking the windowed size (while ignoring the non-windowed size) was
tricky, due to unspecified order of SDL events (e.g. size changes can be
notified before "maximized" events), race conditions when reading window
flags, different behaviors on different platforms...
To simplify the whole resize management, store the old content size (the
frame size, possibly rotated) when it changes while the window is
maximized or fullscreen, so that the new optimal size can be computed on
window restoration.
By default, Ctrl+C just kills the process on Windows. This caused
corrupted video files on recording.
Handle Ctrl+C properly to clean up properly.
Fixes#818 <https://github.com/Genymobile/scrcpy/issues/818>
Scroll events were sent with a mouse input device. When scrolling on a
list, this could cause the whole list to be focused, and drawn with the
focus color as background.
Send scroll events with a touchscreen input device instead (like motion
events).
Fixes#1362 <https://github.com/Genymobile/scrcpy/issues/1362>
Commit 927d655ff66fa439ccd27e054ae6d54a651640cc removed the
iFrameInternal field and constructor argument.
Also remove it from createFormat() arguments.
Now that the server can access the Android settings and clean up
properly, handle the "show touches" option from the server.
The initial state is now correctly restored, even on device
disconnection.
Expose methods to access the Android settings using private APIs.
This allows to read and write settings values immediately without
starting a new process to call "settings".
The window dimensions are integers, so resizing to fit the content may
not be exact.
When computing the optimal size, it could cause to reduce alternatively
the width and height by few pixels, making the "optimal size" unstable.
To avoid this problem, check if the optimal size is already correct
either by keeping the width or the height.
Move the window-to-frame coordinates conversion from the input manager
to the screen.
This will allow to apply more screen-related transformations without
impacting the input manager.
Debian buster (stable) provides Meson 0.49, which is also available in
stretch (oldstable) backports. It's time to abandon Meson 0.37.
Ref: 20b3f101a40cd7455cc5b41e381291504deec5ba
Some compilers warns on uninitialized value in impossible case:
warning: variable 'result' is used uninitialized whenever switch
default is taken [-Wsometimes-uninitialized]
A double-click outside the device content (in the black borders) resizes
so that black borders are removed. But the display rotation was not
taken into account to detect the content.
Use the content size instead of the frame size to fix the issue.
Ref: <https://github.com/Genymobile/scrcpy/issues/898#issuecomment-610993695>
Add Ctrl+Left and Ctrl+Right shortcuts to rotate the display (the
content of the scrcpy window).
Contrary to --lock-video-orientation, the rotation has no impact on
recording, and can be changed dynamically (and immediately).
Fixes#218 <https://github.com/Genymobile/scrcpy/issues/218>
People following default build instructions can be caught off guard by
seeing the build artifacts in the git tree.
Signed-off-by: Harsh Shandilya <me@msfjarvis.dev>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The server may die before connecting to the client. In that case, the
client was blocked indefinitely (until Ctrl+C) on accept().
To avoid the problem, close the server socket once the server process is
dead.
There are a lot of "magic numbers" that we really don't want to extract
as a constant.
Until now, many @SuppressWarnings annotations were added, but it makes
no sense to check for magic number if we silent the warnings everywhere.
Centralize video size management in ScreenInfo.
This allows to always send the correct initial video size to the client
if the video orientation is locked.
Binaries created with MinGW (even a simple Hello World) are detected as
malware by some anti-virus. For some reason, only the 32 bits version of
scrcpy is impacted.
Since users should use the 64 bits version by default anyway, remove the
link to the 32 bits version from the main page.
The 32 bits release is still available in the "releases" tab.
See <https://github.com/Genymobile/scrcpy/issues/1102>
Some workarounds are needed on some devices. But applying them may cause
exceptions on other devices, where they are not necessary anyway.
Do not report these errors in release builds.
Closes#994 <https://github.com/Genymobile/scrcpy/issues/994>
Accept a range of ports to listen to, so that it does not fail if
another instance of scrcpy is currently starting.
The range can be passed via the command line:
scrcpy -p 27183:27186
scrcpy -p 27183 # implicitly 27183:27183, as before
The default is 27183:27199.
