libsdl-org
[SDL3] Adding input and FFB support for Logitech G29(PS3) on hidapi
https://github.com/libsdl-org/SDL/pull/11591 for SDL3
companion test program for SDL3 https://github.com/Kethen/sdl_lg4ff_util/tree/SDL3_lg4ff
WINE has not moved onto SDL3 so it cannot be tried on WINE at the moment
This is an impressive PR. How does this interact with other drivers, like new-lg4ff? Is there any reason we'd want this active when those are available?
This is an impressive PR.
Thanks but a large portion of it was ported from https://github.com/berarma/new-lg4ff
How does this interact with other drivers, like new-lg4ff?
On Linux, it depends on how it was opened:
- If it was opened on /dev/hidraw*, kernel drivers would stay active, and if the wheel is accessed through both drivers, it will be receiving confusing commands from two sources, tho it's generally fine if active effects are on only one side at a time
- If it was opened on /dev/bus/usb/* using libusb, exclusive access should be obtained and kernel drivers will unprobe
Which path would happen depends on which hidapi backend SDL was built to use and device node permissions
On MacOS, if I understand correctly some games also use IOHIDManager to provide wheel support on their own. SDL3 currently does not grab exclusive access https://github.com/libsdl-org/SDL/blob/47429227ab05311b7c5457dad2e748d68097da5f/src/hidapi/SDL_hidapi.c#L1246 No idea if existing games grab exclusive access
On Windows, I have no idea how it would react with the Logitech driver, no idea if any exclusive access would happen when hidapi opens the device from CM_Get_Device_Interface_List fetched interface names
Is there any reason we'd want this active when those are available?
On Linux, evdev does not provide an analog to GetEffectStatus, user is expected to track that as effects are uploaded, I think https://bugzilla.kernel.org/show_bug.cgi?id=219189; otherwise I'd expect the kernel drivers to have better latency profiles
As for availability, unless udev rules were supplied (usb) (hidraw), the evdev backend would be used on SDL
On Windows, once again I have no idea
Added build fixes for platforms with hidapi disabled, without M_PI, and Android
Still need help with MSVC project files
a large portion of it was ported from https://github.com/berarma/new-lg4ff
If code was taken from there, is there permission to relicense it from GPL to zlib?
Thanks @mungewell , @MadCatX and @berarma once again for graciously allowing ported code to be used under Zlib license in SDL
https://github.com/libsdl-org/SDL/pull/11591#issuecomment-2564725610 https://github.com/libsdl-org/SDL/pull/11591#issuecomment-2564761319
WINE has not moved onto SDL3 so it cannot be tried on WINE at the moment
What about sdl2-compat? I've been using it lately as arch forced this move and it works great. It should allow you to test this with Wine.
WINE has not moved onto SDL3 so it cannot be tried on WINE at the moment
What about sdl2-compat? I've been using it lately as arch forced this move and it works great. It should allow you to test this with Wine.
tested that a while back and it didn't quite work in WINE, it just didn't enumerate controllers, and I haven't gotten around to figure out why it didn't work
sdl2-compat does work with the companion test program however last time I tried
WINE has not moved onto SDL3 so it cannot be tried on WINE at the moment
What about sdl2-compat? I've been using it lately as arch forced this move and it works great. It should allow you to test this with Wine.
tested that a while back and it didn't quite work in WINE, it just didn't enumerate controllers, and I haven't gotten around to figure out why it didn't work
Can you investigate this? I'm ready to merge once it's confirmed working.
WINE has not moved onto SDL3 so it cannot be tried on WINE at the moment
What about sdl2-compat? I've been using it lately as arch forced this move and it works great. It should allow you to test this with Wine.
tested that a while back and it didn't quite work in WINE, it just didn't enumerate controllers, and I haven't gotten around to figure out why it didn't work
Can you investigate this? I'm ready to merge once it's confirmed working.
Just gave it a test, as of writing, on Linux, freshly compiled sdl2-compat(4453e2d05881c3616af8ca71f7c4267c60f993d8) can enumerate controllers in wine (10.0 staging and GE-Proton9-25), and can be used with this PR to provide G29 ffb for assetto corsa on dinput8, with the appropriate udev rules
$ lsusb
...
Bus 001 Device 016: ID 046d:c24f Logitech, Inc. G29 Driving Force Racing Wheel [PS3]
...
$ WINEDEBUG="-all,+hid" LD_LIBRARY_PATH="$(realpath ./sdl_test)" wine ... 2>&1 | tee log.txt &
$ cat log.txt
...
0098:trace:hid:process_device_event Received action 605
0098:trace:hid:sdl_add_device Making up serial number for Logitech G29: 03009a206d0400004fc2000000646800.0
0098:trace:hid:sdl_add_device joystick id 0, axis_offset 0, desc {vid 046d, pid c24f, version 6400, input -1, uid 00000000, is_gamepad 0}.
...
0084:trace:hid:sdl_device_physical_device_control iface 0x7fb14802fe40, control 0x9a.
0084:trace:hid:sdl_device_physical_device_set_gain iface 0x7fb14802fe40, percent 0x64.
0084:trace:hid:sdl_device_physical_effect_update iface 0x7fb14802fe40, index 1, params 0x7fb14802ffe0.
...
$ lsof /dev/bus/usb/001/016 sdl_test/*
lsof: WARNING: can't stat() tracefs file system /sys/kernel/debug/tracing
Output information may be incomplete.
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
winedevic 89380 katharine mem REG 0,52 3975143 sdl_test/libSDL3.so.0 (stat: Operation not permitted)
winedevic 89380 katharine mem REG 0,52 3975144 sdl_test/libSDL2-2.0.so.0 (stat: Operation not permitted)
winedevic 89380 katharine 43u CHR 189,15 0t0 1176 /dev/bus/usb/001/016
winedevic 89380 katharine 51u CHR 189,15 0t0 1176 /dev/bus/usb/001/016