Ryujinx-SDL/src/joystick/windows/SDL_rawinputjoystick.c

1824 lines
66 KiB
C
Raw Normal View History

Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
/*
Simple DirectMedia Layer
Copyright (C) 2019 Sam Lantinga <slouken@libsdl.org>
This software is provided 'as-is', without any express or implied
warranty. In no event will the authors be held liable for any damages
arising from the use of this software.
Permission is granted to anyone to use this software for any purpose,
including commercial applications, and to alter it and redistribute it
freely, subject to the following restrictions:
1. The origin of this software must not be misrepresented; you must not
claim that you wrote the original software. If you use this software
in a product, an acknowledgment in the product documentation would be
appreciated but is not required.
2. Altered source versions must be plainly marked as such, and must not be
misrepresented as being the original software.
3. This notice may not be removed or altered from any source distribution.
*/
/*
RAWINPUT Joystick API for better handling XInput-capable devices on Windows.
XInput is limited to 4 devices.
Windows.Gaming.Input does not get inputs from XBox One controllers when not in the foreground.
DirectInput does not get inputs from XBox One controllers when not in the foreground, nor rumble or accurate triggers.
RawInput does not get rumble or accurate triggers.
So, combine them as best we can!
*/
#include "../../SDL_internal.h"
#if SDL_JOYSTICK_RAWINPUT
#include "SDL_assert.h"
#include "SDL_endian.h"
#include "SDL_events.h"
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
#include "SDL_hints.h"
#include "SDL_timer.h"
#include "../usb_ids.h"
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
#include "../SDL_sysjoystick.h"
#include "../../core/windows/SDL_windows.h"
#include "../../core/windows/SDL_hid.h"
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
#include "../hidapi/SDL_hidapijoystick_c.h"
#define SDL_JOYSTICK_RAWINPUT_XINPUT
#ifdef SDL_WINDOWS10_SDK
#define SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
#endif
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
#include "../../core/windows/SDL_xinput.h"
#endif
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
#include "../../core/windows/SDL_windows.h"
typedef struct WindowsGamingInputGamepadState WindowsGamingInputGamepadState;
#define GamepadButtons_GUIDE 0x40000000
#define COBJMACROS
#include "windows.gaming.input.h"
#endif
#if defined(SDL_JOYSTICK_RAWINPUT_XINPUT) || defined(SDL_JOYSTICK_RAWINPUT_GAMING_INPUT)
#define SDL_JOYSTICK_RAWINPUT_MATCHING
#endif
/* #define DEBUG_RAWINPUT */
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
#ifndef RIDEV_EXINPUTSINK
#define RIDEV_EXINPUTSINK 0x00001000
#define RIDEV_DEVNOTIFY 0x00002000
#endif
#ifndef WM_INPUT_DEVICE_CHANGE
#define WM_INPUT_DEVICE_CHANGE 0x00FE
#endif
#ifndef WM_INPUT
#define WM_INPUT 0x00FF
#endif
#ifndef GIDC_ARRIVAL
#define GIDC_ARRIVAL 1
#define GIDC_REMOVAL 2
#endif
/* external variables referenced. */
extern HWND SDL_HelperWindow;
static SDL_bool SDL_RAWINPUT_inited = SDL_FALSE;
static int SDL_RAWINPUT_numjoysticks = 0;
static SDL_bool SDL_RAWINPUT_need_pump = SDL_TRUE;
static void RAWINPUT_JoystickDetect(void);
static void RAWINPUT_PumpMessages(void);
static void RAWINPUT_JoystickClose(SDL_Joystick *joystick);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
typedef struct _SDL_RAWINPUT_Device
{
2020-04-13 22:24:19 +02:00
