Open yoyossef opened 3 years ago
kozec
commented
3 years ago I would be glad to help (since I have both a DualSense and a DualShock 4 controller) with a little bit of guidance.
First of all, try skipping DS4Controller class which uses HID directly and using DS4EvdevController which relies on evdev. Evdev equivalent of hiddrv_test can be found here.
Then it's hopefully just about changing BUTTON_MAP, AXIS_MAP and GYRO_MAP so they match what DualSense outputs.
psstoyanov
commented
3 years ago @kozec , how to use evdevdrv_test ?
I don't see the BUTTON_MAP being different for DualSense when looking at ds4drv.py. The same key codes are being returned from evtest (or at least the main ones - no idea about the Dpad).
Ot does go wonky when you start it - pressing the right trigger to the max looks as if pressing multiple buttons on the UI:
Update 1: This could be a better representation of the issue.
Without SC-Controller running things are looking as expected in evtest:
With SC-Controller running using Xbox Controller setup with the DualSense controller hooked up, evtest gets the following output from the left trigger:
Update 2: Looking further into the evtest capabilities seems to match the axis bindings so far:
Update 3: Forgot to post the system info:
kozec
commented
3 years ago Event: time 1628803562.738729, type 1 (EV_KEY), code 312 (BTN_TL2), value 1 .... Event: time 1628803563.230553, type 1 (EV_KEY), code 312 (BTN_TL2), value 0
Sounds like it actually sends button event when trigger is pulled. 312 is "share" button on DS4. Can't that be reason for virtual buttons getting pressed?
psstoyanov
commented
3 years ago Thank you for the reply, @kozec !
It does spawn a button event when the trigger is pulled. Unlike the Steam Controller with its two-stage triggers (defined click when fully pressed), on DualSense these events are spawned when the trigger is pulled and released regardless of the amount. It could be part of the adaptive trigger system? :thinking:
The naming confuses me a bit :smile: Which is the "share" button on DS4? Both DS4 and DualSense use code 314 (BTN_SELECT) for the button next to the touchpad, from what I can find.
Here is a video of SCC misbehaving with DualSense when only the product id change is made: https://user-images.githubusercontent.com/3979134/129480303-bfa96121-307a-41fa-9afe-1aa01132e18f.mp4
When I mess with the hardcoded HID values, the behaviour changes in terms of what buttons are being triggered through scc :thinking: How were the values for the DS4 here derived? This looks to be the most promising place where adjustments for DualSense would have to be made.
Note:
When I run SC Controller, I have HID disabled in the settings.
While it could be incredibly naive from my part, actually removing the DS4Controller(HIDController) class leaves the SCC unable to detect the DualSense controller (doesn't show up in Controller registration either which is a tad weird). So it doesn't look like avoiding HID entirely is an option.
psstoyanov
commented
3 years ago Small update - I've made a bit of progress with hardcoding new values for the byte_offset. So far, it looks promising.
Currently being stuck on the touchpad. I take it that I should figure out how this commit got to be made: https://github.com/kozec/sc-controller/commit/9a05d1d92610db0888a0dc50353db3732c846f19 If I can understand how to read these two HID descriptors for DualSense and DS4 things will be easier :smiley: I'm not looking forward to the gyro afterwards.
Update: All buttons (touchpad excluded) now work! :tada: Will prepare an initial PR later tonight.
psstoyanov
commented
3 years ago This is very promising information for the touchpad but I can't really process it at the moment: https://github.com/nondebug/dualsense/blob/ac25625c5d3b9f067ec92c96e1e0738e5c5f39cc/dualsense-explorer.html#L208
psstoyanov
commented
3 years ago Ehh.... I'm still fighting with the Touchpad. On the bright side, the Gyro is working.
Interestingly, it can detect touchpad button press but dealing with the X/Y touch proves to be a bit difficult for me so far.
psstoyanov
commented
3 years ago @kozec how did you reach to this commit? https://github.com/kozec/sc-controller/commit/9a05d1d92610db0888a0dc50353db3732c846f19 I'm curious as to what is the proper method of exploring the HID return values and their manipulation.
