Turn screen off when closing scrcpy #824

npes87184 · 2019-09-28T03:47:33Z

rom1v

Thank you for the PR.

I think this should be an option: I often plug my device just to type or copy-paste things from the computer, I don't want it to be turned off just because I close scrcpy.

rom1v · 2019-09-28T10:18:21Z

app/src/control_msg.h

@@ -25,6 +25,7 @@ enum control_msg_type {
    CONTROL_MSG_TYPE_GET_CLIPBOARD,
    CONTROL_MSG_TYPE_SET_CLIPBOARD,
    CONTROL_MSG_TYPE_SET_SCREEN_POWER_MODE,
+    CONTROL_MSG_TYPE_TURN_SCREEN_OFF,


We need to distinguish between "turn screen off but keep device on" (like here) and "turn screen off and device off (standby)". (There is already some naming confusion in the current codebase, if you have any suggestion…)

Maybe "lock screen" vs "turn screen off"?

rom1v · 2019-09-28T10:21:55Z

I merged the first commit fixing the log 👍

npes87184 · 2019-09-29T02:26:42Z

I have pushed v2:

Let this feature be optional.
use "lock screen"

However, I found that server may be killed before screen off.
We may need to introduce some ack mechanism from server to client.

rom1v · 2019-10-31T20:11:34Z

However, I found that server may be killed before screen off.
We may need to introduce some ack mechanism from server to client.

In my TODO list, I plan to implement something similar to #679 to be able to cleanup properly:

The server "forks" itself for full life cycle control (adb may kill it instead preventing required cleanup).

This will be a requirement to merge this PR.

rom1v · 2020-08-03T17:26:55Z

Now that there is a real cleanup, you could implement it 😉

(I suggest --power-off-on-close)

rom1v · 2021-02-15T20:40:20Z

@npes87184 I think several people are interested:

Power off screen on close #714
Add posibilty to keep screen off after disconection #1764
How can I get scrcpy to open automatically as I connect phone to usb? Also screen off upon exit and allow access to phone data on the phone upon connection.. #2103
https://www.reddit.com/r/scrcpy/comments/ljzpdd/turning_off_the_display_automatically_upon/
😉

npes87184 · 2021-02-16T00:11:18Z

Ok, I will look it.

npes87184 · 2021-02-16T05:40:01Z

Hi, @rom1v

There is a problem about power off screen in CleanUp process. In this time, injectEvent will consider about displayId to inject to correct target. However, it is diffiuclt to allocate Device in the process of CleanUp. As an result, I cannot trigger power off screen in that stage. Do you have any good advice?

Maybe CleanUp process should have the same environment just likes the main server?

Thank you.

rom1v · 2021-02-16T10:47:16Z

As a quick-and-dirty solution, maybe you could expose a static injectEvent() taking a displayId:

diff --git a/server/src/main/java/com/genymobile/scrcpy/Device.java b/server/src/main/java/com/genymobile/scrcpy/Device.java
index a8fdf677..9fb365ff 100644
--- a/server/src/main/java/com/genymobile/scrcpy/Device.java
+++ b/server/src/main/java/com/genymobile/scrcpy/Device.java
@@ -157,16 +157,20 @@ public final class Device {
         return supportsInputEvents;
     }
 
+    public static boolean injectEvent(InputEvent inputEvent, int mode, int displayId) {
+        if (displayId != 0 && !InputManager.setDisplayId(inputEvent, displayId)) {
+            return false;
+        }
+
+        return SERVICE_MANAGER.getInputManager().injectInputEvent(inputEvent, mode);
+    }
+
     public boolean injectEvent(InputEvent inputEvent, int mode) {
         if (!supportsInputEvents()) {
             throw new AssertionError("Could not inject input event if !supportsInputEvents()");
         }
 
-        if (displayId != 0 && !InputManager.setDisplayId(inputEvent, displayId)) {
-            return false;
-        }
-
-        return SERVICE_MANAGER.getInputManager().injectInputEvent(inputEvent, mode);
+        return injectEvent(inputEvent, mode, displayId);
     }
 
     public boolean injectEvent(InputEvent event) {

Then, you have to transmit both the displayId and the flag to NOT turn off if control is disabled.