SDL_atomic_t refcount;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
char *name;
Uint16 vendor_id;
Uint16 product_id;
Uint16 version;
SDL_JoystickGUID guid;
SDL_bool is_xinput;
PHIDP_PREPARSED_DATA preparsed_data;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
HANDLE hDevice;
SDL_Joystick *joystick;
SDL_JoystickID joystick_id;
struct _SDL_RAWINPUT_Device *next;
} SDL_RAWINPUT_Device;
struct joystick_hwdata
{
SDL_bool is_xinput;
PHIDP_PREPARSED_DATA preparsed_data;
ULONG max_data_length;
HIDP_DATA *data;
USHORT *button_indices;
USHORT *axis_indices;
USHORT *hat_indices;
SDL_bool guide_hack;
SDL_bool trigger_hack;
USHORT trigger_hack_index;
#ifdef SDL_JOYSTICK_RAWINPUT_MATCHING
Uint32 match_state; /* Low 16 bits for button states, high 16 for 4 4bit axes */
Uint32 last_state_packet;
#endif
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
SDL_bool xinput_enabled;
SDL_bool xinput_correlated;
Uint8 xinput_correlation_id;
Uint8 xinput_correlation_count;
Uint8 xinput_uncorrelate_count;
Uint8 xinput_slot;
#endif
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
SDL_bool wgi_correlated;
Uint8 wgi_correlation_id;
Uint8 wgi_correlation_count;
Uint8 wgi_uncorrelate_count;
WindowsGamingInputGamepadState *wgi_slot;
#endif
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
SDL_RAWINPUT_Device *device;
};
typedef struct joystick_hwdata RAWINPUT_DeviceContext;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
SDL_RAWINPUT_Device *SDL_RAWINPUT_devices;
static const Uint16 subscribed_devices[] = {
USB_USAGE_GENERIC_GAMEPAD,
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
/* Don't need Joystick for any devices we're handling here (XInput-capable)
USB_USAGE_GENERIC_JOYSTICK,
USB_USAGE_GENERIC_MULTIAXISCONTROLLER,
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
*/
};
#ifdef SDL_JOYSTICK_RAWINPUT_MATCHING
static struct {
Uint32 last_state_packet;
SDL_Joystick *joystick;
SDL_Joystick *last_joystick;
} guide_button_candidate;
typedef struct WindowsMatchState {
SHORT match_axes[4];
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
WORD xinput_buttons;
#endif
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
Uint32 wgi_buttons;
#endif
SDL_bool any_data;
} WindowsMatchState;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
static void RAWINPUT_FillMatchState(WindowsMatchState *state, Uint32 match_state)
{
int ii;
state->any_data = SDL_FALSE;
/* SHORT state->match_axes[4] = {
(match_state & 0x000F0000) >> 4,
(match_state & 0x00F00000) >> 8,
(match_state & 0x0F000000) >> 12,
(match_state & 0xF0000000) >> 16,
}; */
for (ii = 0; ii < 4; ii++) {
state->match_axes[ii] = (match_state & (0x000F0000 << (ii * 4))) >> (4 + ii * 4);
if ((Uint32)(state->match_axes[ii] + 0x1000) > 0x2000) { /* match_state bit is not 0xF, 0x1, or 0x2 */
state->any_data = SDL_TRUE;
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
/* Match axes by checking if the distance between the high 4 bits of axis and the 4 bits from match_state is 1 or less */
#define XInputAxesMatch(gamepad) (\
(Uint32)(gamepad.sThumbLX - state->match_axes[0] + 0x1000) <= 0x2fff && \
(Uint32)(~gamepad.sThumbLY - state->match_axes[1] + 0x1000) <= 0x2fff && \
(Uint32)(gamepad.sThumbRX - state->match_axes[2] + 0x1000) <= 0x2fff && \
(Uint32)(~gamepad.sThumbRY - state->match_axes[3] + 0x1000) <= 0x2fff)
/* Explicit
#define XInputAxesMatch(gamepad) (\
SDL_abs((Sint8)((gamepad.sThumbLX & 0xF000) >> 8) - ((match_state & 0x000F0000) >> 12)) <= 0x10 && \
SDL_abs((Sint8)((~gamepad.sThumbLY & 0xF000) >> 8) - ((match_state & 0x00F00000) >> 16)) <= 0x10 && \
