it works perfectly (on uhid), thank you.
I changed the vid:pid and the controller name to match an xbox 360 and thats it, no modifications required on android itself:

public final class UhidManager {
         ByteBuffer buf = ByteBuffer.allocate(280 + reportDesc.length).order(ByteOrder.nativeOrder());
         buf.putInt(UHID_CREATE2);
 
-        String actualName = name.isEmpty() ? "scrcpy" : "scrcpy: " + name;
+        int vid = 0;
+        int pid = 0;
+        String actualName;
+        if ("Xbox One Controller".equals(name)) {
+            actualName = "Microsoft X-Box 360 Pad";
+            vid = 0x045e;
+            pid = 0x028e;
+        } else {
+            actualName = name.isEmpty() ? "scrcpy" : "scrcpy: " + name;
+        }
         byte[] utf8Name = actualName.getBytes(StandardCharsets.UTF_8);
         int len = StringUtils.getUtf8TruncationIndex(utf8Name, 127);
         assert len <= 127;
@@ -183,8 +192,8 @@ public final class UhidManager {
 
         buf.putShort((short) reportDesc.length);
         buf.putShort(BUS_VIRTUAL);
-        buf.putInt(0); // vendor id
-        buf.putInt(0); // product id
+        buf.putInt(vid); // vendor id
+        buf.putInt(pid); // product id
         buf.putInt(0); // version
         buf.putInt(0); // country;
         buf.put(reportDesc);

aoa is gonna be broken as long as we cant change the vid:pid on it (or you could revert back to the old descriptor for aoa).

im doing the "if ("Xbox One Controller".equals(name)){" check so i dont affect other devices like a mouse or a keyboard, i dont know if it would work for people with other controllers (or other os).

but here's the modified server file with that check removed if anyone wants to try it (probably breaks the uhid keyboard):
scrcpy-server.zip

I changed the vid:pid and the controller name to match an xbox 360 and thats it

What if you keep 0:0, does it also work?

I changed the vid:pid and the controller name to match an xbox 360 and thats it

What if you keep 0:0, does it also work?

well, yes, but the rstick doesnt work and the triggers act as the rstick instead.

well, yes, but the rstick doesnt work and the triggers act as the rstick instead.

Is that also true with "Game Controller Tester" (uk.co.powgames.gamecondiag)? On my devices, it works correctly in Gamepad testers with 0:0.

Can it be related to your changes in .kl files? (#5362 (comment))

well, yes, but the rstick doesnt work and the triggers act as the rstick instead.

Is that also true with "Game Controller Tester" (uk.co.powgames.gamecondiag)? On my devices, it works correctly in Gamepad testers with 0:0.

yup, its the same, which gamepad(s) did you try?

Can it be related to your changes in .kl files? (#5362 (comment))

nope, i reverted all of those changes, you dont need keylayout modifications to use xbox360 vid:pids with this update.

yup, its the same, which gamepad(s) did you try?

With a PS4 controller, but that should not make any difference, since scrcpy uses the events received from SDL to generate UHID events.

The difference is probably the drivers on the device. On my devices, it works correctly with both 0:0 or 045e:028e (but not with the ids of the PS4 controller, because it is expected to behave differently I guess).

The difference is probably the drivers on the device. On my devices, it works correctly with both 0:0 or 045e:028e (but not with the ids of the PS4 controller, because it is expected to behave differently I guess).

how do games react to input? do they see your gamepad as an xbox or a ps one?
some ps4 keylayouts use generic identifiers so thats probably the issue you're having i reckon.
you can try the dualsense one: Vendor=0x054c, Product=0x0ce6
its axis identifiers are identical to xbox's, buttons might not work though.

With 054c:0ce6, in Gamepad Tester, it is detected as a "PS5 Controller".
The right stick and L2/R2 axis are inverted (right stick changes L2/R2, and vice-versa).

With 0:0 or 045e:028e, it works correctly.

I will add an option --gamepad-uhid-ids=xxxx:xxxx, and use 045e:028e by default.

Or just force 045e:028e without option (setting another value would not change the HID reports, so it makes no sense to expose it).

I updated the PR.

Please test the latest version.