Closes#951 <https://github.com/Genymobile/scrcpy/issues/951>
Start the server socket in enable_tunnel() directly.
For the caller point of view, enabling the tunnel opens a port (either
the server socket locally or the "adb forward" process).
The platform-specific code for net.c was implemented in sys/*/net.c.
But the differences are quite limited, so use ifdef-blocks in the single
net.c instead.
The file 'E:\安安\scrcpy-win64-v1.12.1-1-g31bd950\scrcpy-server'
exists, however, it will show msg as follow:
INFO: scrcpy 1.12.1 <https://github.com/Genymobile/scrcpy>
stat: No such file or directory
ERROR: 'E:\安安\scrcpy-win64-v1.12.1-1-g31bd950\scrcpy-server' does
not exist or is not a regular file
Press any key to continue...
This patch fixes it.
Signed-off-by: Yu-Chen Lin <npes87184@gmail.com>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The function get_server_path() sometimes returned an owned string,
sometimes a non-owned string.
Always return an allocated (owned) string, and free it after usage.
A proper solution could be to use "long long" instead (guaranteed to be
at least 64 bits), but it adds its own problems (e.g. "%lld" is not
supported as a printf format on all platforms).
In practice, we don't need such high values, so keep it simple.
Fixes#995 <https://github.com/Genymobile/scrcpy/issues/995>
If SCRCPY_SERVER_PATH points to a directory, then a directory will be
pushed to /data/local/tmp/scrcpy-server.jar.
When executing it, app_process will just abort and leave the directory
on the device, causing scrcpy to always fail.
To avoid the problem, check that the server is a regular file before
pushing it.
Closes#956 <https://github.com/Genymobile/scrcpy/issues/956>
Internally, a failure to invoke a method via reflection was partially
managed using exceptions, partially using a null return value.
Handle all errors at the same place, by not catching
NoSuchMethodException too early.
> Movements down (scroll backward) generate negative y values and up
> (scroll forward) generate positive y values.
> If direction is SDL_MOUSEWHEEL_FLIPPED the values in x and y will be
> opposite. Multiply by -1 to change them back.
<https://wiki.libsdl.org/SDL_MouseWheelEvent#Remarks>
The x and y values already take the scrolling configuration into
account. Reversing the values when the direction is flipped cancels the
scrolling configuration.
Therefore, just ignore the direction field.
Fixes <https://github.com/Genymobile/scrcpy/issues/966>
When an ApplicationInfo is set (commit
90293240cc622bb58cb1de741f86cbc0889c03e8), some devices (Nvidia Shield
TV) attempt to access the Application object, causing a
NullPointerException.
As a workaround, initialize an Application object.
Fixes <https://github.com/Genymobile/scrcpy/issues/940>
Commit 3da95b52bd21ae0f4d81ee933caea63b78191deb renamed
'scrcpy-server.jar' to 'scrcpy-server' to avoid issues on the client
side.
However, removing the extension may cause issues with app_process, so
restore the extension only on the device side.
Fixes <https://github.com/Genymobile/scrcpy/issues/944>
Since commit b963a3b9d56f744cceba2e19cba3f9fef858c058, the server uses
BuildConfig.VERSION_NAME.
Generate this field manually for building without gradle.
The KEY_FRAME_RATE parameter value is necessary for the configuration of
the encoder, but its actual value does not impact the frame rate (only
resources used by the encoder).
Therefore, it's an internal detail and should not be exposed by the
ScreenEncoder class.
- [ ] I have checked that a similar [feature request](https://github.com/Genymobile/scrcpy/issues?q=is%3Aopen+is%3Aissue+label%3A%22feature+request%22) does not already exist.
**Is your feature request related to a problem? Please describe.**
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
**Describe the solution you'd like**
A clear and concise description of what you want to happen.
**Describe alternatives you've considered**
A clear and concise description of any alternative solutions or features you've considered.
**Additional context**
Add any other context or screenshots about the feature request here.
On some devices (especially Xiaomi), you might get the following error:
```
java.lang.SecurityException: Injecting input events requires the caller (or the source of the instrumentation, if any) to have the INJECT_EVENTS permission.