SDL_abs((Sint8)((gamepad.sThumbRX & 0xF000) >> 8) - ((match_state & 0x0F000000) >> 20)) <= 0x10 && \
SDL_abs((Sint8)((~gamepad.sThumbRY & 0xF000) >> 8) - ((match_state & 0xF0000000) >> 24)) <= 0x10) */
state->xinput_buttons =
/* Bitwise map .RLDUWVQTS.KYXBA -> YXBA..WVQTKSRLDU */
match_state << 12 | (match_state & 0x0780) >> 1 | (match_state & 0x0010) << 1 | (match_state & 0x0040) >> 2 | (match_state & 0x7800) >> 11;
/* Explicit
((match_state & (1<<SDL_CONTROLLER_BUTTON_A)) ? XINPUT_GAMEPAD_A : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_B)) ? XINPUT_GAMEPAD_B : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_X)) ? XINPUT_GAMEPAD_X : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_Y)) ? XINPUT_GAMEPAD_Y : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_BACK)) ? XINPUT_GAMEPAD_BACK : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_START)) ? XINPUT_GAMEPAD_START : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_LEFTSTICK)) ? XINPUT_GAMEPAD_LEFT_THUMB : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_RIGHTSTICK)) ? XINPUT_GAMEPAD_RIGHT_THUMB: 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_LEFTSHOULDER)) ? XINPUT_GAMEPAD_LEFT_SHOULDER : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_RIGHTSHOULDER)) ? XINPUT_GAMEPAD_RIGHT_SHOULDER : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_DPAD_UP)) ? XINPUT_GAMEPAD_DPAD_UP : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_DPAD_DOWN)) ? XINPUT_GAMEPAD_DPAD_DOWN : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_DPAD_LEFT)) ? XINPUT_GAMEPAD_DPAD_LEFT : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_DPAD_RIGHT)) ? XINPUT_GAMEPAD_DPAD_RIGHT : 0);
*/
if (state->xinput_buttons)
state->any_data = SDL_TRUE;
#endif
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
/* Match axes by checking if the distance between the high 4 bits of axis and the 4 bits from match_state is 1 or less */
#define WindowsGamingInputAxesMatch(gamepad) (\
(Uint16)(((Sint16)(gamepad.LeftThumbstickX * SDL_MAX_SINT16) & 0xF000) - state->match_axes[0] + 0x1000) <= 0x2fff && \
(Uint16)((~(Sint16)(gamepad.LeftThumbstickY * SDL_MAX_SINT16) & 0xF000) - state->match_axes[1] + 0x1000) <= 0x2fff && \
(Uint16)(((Sint16)(gamepad.RightThumbstickX * SDL_MAX_SINT16) & 0xF000) - state->match_axes[2] + 0x1000) <= 0x2fff && \
(Uint16)((~(Sint16)(gamepad.RightThumbstickY * SDL_MAX_SINT16) & 0xF000) - state->match_axes[3] + 0x1000) <= 0x2fff)
state->wgi_buttons =
/* Bitwise map .RLD UWVQ TS.K YXBA -> ..QT WVRL DUYX BAKS */
/* RStick/LStick (QT) RShould/LShould (WV) DPad R/L/D/U YXBA bac(K) (S)tart */
(match_state & 0x0180) << 5 | (match_state & 0x0600) << 1 | (match_state & 0x7800) >> 5 | (match_state & 0x000F) << 2 | (match_state & 0x0010) >> 3 | (match_state & 0x0040) >> 6;
/* Explicit
((match_state & (1<<SDL_CONTROLLER_BUTTON_A)) ? GamepadButtons_A : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_B)) ? GamepadButtons_B : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_X)) ? GamepadButtons_X : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_Y)) ? GamepadButtons_Y : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_BACK)) ? GamepadButtons_View : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_START)) ? GamepadButtons_Menu : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_LEFTSTICK)) ? GamepadButtons_LeftThumbstick : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_RIGHTSTICK)) ? GamepadButtons_RightThumbstick: 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_LEFTSHOULDER)) ? GamepadButtons_LeftShoulder: 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_RIGHTSHOULDER)) ? GamepadButtons_RightShoulder: 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_DPAD_UP)) ? GamepadButtons_DPadUp : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_DPAD_DOWN)) ? GamepadButtons_DPadDown : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_DPAD_LEFT)) ? GamepadButtons_DPadLeft : 0) |