What do you think?

npes87184 · 2021-02-17T05:53:28Z

As a quick-and-dirty solution, maybe you could expose a static injectEvent() taking a displayId:

diff --git a/server/src/main/java/com/genymobile/scrcpy/Device.java b/server/src/main/java/com/genymobile/scrcpy/Device.java
index a8fdf677..9fb365ff 100644
--- a/server/src/main/java/com/genymobile/scrcpy/Device.java
+++ b/server/src/main/java/com/genymobile/scrcpy/Device.java
@@ -157,16 +157,20 @@ public final class Device {
         return supportsInputEvents;
     }
 
+    public static boolean injectEvent(InputEvent inputEvent, int mode, int displayId) {
+        if (displayId != 0 && !InputManager.setDisplayId(inputEvent, displayId)) {
+            return false;
+        }
+
+        return SERVICE_MANAGER.getInputManager().injectInputEvent(inputEvent, mode);
+    }
+
     public boolean injectEvent(InputEvent inputEvent, int mode) {
         if (!supportsInputEvents()) {
             throw new AssertionError("Could not inject input event if !supportsInputEvents()");
         }
 
-        if (displayId != 0 && !InputManager.setDisplayId(inputEvent, displayId)) {
-            return false;
-        }
-
-        return SERVICE_MANAGER.getInputManager().injectInputEvent(inputEvent, mode);
+        return injectEvent(inputEvent, mode, displayId);
     }
 
     public boolean injectEvent(InputEvent event) {

Then, you have to transmit both the displayId and the flag to NOT turn off if control is disabled.

What do you think?

Oh, yes. This solution works. However, I am thinking about future. It will have more and more things that need to be done in CleanUp. If we need to pass variables likes this in each time, it is very annoy and dirty.

For infrastructure, in my opinion, in CleanUp process, I prefer to prepare a environment similar to the main process.
It just needs to handle the options correctly. Maybe it can add two methods, toFile and fromFile, to the class options.

What do you think?

After the struct screen is initialized, the window and the renderer are necessarily valid, so there is no need o check in screen_destroy().

This paves the way to handle EVENT_NEW_FRAME from screen.c, by allowing to call screen_update_frame() without an explicit video_buffer instance.

Now that all screen-related events are handled from screen.c, there is no need for a separate method for window events.

rom1v · 2021-02-18T00:20:58Z

It just needs to handle the options correctly.

It could be great to serialize some structure between the main process and the cleanup process indeed.

However, this structure should not be the existing Options, which is neither necessary (not all fields are meaningful for the cleanup process) nor sufficient (the cleanup actions may for example depend on the initial state of the device, which is not stored in the options).

(I don't know if in the end such a structure would be helpful compared to passing each option as argument.)

But yes, it would be great to share the common part of the Device initialization between the main and cleanup process. For now, Device is quite a mess, and it's not clear semantically what this class represents.

npes87184 · 2021-02-18T01:57:44Z

Ok, I will pass displayId in this time. The refactoring of Device and the environment preparation of CleanUp should be another problems (or features).

Thank you for your help.

Signed-off-by: Yu-Chen Lin <[email protected]>

npes87184 · 2021-02-21T00:46:06Z

Hi, @rom1v

I have added two commits about this features.

First, I export a series static methods for powering off. Second, make CleanUp process respects the power off setting.

Thanks.

rom1v · 2021-02-21T10:39:55Z

Thank you very much for your work 👍

I just tested, and I found a problem (on a Nexus 5): the "power off" cleanup seems to interact badly with the "restore normal power mode" cleanup, which is executed whenever the screen is on on exit. As a result, --power-off-on-close works randomly (on this device). (EDIT: this may not be the problem: #824 (comment))

I tried with these changes:

diff --git a/server/src/main/java/com/genymobile/scrcpy/CleanUp.java b/server/src/main/java/com/genymobile/scrcpy/CleanUp.java
index 7dda83fb..26012409 100644
--- a/server/src/main/java/com/genymobile/scrcpy/CleanUp.java
+++ b/server/src/main/java/com/genymobile/scrcpy/CleanUp.java
@@ -84,16 +84,14 @@ public final class CleanUp {
             }
         }
 