```
In that case, you need to enable [an additional option][control] `USB debugging
(Security Settings)` (this is an item different from `USB debugging`) to control
it using a keyboard and mouse. Rebooting the device is necessary once this
<pathstyle="opacity:0.2"d="m 16.846877,12 c -1.116351,0 -2.227419,0.912229 -2.015075,2 l 3.122973,16 -5.596557,11.109375 C 11.959876,41.871734 11.885244,42.336988 12.177176,43 c 0.278672,0.632897 0.998812,1 1.747448,1 H 24 34.075375 c 0.748637,0 1.468777,-0.367103 1.747448,-1 0.291932,-0.663012 0.217302,-1.128266 -0.181041,-1.890625 L 30.045225,30 33.168198,14 c 0.212344,-1.087771 -0.898724,-2 -2.015075,-2 H 24 Z"/>
<pathstyle="fill:#cccccc"d="m 16.846877,11 c -1.116351,0 -2.227419,0.912229 -2.015075,2 l 3.122973,16 -5.596557,11.109375 C 11.959876,40.871734 11.885244,41.336988 12.177176,42 c 0.278672,0.632897 0.998812,1 1.747448,1 H 24 34.075375 c 0.748637,0 1.468777,-0.367103 1.747448,-1 0.291932,-0.663012 0.217302,-1.128266 -0.181041,-1.890625 L 30.045225,29 33.168198,13 c 0.212344,-1.087771 -0.898724,-2 -2.015075,-2 H 24 Z"/>
<pathstyle="fill:#e4e4e4"d="m 4,33 v 2 c 0,1.108 0.892,2 2,2 h 36 c 1.108,0 2,-0.892 2,-2 v -2 z"/>
<pathstyle="opacity:0.1"d="m 11.494141,15 a 1.5,1.5 0 0 0 -0.832032,0.255859 1.5,1.5 0 0 0 -0.40625,2.082032 l 3.13086,4.654297 C 10.404945,24.606192 8.4012371,28.299159 8.0019531,32.460938 7.9284599,34.000879 9.5546875,34 9.5546875,34 H 38.40625 c 0,0 1.672856,-3.38e-4 1.591797,-1.617188 -0.416529,-4.131451 -2.415618,-7.796833 -5.380859,-10.394531 l 3.126953,-4.65039 a 1.5,1.5 0 0 0 -0.40625,-2.082032 1.5,1.5 0 0 0 -1.125,-0.228515 1.5,1.5 0 0 0 -0.957032,0.634765 l -3.072265,4.566407 C 29.78649,18.814887 26.990024,18 24.001953,18 c -2.989385,0 -5.786177,0.815488 -8.183594,2.230469 l -3.074218,-4.56836 A 1.5,1.5 0 0 0 11.787109,15.027344 1.5,1.5 0 0 0 11.494141,15 Z"/>
<pathstyle="fill:#077063"d="M 6,5 C 4.892,5 4,5.892 4,7 V 33 H 44 V 7 C 44,5.892 43.108,5 42,5 Z"/>
<pathstyle="opacity:0.1;fill:#ffffff"d="M 6,5 C 4.892,5 4,5.892 4,7 V 8 C 4,6.892 4.892,6 6,6 h 36 c 1.108,0 2,0.892 2,2 V 7 C 44,5.892 43.108,5 42,5 Z"/>
<pathstyle="fill:#30dd81"d="m 24.002386,17.000034 c -8.355868,0 -15.2214979,6.346843 -15.9999669,14.460906 C 7.9289259,33.000882 9.5544999,33 9.5544999,33 H 38.406003 c 0,0 1.672201,-3.35e-4 1.591142,-1.617185 C 39.182807,23.305596 32.331836,17.000034 24.002386,17.000034 Z"/>
Duplicate audio (capture and keep playing on the device).
This feature is only available with --audio-source=playback.
.TP
.BI"\-\-audio\-encoder "name
Use a specific MediaCodec audio encoder (depending on the codec provided by \fB\-\-audio\-codec\fR).
The available encoders can be listed by \fB\-\-list\-encoders\fR.
.TP
.BI"\-\-audio\-source "source
Select the audio source (output, mic or playback).
The "output" source forwards the whole audio output, and disables playback on the device.