((match_state & (1<<SDL_CONTROLLER_BUTTON_DPAD_RIGHT)) ? GamepadButtons_DPadRight : 0); */
if (state->wgi_buttons)
state->any_data = SDL_TRUE;
#endif
}
#endif /* SDL_JOYSTICK_RAWINPUT_MATCHING */
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
static struct {
XINPUT_STATE_EX state;
SDL_bool connected; /* Currently has an active XInput device */
SDL_bool used; /* Is currently mapped to an SDL device */
Uint8 correlation_id;
} xinput_state[XUSER_MAX_COUNT];
static SDL_bool xinput_device_change = SDL_TRUE;
static SDL_bool xinput_state_dirty = SDL_TRUE;
static void
RAWINPUT_UpdateXInput()
{
DWORD user_index;
if (xinput_device_change) {
for (user_index = 0; user_index < XUSER_MAX_COUNT; user_index++) {
XINPUT_CAPABILITIES capabilities;
xinput_state[user_index].connected = (XINPUTGETCAPABILITIES(user_index, XINPUT_FLAG_GAMEPAD, &capabilities) == ERROR_SUCCESS) ? SDL_TRUE : SDL_FALSE;
}
xinput_device_change = SDL_FALSE;
xinput_state_dirty = SDL_TRUE;
}
if (xinput_state_dirty) {
xinput_state_dirty = SDL_FALSE;
for (user_index = 0; user_index < SDL_arraysize(xinput_state); ++user_index) {
if (xinput_state[user_index].connected) {
if (XINPUTGETSTATE(user_index, &xinput_state[user_index].state) != ERROR_SUCCESS) {
xinput_state[user_index].connected = SDL_FALSE;
}
}
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
static void
RAWINPUT_MarkXInputSlotUsed(Uint8 xinput_slot)
{
if (xinput_slot != XUSER_INDEX_ANY) {
xinput_state[xinput_slot].used = SDL_TRUE;
}
}
static void
RAWINPUT_MarkXInputSlotFree(Uint8 xinput_slot)
{
if (xinput_slot != XUSER_INDEX_ANY) {
xinput_state[xinput_slot].used = SDL_FALSE;
}
}
static SDL_bool
RAWINPUT_MissingXInputSlot()
{
int ii;
for (ii = 0; ii < SDL_arraysize(xinput_state); ii++) {
if (xinput_state[ii].connected && !xinput_state[ii].used) {
return SDL_TRUE;
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
return SDL_FALSE;
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
static SDL_bool
RAWINPUT_XInputSlotMatches(const WindowsMatchState *state, Uint8 slot_idx)
{
if (xinput_state[slot_idx].connected) {
WORD xinput_buttons = xinput_state[slot_idx].state.Gamepad.wButtons;
if ((xinput_buttons & ~XINPUT_GAMEPAD_GUIDE) == state->xinput_buttons && XInputAxesMatch(xinput_state[slot_idx].state.Gamepad)) {
return SDL_TRUE;
}
}
return SDL_FALSE;
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
static SDL_bool
RAWINPUT_GuessXInputSlot(const WindowsMatchState *state, Uint8 *correlation_id, Uint8 *slot_idx)
{
int user_index;
int match_count;
*slot_idx = 0;
match_count = 0;
for (user_index = 0; user_index < XUSER_MAX_COUNT; ++user_index) {
if (!xinput_state[user_index].used && RAWINPUT_XInputSlotMatches(state, user_index)) {
++match_count;
*slot_idx = (Uint8)user_index;
/* Incrementing correlation_id for any match, as negative evidence for others being correlated */
*correlation_id = ++xinput_state[user_index].correlation_id;
}
}
/* Only return a match if we match exactly one, and we have some non-zero data (buttons or axes) that matched.