-        if (restoreNormalPowerMode) {
-            Ln.i("Restoring normal power mode");
-            if (Device.isScreenOn()) {
+        if (Device.isScreenOn()) {
+            if (powerOffScreen) {
+                Ln.i("Power off screen");
+                Device.powerOffScreen(displayId);
+            } else if (restoreNormalPowerMode) {
+                Ln.i("Restoring normal power mode");
                 Device.setScreenPowerMode(Device.POWER_MODE_NORMAL);
             }
         }
-
-        if (powerOffScreen) {
-            Ln.i("Power off screen");
-            Device.powerOffScreen(displayId);
-        }
     }
 }

It improves the behavior.

But I have an additional issue that can I cannot explain: if I keep touching the device (I keep one finger on the device screen) while I close the scrcpy window, it blocks and never power off the screen. It is unexpected, because if I keep one finger on the screen and press MOD+p to power off the screen, then it works. 😕

If I add some logs (and even if I always return true) in Device.isScreenOn():

diff --git a/server/src/main/java/com/genymobile/scrcpy/CleanUp.java b/server/src/main/java/com/genymobile/scrcpy/CleanUp.java
index 26012409..2c54558c 100644
--- a/server/src/main/java/com/genymobile/scrcpy/CleanUp.java
+++ b/server/src/main/java/com/genymobile/scrcpy/CleanUp.java
@@ -84,7 +84,9 @@ public final class CleanUp {
             }
         }
 
+        Ln.i("CLEANUP #1");
         if (Device.isScreenOn()) {
+            Ln.i("CLEANUP #2");
             if (powerOffScreen) {
                 Ln.i("Power off screen");
                 Device.powerOffScreen(displayId);
@@ -93,5 +95,6 @@ public final class CleanUp {
                 Device.setScreenPowerMode(Device.POWER_MODE_NORMAL);
             }
         }
+        Ln.i("CLEANUP #3");
     }
 }
diff --git a/server/src/main/java/com/genymobile/scrcpy/Device.java b/server/src/main/java/com/genymobile/scrcpy/Device.java
index 624c9fa0..3ee248d3 100644
--- a/server/src/main/java/com/genymobile/scrcpy/Device.java
+++ b/server/src/main/java/com/genymobile/scrcpy/Device.java
@@ -212,7 +212,7 @@ public final class Device {
     }
 
     public static boolean isScreenOn() {
-        return SERVICE_MANAGER.getPowerManager().isScreenOn();
+        return true; //SERVICE_MANAGER.getPowerManager().isScreenOn();
     }
 
     public synchronized void setRotationListener(RotationListener rotationListener) {

Then it blocks:

02-21 11:24:23.251 21036 21056 D scrcpy  : Device message sender stopped
02-21 11:24:23.485 21054 21054 I scrcpy  : Cleaning up
02-21 11:24:23.485 21054 21054 I scrcpy  : CLEANUP #1

If I replace Device.isScreenOn() by true, then it prints CLEANUP #2, but and Power off screen, but does not power off the screen. 🤔

rom1v · 2021-02-21T10:46:13Z

In fact, it seems the cleanup process is killed (the process does not appear in ps).

Btw, if I add a SystemClock.sleep(1000); (even if I don't touch the device), the process is closed before CLEANUP #1.
The fact that I touch the screen is irrelevant (it probably just change the timings and makes it more probable).

It's probably due to the OMXNodeInstance: !!! Observer died:

02-21 11:45:31.420 22574 22593 D scrcpy  : Controller stopped
02-21 11:45:32.469 22592 22592 I scrcpy  : Cleaning up
02-21 11:45:32.478 28310 29411 E OMXNodeInstance: !!! Observer died. Quickly, do something, ... anything...
02-21 11:45:32.590 28310 29411 W GraphicBufferSource: Dropped back down to Loaded without Executing
02-21 11:45:32.624 28310 29411 I OMX-VENC: Component Deinit

(I don't have this problem on my OnePlus 7 Pro)

npes87184 · 2021-02-21T13:36:31Z

In fact, it seems the cleanup process is killed (the process does not appear in ps).