The "playback" source captures the audio playback (Android apps can opt-out, so the whole output is not necessarily captured).
The "mic" source captures the microphone.
Default is output.
.TP
.BI"\-\-audio\-output\-buffer "ms
Configure the size of the SDL audio output buffer (in milliseconds).
If you get "robotic" audio playback, you should test with a higher value (10). Do not change this setting otherwise.
Default is 5.
.TP
.BI"\-b, \-\-video\-bit\-rate "value
Encode the video at the given bit rate, expressed in bits/s. Unit suffixes are supported: '\fBK\fR' (x1000) and '\fBM\fR' (x1000000).
Default is 8M (8000000).
.TP
.BI"\-\-camera\-ar "ar
Select the camera size by its aspect ratio (+/- 10%).
Possible values are "sensor" (use the camera sensor aspect ratio), "\fInum\fR:\fIden\fR" (e.g. "4:3") and "\fIvalue\fR" (e.g. "1.6").
.TP
.BI"\-\-camera\-facing "facing
Select the device camera by its facing direction.
Possible values are "front", "back" and "external".
.TP
.BI"\-\-camera\-fps "fps
Specify the camera capture frame rate.
If not specified, Android's default frame rate (30 fps) is used.
.TP
.B\-\-camera\-high\-speed
Enable high-speed camera capture mode.
This mode is restricted to specific resolutions and frame rates, listed by \fB\-\-list\-camera\-sizes\fR.
.TP
.BI"\-\-camera\-id "id
Specify the device camera id to mirror.
The available camera ids can be listed by \fB\-\-list\-cameras\fR.
.TP
.BI"\-\-camera\-size "width\fRx\fIheight
Specify an explicit camera capture size.
.TP
.BI"\-\-capture\-orientation "value
Possible values are 0, 90, 180, 270, flip0, flip90, flip180 and flip270, possibly prefixed by '@'.
The number represents the clockwise rotation in degrees; the "flip" keyword applies a horizontal flip before the rotation.
If a leading '@' is passed (@90) for display capture, then the rotation is locked, and is relative to the natural device orientation.
If '@' is passed alone, then the rotation is locked to the initial device orientation.
Default is 0.
.TP
.BI"\-\-crop "width\fR:\fIheight\fR:\fIx\fR:\fIy
Crop the device screen on the server.
The values are expressed in the device natural orientation (typically, portrait for a phone, landscape for a tablet). Any
.B\-\-max\-size
value is computed on the cropped size.
The values are expressed in the device natural orientation (typically, portrait for a phone, landscape for a tablet).
.TP
.B\-d,\-\-select\-usb
Use USB device (if there is exactly one, like adb -d).
Also see \fB\-e\fR (\fB\-\-select\-tcpip\fR).
.TP
.BI"\-\-disable\-screensaver"
Disable screensaver while scrcpy is running.
.TP
.BI"\-\-display\-id "id
Specify the device display id to mirror.
The available display ids can be listed by \fB\-\-list\-displays\fR.
Default is 0.
.TP
.BI"\-\-display\-orientation "value
Set the initial display orientation.
Possible values are 0, 90, 180, 270, flip0, flip90, flip180 and flip270. The number represents the clockwise rotation in degrees; the "flip" keyword applies a horizontal flip before the rotation.
Default is 0.
.TP
.B\-e,\-\-select\-tcpip
Use TCP/IP device (if there is exactly one, like adb -e).
Also see \fB\-d\fR (\fB\-\-select\-usb\fR).
.TP
.B\-f,\-\-fullscreen
Start in fullscreen.
.TP
.BI"\-F, \-\-record\-format "format
Force recording format (either mp4 or mkv).
.B\-\-force\-adb\-forward
Do not attempt to use "adb reverse" to connect to the device.
.TP
.B\-G
Same as \fB\-\-gamepad=uhid\fR, or \fB\-\-keyboard=aoa\fR if \fB\-\-otg\fR is set.
.TP
.BI"\-\-gamepad "mode
Select how to send gamepad inputs to the device.
Possible values are "disabled", "uhid" and "aoa":
- "disabled" does not send gamepad inputs to the device.