Note that we're still invalidating *other* potential correlations if we have more than one match or we have no
data. */
if (match_count == 1 && state->any_data) {
return SDL_TRUE;
}
return SDL_FALSE;
}
#endif /* SDL_JOYSTICK_RAWINPUT_XINPUT */
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
typedef struct WindowsGamingInputGamepadState {
__x_ABI_CWindows_CGaming_CInput_CIGamepad *gamepad;
struct __x_ABI_CWindows_CGaming_CInput_CGamepadReading state;
RAWINPUT_DeviceContext *correlated_context;
SDL_bool used; /* Is currently mapped to an SDL device */
SDL_bool connected; /* Just used during update to track disconnected */
Uint8 correlation_id;
struct __x_ABI_CWindows_CGaming_CInput_CGamepadVibration vibration;
} WindowsGamingInputGamepadState;
static struct {
WindowsGamingInputGamepadState **per_gamepad;
int per_gamepad_count;
SDL_bool initialized;
SDL_bool dirty;
SDL_bool need_device_list_update;
int ref_count;
__x_ABI_CWindows_CGaming_CInput_CIGamepadStatics *gamepad_statics;
} wgi_state;
static void
RAWINPUT_MarkWindowsGamingInputSlotUsed(WindowsGamingInputGamepadState *wgi_slot, RAWINPUT_DeviceContext *ctx)
{
wgi_slot->used = SDL_TRUE;
wgi_slot->correlated_context = ctx;
}
static void
RAWINPUT_MarkWindowsGamingInputSlotFree(WindowsGamingInputGamepadState *wgi_slot)
{
wgi_slot->used = SDL_FALSE;
wgi_slot->correlated_context = NULL;
}
static SDL_bool
RAWINPUT_MissingWindowsGamingInputSlot()
{
int ii;
for (ii = 0; ii < wgi_state.per_gamepad_count; ii++) {
if (!wgi_state.per_gamepad[ii]->used) {
return SDL_TRUE;
}
}
return SDL_FALSE;
}
static void
RAWINPUT_UpdateWindowsGamingInput()
{
int ii;
if (!wgi_state.gamepad_statics)
return;
if (!wgi_state.dirty)
return;
wgi_state.dirty = SDL_FALSE;
if (wgi_state.need_device_list_update) {
wgi_state.need_device_list_update = SDL_FALSE;
for (ii = 0; ii < wgi_state.per_gamepad_count; ii++) {
wgi_state.per_gamepad[ii]->connected = SDL_FALSE;
}
HRESULT hr;
__FIVectorView_1_Windows__CGaming__CInput__CGamepad *gamepads;
hr = __x_ABI_CWindows_CGaming_CInput_CIGamepadStatics_get_Gamepads(wgi_state.gamepad_statics, &gamepads);
if (SUCCEEDED(hr)) {
unsigned int num_gamepads;
hr = __FIVectorView_1_Windows__CGaming__CInput__CGamepad_get_Size(gamepads, &num_gamepads);
if (SUCCEEDED(hr)) {
unsigned int i;
for (i = 0; i < num_gamepads; ++i) {
__x_ABI_CWindows_CGaming_CInput_CIGamepad *gamepad;
hr = __FIVectorView_1_Windows__CGaming__CInput__CGamepad_GetAt(gamepads, i, &gamepad);
if (SUCCEEDED(hr)) {
SDL_bool found = SDL_FALSE;
int jj;
for (jj = 0; jj < wgi_state.per_gamepad_count ; jj++) {
if (wgi_state.per_gamepad[jj]->gamepad == gamepad) {
found = SDL_TRUE;
wgi_state.per_gamepad[jj]->connected = SDL_TRUE;
break;
}
}
if (!found) {
/* New device, add it */
wgi_state.per_gamepad_count++;
wgi_state.per_gamepad = SDL_realloc(wgi_state.per_gamepad, sizeof(wgi_state.per_gamepad[0]) * wgi_state.per_gamepad_count);
if (!wgi_state.per_gamepad) {
SDL_OutOfMemory();
return;
}
WindowsGamingInputGamepadState *gamepad_state = SDL_calloc(1, sizeof(*gamepad_state));
if (!gamepad_state) {
SDL_OutOfMemory();
return;
}
wgi_state.per_gamepad[wgi_state.per_gamepad_count - 1] = gamepad_state;
gamepad_state->gamepad = gamepad;
gamepad_state->connected = SDL_TRUE;
} else {
/* Already tracked */
__x_ABI_CWindows_CGaming_CInput_CIGamepad_Release(gamepad);
}
}
}
for (ii = wgi_state.per_gamepad_count - 1; ii >= 0; ii--) {
WindowsGamingInputGamepadState *gamepad_state = wgi_state.per_gamepad[ii];
if (!gamepad_state->connected) {
/* Device missing, must be disconnected */
if (gamepad_state->correlated_context) {
gamepad_state->correlated_context->wgi_correlated = SDL_FALSE;
gamepad_state->correlated_context->wgi_slot = NULL;
}