Here are binaries for Linux and Windows:

• scrcpy-linux-x86_64-gamepad_fixes.tar.gz SHA-256: f81fe2e3428018b403e401fd87483718ec2452b35f4586a7d8c612c4ccaac4d
• scrcpy-win64-gamepad_fixes.zip SHA-256: 9e2f0a521ba191f8b098ccfc0c3f8ff3d406337c3849a076f754ecd0dfdaa53

@LHLaurini @yume-chan I'm also interested in your feedback if you have some time, since you worked on gamepads in the past 😉 Thank you.

I updated the PR.

Please test the latest version.

Here are binaries for Linux and Windows:

• scrcpy-linux-x86_64-gamepad_fixes.tar.gz SHA-256: f81fe2e3428018b403e401fd87483718ec2452b35f4586a7d8c612c4ccaac4d
• scrcpy-win64-gamepad_fixes.zip SHA-256: 9e2f0a521ba191f8b098ccfc0c3f8ff3d406337c3849a076f754ecd0dfdaa53

it works perfectly, but i think we should change the controller name as well, since some games seem to pay attention to that:

diff --git a/app/src/uhid/gamepad_uhid.c b/app/src/uhid/gamepad_uhid.c
index 2a063af5..b9016b39 100644
--- a/app/src/uhid/gamepad_uhid.c
+++ b/app/src/uhid/gamepad_uhid.c
@@ -36,7 +36,7 @@ sc_gamepad_uhid_send_open(struct sc_gamepad_uhid *gamepad,
     msg.uhid_create.id = hid_open->hid_id;
     msg.uhid_create.vendor_id = SC_GAMEPAD_UHID_VENDOR_ID;
     msg.uhid_create.product_id = SC_GAMEPAD_UHID_PRODUCT_ID;
-    msg.uhid_create.name = hid_open->name;
+    msg.uhid_create.name = "Microsoft X-Box 360 Pad";
     msg.uhid_create.report_desc = hid_open->report_desc;
     msg.uhid_create.report_desc_size = hid_open->report_desc_size;
 
diff --git a/server/src/main/java/com/genymobile/scrcpy/control/UhidManager.java b/server/src/main/java/com/genymobile/scrcpy/control/UhidManager.java
index 1d7678ec..c4867a3f 100644
--- a/server/src/main/java/com/genymobile/scrcpy/control/UhidManager.java
+++ b/server/src/main/java/com/genymobile/scrcpy/control/UhidManager.java
@@ -174,7 +174,7 @@ public final class UhidManager {
         ByteBuffer buf = ByteBuffer.allocate(280 + reportDesc.length).order(ByteOrder.nativeOrder());
         buf.putInt(UHID_CREATE2);
 
-        String actualName = name.isEmpty() ? "scrcpy" : "scrcpy: " + name;
+        String actualName = name.isEmpty() ? "scrcpy" : name;
         byte[] utf8Name = actualName.getBytes(StandardCharsets.UTF_8);
         int len = StringUtils.getUtf8TruncationIndex(utf8Name, 127);
         assert len <= 127;

i think we should change the controller name as well, since some games seem to pay attention to that

Did you observe a difference in behavior with just a different name?

If possible, I'd like to keep exposing the real controller name, but if it causes issues, ok.

Or just force 045e:028e without option (setting another value would not change the HID reports, so it makes no sense to expose it).

I believe having that option would only be useful for games and controllers that support stuff like gyro, adaptive triggers and haptic feedback. While I did find some references online saying some games support these features, I didn't find which games support them. In any case, I don't think scrcpy currently supports these features anyway, so it doesn't matter for now.

i think we should change the controller name as well, since some games seem to pay attention to that

Did you observe a difference in behavior with just a different name?

yes, in human fall flat (#5362), the game didnt recognize the controller cause of its name, changing the name fixed the issue.

@Withoutruless Thank you 👍 I added a commit under your authorship: 7418fd0.

(and I added a log when a gamepad is added or removed: 548d2c7)

Please confirm that everything works.

(scrcpy 3.1 including this PR is ready. I will publish it soon 🚀)

@rom1v Hello! It seems there's still a problem with some gamepads. The triggers on Logitech F310 don't work in either mode, and the Y and X buttons are for some reason swapped. Tested on 3.1 release version and gamepad_fixes version from this conversation.

On hardwaretester.com website opened on my phone I can see that pulling LT does nothing, pulling RT changes Axis 3 (RS Y-axis) value from -1 to 1. B6 and B7 (correct inputs for LT/RT) don't activate at all. Other inputs work fine. On PC (win-64) this gamepad is identified as Xbox 360 Controller and on my phone as ffff, if it helps.

Could you please look into it?