Btw, if I add a SystemClock.sleep(1000); (even if I don't touch the device), the process is closed before CLEANUP #1.
The fact that I touch the screen is irrelevant (it probably just change the timings and makes it more probable).

It's probably due to the OMXNodeInstance: !!! Observer died:
02-21 11:45:31.420 22574 22593 D scrcpy  : Controller stopped
02-21 11:45:32.469 22592 22592 I scrcpy  : Cleaning up
02-21 11:45:32.478 28310 29411 E OMXNodeInstance: !!! Observer died. Quickly, do something, ... anything...
02-21 11:45:32.590 28310 29411 W GraphicBufferSource: Dropped back down to Loaded without Executing
02-21 11:45:32.624 28310 29411 I OMX-VENC: Component Deinit
(I don't have this problem on my OnePlus 7 Pro)

Hmm, I am not familiar with OMXNodeInstance: !!! Observer died.

BTW, I don't notice this problem in my note10+ with android 11.

diff --git a/server/src/main/java/com/genymobile/scrcpy/CleanUp.java b/server/src/main/java/com/genymobile/scrcpy/CleanUp.java
index 7dda83fb..26012409 100644
--- a/server/src/main/java/com/genymobile/scrcpy/CleanUp.java
+++ b/server/src/main/java/com/genymobile/scrcpy/CleanUp.java
@@ -84,16 +84,14 @@ public final class CleanUp {
             }
         }
 
-        if (restoreNormalPowerMode) {
-            Ln.i("Restoring normal power mode");
-            if (Device.isScreenOn()) {
+        if (Device.isScreenOn()) {
+            if (powerOffScreen) {
+                Ln.i("Power off screen");
+                Device.powerOffScreen(displayId);
+            } else if (restoreNormalPowerMode) {
+                Ln.i("Restoring normal power mode");
                 Device.setScreenPowerMode(Device.POWER_MODE_NORMAL);
             }
         }
-
-        if (powerOffScreen) {
-            Ln.i("Power off screen");
-            Device.powerOffScreen(displayId);
-        }
     }
 }

I will apply this change as it is more correct.

Thank you for your test and survey.

Signed-off-by: Yu-Chen Lin <[email protected]>

rom1v · 2021-03-16T20:16:06Z

Thank you 👍 Merged into dev

PR #824 <#824> Signed-off-by: Yu-Chen Lin <[email protected]> Signed-off-by: Romain Vimont <[email protected]>

This avoids to pass each option as individual parameter and parse them manually (it's still "manual" in the Parcelable implementation). Refs #824 <#824 (comment)>

This avoids to pass each option as individual parameter and parse them manually (it's still "manual" in the Parcelable implementation). Refs #824 <#824 (comment)> Reviewed-by: Yu-Chen Lin <[email protected]>

The option was not documented. Refs #824 <#824>

rom1v reviewed Sep 28, 2019

View reviewed changes

npes87184 force-pushed the master branch from 7ec0c29 to da7a2c2 Compare September 29, 2019 02:24

rom1v added 6 commits February 17, 2021 09:54

Improve error handling in screen initialization

44aa7f2

After the struct screen is initialized, the window and the renderer are necessarily valid, so there is no need o check in screen_destroy().

Reference video buffer from screen

b1734ab

This paves the way to handle EVENT_NEW_FRAME from screen.c, by allowing to call screen_update_frame() without an explicit video_buffer instance.

Handle screen-related events from screen.c

e1dbe4f

Inline window events handling

98628f2

Now that all screen-related events are handled from screen.c, there is no need for a separate method for window events.