I'm not sure if DualSense will require changes to DS4TOUCHPAD but I don't understand how the values were reached either.
If I can figure out how to see any printf statements from hiddrv.c, I might be able to see what I'm doing wrong without that much guidance.
psstoyanov
commented
3 years ago Found the touchpad byte_offset values for X&&Y on DualSense.
Printing the results from hiddrv.c in grab_value posts the correct results.
However trying to use the touchpad through sc-controller is very janky - quite a few jumps left and right. I don't have DS4 to know if that is the expected behaviour there or if I just need to tinker with the config values to make it smooth.
Will push another commit later.
kozec
commented
3 years ago I'm curious as to what is the proper method of exploring the HID return values and their manipulation.
I can't say whether it is proper method, but I've just ran loop printing out packets converted to hex and guessed which one is which from how they were changing.
However trying to use the touchpad through sc-controller is very janky - quite a few jumps left and right.
That may be expected. Smoothing is applied much later in mapper, using smooth modifier.
psstoyanov
commented
3 years ago Thanks, @kozec !
I wouldn't use the touchpad as a touchpad without further changes for now. It looks as if the jump occurs when lifting and then pressing down (could be re-using the old value as a comparison causing the spike). That would be a task for the future, I guess.
Is there rumble/ haptics support for DS4? Not sure if this will require entirely new pieces of code for DualSense or if I'm missing something while looking at the DS4 config.
LED - possibly new struct inside hiddrv.c?
Note:
This looks like an interesting source when it comes to the value that are spat from HID: https://controllers.fandom.com/wiki/Sony_DualSense/Data_Structures
And then there is the GUI - I should check #595
Something simpler for tonight - checking if it works through Bluetooth.
PS: Doesn't look optimistic:
~ sdl2-jstest -r 0
Unable to open haptic on joystick 0
SDL_Error: Haptic: Joystick isn't a haptic device.Well, the Bluetooth evdev config is all over the place. But it looks like as if only the Dpad and right stick will need some helping hand there. Or is a separate HID driver for Bluetooth preferable?
psstoyanov
commented
3 years ago Curious - fftest does work on the same machine:
So at least partial rumble support should be possible :thinking:
kozec
commented
3 years ago Is there rumble/ haptics support for DS4? Not sure if this will require entirely new pieces of code for DualSense or if I'm missing something while looking at the DS4 config.
No, but it should be possible for both DS4 and DSS.
It would require only handler like this, but with values DS4/DSS expects.
Or is a separate HID driver for Bluetooth preferable?
I'd say only if it allows for cleaner code.
psstoyanov
commented
3 years ago Thank you for the explanation, @kozec ! Not sure if rumble is something I can tackle yet.
Now this is both great and a bit weird at the same time - touchpad in Bluetooth mode.
My previous check was on one of the SBC devices I have and Bluetooth worked well enough. What I didn't notice is the older kernel the device was running explaining the jankier setup.
Meanwhile on the tower, I finally figured out that I had to enable the bluetooth service (cough), DualSense is detected immediately just like the wired connection. The strange part is the touchpad - it works great in Bluetooth mode! No jumping around or strange behaviour like in USB connection. Maybe the byte_offset is different in wired connection?
Tomorrow, I will play a bit with the GUI and maybe touchpad in wired mode.
psstoyanov
commented
3 years ago I think so far things are looking good.
I've figured out which piece was missing for the erratic touchpad behaviour in USB mode (missed the override earlier).Stil far slower cursor movement while in USB mode compared to Bluetooth but it's smooth - no random spikes. Added separate UI items for Glade.
USB HID mode does highjack the audio portion from the controller (headphones/mic don't work while emulation is on). Not sure if that is something I can resolve.