- "uhid" simulates physical HID gamepads using the Linux HID kernel module on the device.
- "aoa" simulates physical HID gamepads using the AOAv2 protocol. It may only work over USB.
Also see \fB\-\-keyboard\f and R\fB\-\-mouse\fR.
.TP
.B\-h,\-\-help
Print this help.
.TP
.B\-K
Same as \fB\-\-keyboard=uhid\fR, or \fB\-\-keyboard=aoa\fR if \fB\-\-otg\fR is set.
.TP
.BI"\-\-keyboard "mode
Select how to send keyboard inputs to the device.
Possible values are "disabled", "sdk", "uhid" and "aoa":
- "disabled" does not send keyboard inputs to the device.
- "sdk" uses the Android system API to deliver keyboard events to applications.
- "uhid" simulates a physical HID keyboard using the Linux HID kernel module on the device.
- "aoa" simulates a physical HID keyboard using the AOAv2 protocol. It may only work over USB.
For "uhid" and "aoa", the keyboard layout must be configured (once and for all) on the device, via Settings -> System -> Languages and input -> Physical keyboard. This settings page can be started directly using the shortcut MOD+k (except in OTG mode), or by executing:
adb shell am start -a android.settings.HARD_KEYBOARD_SETTINGS
This option is only available when the HID keyboard is enabled (or a physical keyboard is connected).
Also see \fB\-\-mouse\fR and \fB\-\-gamepad\fR.
.TP
.B\-\-kill\-adb\-on\-close
Kill adb when scrcpy terminates.
.TP
.B\-\-legacy\-paste
Inject computer clipboard text as a sequence of key events on Ctrl+v (like MOD+Shift+v).
This is a workaround for some devices not behaving as expected when setting the device clipboard programmatically.
.TP
.B\-\-list\-apps
List Android apps installed on the device.
.TP
.B\-\-list\-camera\-sizes
List the valid camera capture sizes.
.TP
.B\-\-list\-cameras
List cameras available on the device.
.TP
.B\-\-list\-encoders
List video and audio encoders available on the device.
.TP
.B\-\-list\-displays
List displays available on the device.
.TP
.BI"\-m, \-\-max\-size "value
Limit both the width and height of the video to \fIvalue\fR. The other dimension is computed so that the device aspect\-ratio is preserved.
Default is 0 (unlimited).
.TP
.B\-M
Same as \fB\-\-mouse=uhid\fR, or \fB\-\-mouse=aoa\fR if \fB\-\-otg\fR is set.
.TP
.BI"\-\-max\-fps "value
Limit the framerate of screen capture (officially supported since Android 10, but may work on earlier versions).
.TP
.BI"\-\-mouse "mode
Select how to send mouse inputs to the device.
Possible values are "disabled", "sdk", "uhid" and "aoa":
- "disabled" does not send mouse inputs to the device.
- "sdk" uses the Android system API to deliver mouse events to applications.
- "uhid" simulates a physical HID mouse using the Linux HID kernel module on the device.
- "aoa" simulates a physical mouse using the AOAv2 protocol. It may only work over USB.
In "uhid" and "aoa" modes, the computer mouse is captured to control the device directly (relative mouse mode).
LAlt, LSuper or RSuper toggle the capture mode, to give control of the mouse back to the computer.
Also see \fB\-\-keyboard\fR and \fB\-\-gamepad\fR.
.TP
.BI"\-\-mouse\-bind "xxxx[:xxxx]
Configure bindings of secondary clicks.
The argument must be one or two sequences (separated by ':') of exactly 4 characters, one for each secondary click (in order: right click, middle click, 4th click, 5th click).
The first sequence defines the primary bindings, used when a mouse button is pressed alone. The second sequence defines the secondary bindings, used when a mouse button is pressed while the Shift key is held.
If the second sequence of bindings is omitted, then it is the same as the first one.
Each character must be one of the following:
- '+': forward the click to the device
- '-': ignore the click
- 'b': trigger shortcut BACK (or turn screen on if off)
Create a new display with the specified resolution and density. If not provided, they default to the main display dimensions and DPI.
Default is 27183.