Handle im-related events from input_manager.c

5c4c28c

Enable NDEBUG via Meson built-in option

dce0867

Export method to power off screen in Device

9cb8c92

Signed-off-by: Yu-Chen Lin <[email protected]>

npes87184 force-pushed the master branch from da7a2c2 to f399cb6 Compare February 21, 2021 00:43

npes87184 changed the base branch from master to dev February 21, 2021 00:44

Support power off on close

a9bd612

Signed-off-by: Yu-Chen Lin <[email protected]>

npes87184 force-pushed the master branch from f399cb6 to a9bd612 Compare February 21, 2021 14:00

rom1v force-pushed the dev branch from 2630512 to b16b65a Compare February 25, 2021 21:19

rom1v closed this Mar 16, 2021

rom1v pushed a commit that referenced this pull request Mar 16, 2021

Export static method to power off screen in Device

fb0bcae

PR #824 <#824> Signed-off-by: Yu-Chen Lin <[email protected]> Signed-off-by: Romain Vimont <[email protected]>

rom1v pushed a commit that referenced this pull request Mar 16, 2021

Support power off on close

1d615a0

PR #824 <#824> Signed-off-by: Yu-Chen Lin <[email protected]> Signed-off-by: Romain Vimont <[email protected]>

This was referenced Mar 25, 2021

Screen waking when app closes #2216

Closed

Terminating scrcpy when screen is off (normally via POWER) would turn it on with all-black display #1667

Open

rom1v added a commit that referenced this pull request Apr 30, 2021

Serialize clean-up configuration

e55c095

This avoids to pass each option as individual parameter and parse them manually (it's still "manual" in the Parcelable implementation). Refs #824 <#824 (comment)>

rom1v mentioned this pull request Apr 30, 2021

Serialize clean-up configuration #2286

Closed

rom1v mentioned this pull request Oct 27, 2021

Power off screen on close #714

Closed

rom1v added a commit that referenced this pull request Nov 7, 2021

Document --power-off-on-close

30d40f4

The option was not documented. Refs #824 <#824>

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Turn screen off when closing scrcpy #824

Turn screen off when closing scrcpy #824

npes87184 commented Sep 28, 2019

rom1v left a comment

rom1v Sep 28, 2019

npes87184 Sep 28, 2019 •

edited

Loading

rom1v commented Sep 28, 2019

npes87184 commented Sep 29, 2019

rom1v commented Oct 31, 2019

rom1v commented Aug 3, 2020

rom1v commented Feb 15, 2021

npes87184 commented Feb 16, 2021

npes87184 commented Feb 16, 2021 •

edited

Loading

rom1v commented Feb 16, 2021

npes87184 commented Feb 17, 2021 •

edited

Loading

rom1v commented Feb 18, 2021 •

edited

Loading

npes87184 commented Feb 18, 2021

npes87184 commented Feb 21, 2021

rom1v commented Feb 21, 2021 •

edited

Loading

rom1v commented Feb 21, 2021 •

edited

Loading

npes87184 commented Feb 21, 2021

rom1v commented Mar 16, 2021

Turn screen off when closing scrcpy #824

Turn screen off when closing scrcpy #824

Conversation

npes87184 commented Sep 28, 2019

rom1v left a comment

Choose a reason for hiding this comment

rom1v Sep 28, 2019

Choose a reason for hiding this comment

npes87184 Sep 28, 2019 • edited Loading

Choose a reason for hiding this comment

rom1v commented Sep 28, 2019

npes87184 commented Sep 29, 2019

rom1v commented Oct 31, 2019

rom1v commented Aug 3, 2020

rom1v commented Feb 15, 2021

npes87184 commented Feb 16, 2021

npes87184 commented Feb 16, 2021 • edited Loading

rom1v commented Feb 16, 2021

npes87184 commented Feb 17, 2021 • edited Loading

rom1v commented Feb 18, 2021 • edited Loading

npes87184 commented Feb 18, 2021

npes87184 commented Feb 21, 2021

rom1v commented Feb 21, 2021 • edited Loading

rom1v commented Feb 21, 2021 • edited Loading

npes87184 commented Feb 21, 2021

rom1v commented Mar 16, 2021

npes87184 Sep 28, 2019 •

edited

Loading

npes87184 commented Feb 16, 2021 •

edited

Loading

npes87184 commented Feb 17, 2021 •

edited

Loading

rom1v commented Feb 18, 2021 •

edited

Loading

rom1v commented Feb 21, 2021 •

edited

Loading

rom1v commented Feb 21, 2021 •

edited

Loading