@kozec , is there something that you would like to see changed in the merge request? I don't know if it's good but at least some basic support is present.
@yoyossef , it would be great if you can try the changes I've made so far.
lemonxah
commented
3 years ago @psstoyanov do i just build and install your dualsense branch? i want to start testing this as i dont have any other controller i only have a dualsense
psstoyanov
commented
3 years ago @lemonxah , yup, that should be enough :+1:
psstoyanov
commented
3 years ago @lemonxah , as a note - you can build and run directly instead of installing. I found this to be a better way and would recommend it as it won't affect your system in case you don't like it.
As a side note - the daemon can continue to run if you kill the GUI app. It's good to double-check through htop or similar if scc is still running in the background.
To run just use the run.sh
lemonxah
commented
3 years ago @psstoyanov is your controller working via bluetooth or only usb?
psstoyanov
commented
3 years ago @lemonxah - both. However I am running on kernel 5.13 - if you are on kernel older than 5.12, then I don't know how it will react. The DualSense kernel driver was merged in 5.12 if I recall correctly.
lemonxah
commented
3 years ago @psstoyanov i am on 5.13 manjaro with the sony hid driver installed, but mine isn't workoing with bluetooth, is there any extra steps that you have taken to make it work over blue tooth? the controller connects to my pc fine it works on steam with bluetooth, but i shut down steam before i was running this shoudl steam be running?
psstoyanov
commented
3 years ago For Bluetooth, I just pair the controller from the settings and then it gets picked up sc-controller automatically from what I can see. No idea if hooking it up through USB first makes a difference.
lemonxah
commented
3 years ago weird same setup, but did't it up and then i plugged it in and it works right away .. ok cool let me debug some more thanks, at least i can play lutris games now thanks
psstoyanov
commented
3 years ago @lemonxah , maybe I have system side sdl2 config for DualSense present as well that is affecting the Bluetooth connection (I haven't updated the sdl2 config in sc-controller to include it). I will try to see if that is the case over the weekend.
psstoyanov
commented
3 years ago @lemonxah , ahh, I've just rebuilt the branch cleanly. I didn't commit my sdl2 configs and might've remained there when I was running it previously. Will check if Bluetooth will work once I add them back.
Update:
Huh, doesn't matter if I plug then in gamecontrollerdb.txt(though I will make another commit containing those just in case).
Starting with Bluetooth mode not being recognised -> disable DualSense support and restart emulation -> re-enable and restart emulation again got it recognised. Not sure why that is :thinking:
lemonxah
commented
3 years ago i have followed those steps mulitple times and its still not recognised in bluetooth,
psstoyanov
commented
3 years ago @lemonxah , I can replicate it with my Pinebook Pro running Manjaro ARM. I will try to see if I can figure out why this happens.
~btw, when you use Bluetooth connection, does it list it as 054c 0xce6 ?~ Doesn't seem related
psstoyanov
commented
3 years ago Weird, on the PBP the touchpad isn't detected in BT mode as well. Not sure if it's related or if you observe the same.
I did miss a line here specifically for Bluetooth (it doesn't solve the issue for me): https://github.com/kozec/sc-controller/blob/ce92c773b8b26f6404882e9209aff212c4053170/scripts/69-sc-controller.rules#L12 Will add it in a bit.
psstoyanov
commented
3 years ago @lemonxah , made a couple of changes. Give them a try and lets see if it will resolve it at least for you.
psstoyanov
commented
3 years ago Update - on the PBP, DualSense is displayed just as generic controller through evdev whereas on the PC tower it's listed as 3 devices in BT mode as well as USB mode :confused:
psstoyanov
commented
3 years ago 054C:09CC is reported by DualShock 4 and 054C:0CE6 is DualSense - that should be normal.
Thanks for the log! Will try to see if something stands out
tonibgd
commented
2 years ago Great work on the PR @psstoyanov , I pulled your branch and tried in on my machine. Wired works without a problem, Bluetooth doesn't work at all for me. Seems like the controller is not being detected when paired over Bluetooth.