Examples:
\-\-new\-display=1920x1080
\-\-new\-display=1920x1080/420
\-\-new\-display # main display size and density
\-\-new\-display=/240 # main display size and 240 dpi
.TP
.B\-\-no\-audio
Disable audio forwarding.
.TP
.B\-\-no\-audio\-playback
Disable audio playback on the computer.
.TP
.B\-\-no\-cleanup
By default, scrcpy removes the server binary from the device and restores the device state (show touches, stay awake and power mode) on exit.
This option disables this cleanup.
.TP
.B\-\-no\-clipboard\-autosync
By default, scrcpy automatically synchronizes the computer clipboard to the device clipboard before injecting Ctrl+v, and the device clipboard to the computer clipboard whenever it changes.
This option disables this automatic synchronization.
.TP
.B\-\-no\-downsize\-on\-error
By default, on MediaCodec error, scrcpy automatically tries again with a lower definition.
This option disables this behavior.
.TP
.B\-\-no\-key\-repeat
Do not forward repeated key events when a key is held down.
.TP
.B\-\-no\-mipmaps
If the renderer is OpenGL 3.0+ or OpenGL ES 2.0+, then mipmaps are automatically generated to improve downscaling quality. This option disables the generation of mipmaps.
.TP
.B\-\-no\-mouse\-hover
Do not forward mouse hover (mouse motion without any clicks) events.
.TP
.B\-\-no\-power\-on
Do not power on the device on start.
.TP
.B\-\-no\-vd\-destroy\-content
Disable virtual display "destroy content on removal" flag.
With this option, when the virtual display is closed, the running apps are moved to the main display rather than being destroyed.
.TP
.B\-\-no\-vd\-system\-decorations
Disable virtual display system decorations flag.
.TP
.B\-\-no\-video
Disable video forwarding.
.TP
.B\-\-no\-video\-playback
Disable video playback on the computer.
.TP
.B\-\-no\-window
Disable scrcpy window. Implies --no-video-playback and --no-control.
.TP
.BI"\-\-orientation "value
Same as --display-orientation=value --record-orientation=value.
.TP
.B\-\-otg
Run in OTG mode: simulate physical keyboard and mouse, as if the computer keyboard and mouse were plugged directly to the device via an OTG cable.
In this mode, adb (USB debugging) is not necessary, and mirroring is disabled.
LAlt, LSuper or RSuper toggle the mouse capture mode, to give control of the mouse back to the computer.
If any of \fB\-\-hid\-keyboard\fR or \fB\-\-hid\-mouse\fR is set, only enable keyboard or mouse respectively, otherwise enable both.
It may only work over USB.
See \fB\-\-keyboard\fR, \fB\-\-mouse\fR and \fB\-\-gamepad\fR.
.TP
.BI"\-p, \-\-port "port\fR[:\fIport\fR]
Set the TCP port (range) used by the client to listen.
Default is 27183:27199.
.TP
\fB\-\-pause\-on\-exit\fR[=\fImode\fR]
Configure pause on exit. Possible values are "true" (always pause on exit), "false" (never pause on exit) and "if-error" (pause only if an error occurred).
This is useful to prevent the terminal window from automatically closing, so that error messages can be read.
Default is "false".
Passing the option without argument is equivalent to passing "true".
.TP
.B\-\-power\-off\-on\-close
Turn the device screen off when closing scrcpy.
.TP
.B\-\-prefer\-text
@ -68,11 +436,15 @@ Inject alpha characters and space as text events instead of key events.
This avoids issues when combining multiple keys to enter special characters,
but breaks the expected behavior of alpha keys in games (typically WASD).
.TP
.B"\-\-print\-fps
Start FPS counter, to print framerate logs to the console. It can be started or stopped at any time with MOD+i.
.TP
.BI"\-\-push\-target "path
Set the target directory for pushing files to the device by drag & drop. It is passed as\-is to "adb push".
Default is "/sdcard/".
Default is "/sdcard/Download/".
.TP
.BI"\-r, \-\-record "file
@ -80,12 +452,36 @@ Record screen to
.IRfile.
The format is determined by the
.B\-F/\-\-record\-format
option if set, or by the file extension (.mp4 or .mkv).