Bluetooth dmesg:
[ 86.905921] hid-generic 0005:054C:0CE6.0008: unknown main item tag 0x0
[ 86.906128] input: Wireless Controller as /devices/pci0000:00/0000:00:14.0/usb1/1-10/1-10:1.0/bluetooth/hci0/hci0:12/0005:054C:0CE6.0008/input/input28
[ 86.907443] hid-generic 0005:054C:0CE6.0008: input,hidraw7: BLUETOOTH HID v1.00 Gamepad [Wireless Controller] on 5c:f3:70:a4:1b:f8
[ 86.967529] playstation 0005:054C:0CE6.0008: unknown main item tag 0x0
[ 86.967892] playstation 0005:054C:0CE6.0008: hidraw7: BLUETOOTH HID v1.00 Gamepad [Wireless Controller] on 5c:f3:70:a4:1b:f8
[ 86.983015] input: Wireless Controller as /devices/pci0000:00/0000:00:14.0/usb1/1-10/1-10:1.0/bluetooth/hci0/hci0:12/0005:054C:0CE6.0008/input/input29
[ 86.984039] input: Wireless Controller Motion Sensors as /devices/pci0000:00/0000:00:14.0/usb1/1-10/1-10:1.0/bluetooth/hci0/hci0:12/0005:054C:0CE6.0008/input/input30
[ 86.985109] input: Wireless Controller Touchpad as /devices/pci0000:00/0000:00:14.0/usb1/1-10/1-10:1.0/bluetooth/hci0/hci0:12/0005:054C:0CE6.0008/input/input31
[ 86.987040] playstation 0005:054C:0CE6.0008: Registered DualSense controller hw_version=0x00000313 fw_version=0x01000037USB dmesg
[ 3646.760289] input: Sony Interactive Entertainment Wireless Controller as /devices/pci0000:00/0000:00:14.0/usb1/1-10/1-10:1.3/0003:054C:0CE6.0008/input/input28
[ 3646.760646] hid-generic 0003:054C:0CE6.0008: input,hidraw7: USB HID v1.11 Gamepad [Sony Interactive Entertainment Wireless Controller] on usb-0000:00:14.0-10/input3
[ 3647.479093] playstation 0003:054C:0CE6.0008: hidraw7: USB HID v1.11 Gamepad [Sony Interactive Entertainment Wireless Controller] on usb-0000:00:14.0-10/input3
[ 3647.538926] input: Sony Interactive Entertainment Wireless Controller as /devices/pci0000:00/0000:00:14.0/usb1/1-10/1-10:1.3/0003:054C:0CE6.0008/input/input29
[ 3647.539209] input: Sony Interactive Entertainment Wireless Controller Motion Sensors as /devices/pci0000:00/0000:00:14.0/usb1/1-10/1-10:1.3/0003:054C:0CE6.0008/input/input30
[ 3647.539783] input: Sony Interactive Entertainment Wireless Controller Touchpad as /devices/pci0000:00/0000:00:14.0/usb1/1-10/1-10:1.3/0003:054C:0CE6.0008/input/input31
[ 3647.541650] playstation 0003:054C:0CE6.0008: Registered DualSense controller hw_version=0x00000313 fw_version=0x01000037I'm using jstest to see if button presses are being registered, here are a few screenshots:
Wired over USB + Emulation enabled (Works):
Bluetooth + Emulation disabled:
Bluetooth + Emulation enabled:
psstoyanov
commented
2 years ago Thanks, @tonibgd , I know BT is... at best dodgy for now. I need to spend some time figuring it out. Not sure if that will happen this month.
tonibgd
commented
2 years ago @psstoyanov I managed to make it work over Bluetooth.
Going through the code i saw that HAVE_EVDEV flag is required for Bluetooth callback registering, made me think this check wasn't passing for some reason, so i started looking into evdev dependencies.
Remembered that this project uses python2, installed python2-evdev from AUR and now it works.
Hope this information is useful.
if HAVE_EVDEV and config["drivers"].get("evdevdrv"):
daemon.get_device_monitor().add_callback("bluetooth",
VENDOR_ID, PRODUCT_ID, make_evdev_device, None)So... the goal would be to figure out a method to notify the user about the presence (or lack of) python2-evdev I guess.
@kozec , is there already per-existing function that is similar or that can be expanded for this purpose?
DominikNovosel
commented
1 year ago Is this still something that's being worked on? Having Dualsense support would be great.
psstoyanov
commented
1 year ago @DominikNovosel , take a look at https://github.com/kozec/sc-controller/pull/648
For me, the base is working okey-ish as long as you have that python2-evdev lib on your system (which could be problematic now that Arch has also dropped python2 support).
Please, check the comments from other people as well and the forks of my brach - iirc there were a couple that were more feature complete than mine but I haven't even got the time to properly take a look at them. In particular take a look at:
DominikNovosel
commented
1 year ago Thanks. Clearly, there's been progress, but it's still not super user-friendly and I'm nowhere near being a full-blown developer that could try and figure this out.
Ryochan7
commented
1 year ago Ended up combining the efforts put out by theY4Kman and psstoyanov and made a branch for DualSense support in the Python 3 fork. Works with USB and Bluetooth. Going to experiment with it some more before I think about putting the code in the python3 branch.
https://github.com/Ryochan7/sc-controller/tree/theY4Kman-dualsense
Ryochan7
commented
1 year ago Going to make an attempt to handle Bluetooth via hidraw rather than evdev like it is now. Finally got the DualSense to accept an output packet with the slightly different calls required in Python to generate valid crc32. Got some of the input report parsed and most standard controls are mappable; it is still lacking dpad, touchpad, and gyro parsing.
Ryochan7
commented
1 year ago Current shenanigans. It is in a mostly working state. DPad and Touchpad support now work correctly. Feedback support now seems to work; I carried over the method from DS5Controller.feedback and made some small modifications to the code (crc32 computation and changed min duration). The main portion that does not work is Gyro Tilt; Gyro Mouse should work fine but I have not tested it. Confused about how to compute a quaternion based on the available IMU data provided by the controller. The Steam Controller provides that data in its input data packets.
https://github.com/Ryochan7/sc-controller/commit/e5508284005cf7042bda010e8a1620d51dde58e6
Ryochan7
commented
1 year ago Was wrong before about gyro working. I didn't have the proper method established so the mapper would use the gyro data. Also, the gyro rotation data was not interpreted as int16 values like they should have. Now Gyro Mouse works and has been tested. Still working on the quaternion problem for some gyro actions in the mapper. Got a somewhat functional routine but it seems unstable so it is commented out.
Hi,
I was wondering if the new DualSense controller could be supported? Official drivers are already on the way to be merged for Linux 5.12 ( https://www.phoronix.com/scan.php?page=news_item&px=Sony-PS5-DualSense-Linux-5.12 ) and it is already available on the AUR to install ( https://aur.archlinux.org/packages/hid-playstation-dkms/ ).
SDL and Steam Input already supports it but it would be nice to at least be able to emulate it as a Xbox 360 controller on non-steam games (for example I can't get it to work on some lutris games as is, or even with Steam Input as an overlay).
The drivers are really similar to the DS4 one, the controller is recognized as multiples devices, a controller, a gyroscope, a touchpad etc.
It would be great to have support for the basic DualSense features for now, to be on par with the DualShock 4 support at least because they are so similar.
I tried to get it working through
scc/drivers/ds4drv.pyby substituting the product id to0ce6instead of09ccbut I couldn't get it to work properly when executingscc/drivers/ds4drv.pyas a script, the controller gets recognized but the inputs are wrongly read or something.I would be glad to help (since I have both a DualSense and a DualShock 4 controller) with a little bit of guidance.