.B\-\-record\-format
option if set, or by the file extension.
.TP
.B\-\-render\-expired\-frames
By default, to minimize latency, scrcpy always renders the last available decoded frame, and drops any previous ones. This flag forces to render all frames, at a cost of a possible increased latency.
.B\-\-raw\-key\-events
Inject key events for all input keys, and ignore text events.
.TP
.BI"\-\-record\-format "format
Force recording format (mp4, mkv, m4a, mka, opus, aac, flac or wav).
.TP
.BI"\-\-record\-orientation "value
Set the record orientation.
Possible values are 0, 90, 180 and 270. The number represents the clockwise rotation in degrees.
Default is 0.
.TP
.BI"\-\-render\-driver "name
Request SDL to use the given render driver (this is just a hint).
Supported names are currently "direct3d", "opengl", "opengles2", "opengles", "metal" and "software".
<https://wiki.libsdl.org/SDL_HINT_RENDER_DRIVER>
.TP
.B\-\-require\-audio
By default, scrcpy mirrors only the video if audio capture fails on the device. This option makes scrcpy fail if audio is enabled but does not work.
.TP
.BI"\-s, \-\-serial "number
@ -95,113 +491,325 @@ The device serial number. Mandatory only if several devices are connected to adb
.B\-S,\-\-turn\-screen\-off
Turn the device screen off immediately.
.TP
.BI"\-\-shortcut\-mod "key\fR[+...]][,...]
Specify the modifiers to use for scrcpy shortcuts. Possible keys are "lctrl", "rctrl", "lalt", "ralt", "lsuper" and "rsuper".
Several shortcut modifiers can be specified, separated by ','.
For example, to use either LCtrl or LSuper for scrcpy shortcuts, pass "lctrl,lsuper".
Default is "lalt,lsuper" (left-Alt or left-Super).
.TP
.BI"\-\-start\-app "name
Start an Android app, by its exact package name.
Add a '?' prefix to select an app whose name starts with the given name, case-insensitive (retrieving app names on the device may take some time):
scrcpy --start-app=?firefox
Add a '+' prefix to force-stop before starting the app:
Enable "show touches" on start, restore the initial value on exit.
It only shows physical touches (not clicks from scrcpy).
.TP
.B\-T,\-\-always\-on\-top
Make scrcpy window always on top (above other windows).
.BI"\-\-tcpip\fR[=[+]\fIip\fR[:\fIport\fR]]
Configure and connect the device over TCP/IP.
If a destination address is provided, then scrcpy connects to this address before starting. The device must listen on the given TCP port (default is 5555).
If no destination address is provided, then scrcpy attempts to find the IP address and adb port of the current device (typically connected over USB), enables TCP/IP mode if necessary, then connects to this address before starting.
Prefix the address with a '+' to force a reconnection.
.TP
.BI"\-\-time\-limit "seconds
Set the maximum mirroring time, in seconds.
.TP
.BI"\-\-tunnel\-host "ip
Set the IP address of the adb tunnel to reach the scrcpy server. This option automatically enables \fB\-\-force\-adb\-forward\fR.
Default is localhost.
.TP
.BI"\-\-tunnel\-port "port
Set the TCP port of the adb tunnel to reach the scrcpy server. This option automatically enables \fB\-\-force\-adb\-forward\fR.
Default is 0 (not forced): the local port used for establishing the tunnel will be used.
.TP
.B\-v,\-\-version
Print the version of scrcpy.
.TP
.BI\-\-window\-title"text
.BI"\-V, \-\-verbosity "value
Set the log level ("verbose", "debug", "info", "warn" or "error").
Default is "info" for release builds, "debug" for debug builds.
.TP
.BI"\-\-v4l2-sink "/dev/videoN
Output to v4l2loopback device.
.TP
.BI"\-\-v4l2-buffer "ms
Add a buffering delay (in milliseconds) before pushing frames. This increases latency to compensate for jitter.
This option is similar to \fB\-\-video\-buffer\fR, but specific to V4L2 sink.
Default is 0 (no buffering).
.TP
.BI"\-\-video\-buffer "ms
Add a buffering delay (in milliseconds) before displaying video frames.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