__x_ABI_CWindows_CGaming_CInput_CIGamepad_Release(gamepad_state->gamepad);
SDL_free(gamepad_state);
wgi_state.per_gamepad[ii] = wgi_state.per_gamepad[wgi_state.per_gamepad_count - 1];
--wgi_state.per_gamepad_count;
}
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
__FIVectorView_1_Windows__CGaming__CInput__CGamepad_Release(gamepads);
}
} /* need_device_list_update */
for (ii = 0; ii < wgi_state.per_gamepad_count; ii++) {
HRESULT hr = __x_ABI_CWindows_CGaming_CInput_CIGamepad_GetCurrentReading(wgi_state.per_gamepad[ii]->gamepad, &wgi_state.per_gamepad[ii]->state);
if (!SUCCEEDED(hr)) {
wgi_state.per_gamepad[ii]->connected = SDL_FALSE; /* Not used by anything, currently */
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
}
}
static void
RAWINPUT_InitWindowsGamingInput(RAWINPUT_DeviceContext *ctx)
{
wgi_state.need_device_list_update = SDL_TRUE;
wgi_state.ref_count++;
if (!wgi_state.initialized) {
/* I think this takes care of RoInitialize() in a way that is compatible with the rest of SDL */
if (FAILED(WIN_CoInitialize())) {
return;
}
wgi_state.initialized = SDL_TRUE;
wgi_state.dirty = SDL_TRUE;
static const IID SDL_IID_IGamepadStatics = { 0x8BBCE529, 0xD49C, 0x39E9, { 0x95, 0x60, 0xE4, 0x7D, 0xDE, 0x96, 0xB7, 0xC8 } };
HRESULT hr;
HMODULE hModule = LoadLibraryA("combase.dll");
if (hModule != NULL) {
typedef HRESULT (WINAPI *WindowsCreateString_t)(PCNZWCH sourceString, UINT32 length, HSTRING* string);
typedef HRESULT (WINAPI *WindowsDeleteString_t)(HSTRING string);
typedef HRESULT (WINAPI *RoGetActivationFactory_t)(HSTRING activatableClassId, REFIID iid, void** factory);
WindowsCreateString_t WindowsCreateStringFunc = (WindowsCreateString_t)GetProcAddress(hModule, "WindowsCreateString");
WindowsDeleteString_t WindowsDeleteStringFunc = (WindowsDeleteString_t)GetProcAddress(hModule, "WindowsDeleteString");
RoGetActivationFactory_t RoGetActivationFactoryFunc = (RoGetActivationFactory_t)GetProcAddress(hModule, "RoGetActivationFactory");
if (WindowsCreateStringFunc && WindowsDeleteStringFunc && RoGetActivationFactoryFunc) {
LPTSTR pNamespace = L"Windows.Gaming.Input.Gamepad";
HSTRING hNamespaceString;
hr = WindowsCreateStringFunc(pNamespace, SDL_wcslen(pNamespace), &hNamespaceString);
if (SUCCEEDED(hr)) {
RoGetActivationFactoryFunc(hNamespaceString, &SDL_IID_IGamepadStatics, &wgi_state.gamepad_statics);
WindowsDeleteStringFunc(hNamespaceString);
}
}
FreeLibrary(hModule);
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
}
static SDL_bool
RAWINPUT_WindowsGamingInputSlotMatches(const WindowsMatchState *state, WindowsGamingInputGamepadState *slot)
{
Uint32 wgi_buttons = slot->state.Buttons;
if ((wgi_buttons & 0x3FFF) == state->wgi_buttons && WindowsGamingInputAxesMatch(slot->state)) {
return SDL_TRUE;
}
return SDL_FALSE;
}
static SDL_bool
RAWINPUT_GuessWindowsGamingInputSlot(const WindowsMatchState *state, Uint8 *correlation_id, WindowsGamingInputGamepadState **slot)
{
int match_count;
match_count = 0;
for (int user_index = 0; user_index < wgi_state.per_gamepad_count; ++user_index) {
WindowsGamingInputGamepadState *gamepad_state = wgi_state.per_gamepad[user_index];
if (RAWINPUT_WindowsGamingInputSlotMatches(state, gamepad_state)) {
++match_count;
*slot = gamepad_state;
/* Incrementing correlation_id for any match, as negative evidence for others being correlated */
*correlation_id = ++gamepad_state->correlation_id;
}
}
/* Only return a match if we match exactly one, and we have some non-zero data (buttons or axes) that matched.
Note that we're still invalidating *other* potential correlations if we have more than one match or we have no
data. */
if (match_count == 1 && state->any_data) {
return SDL_TRUE;
}
return SDL_FALSE;
}
static void
RAWINPUT_QuitWindowsGamingInput(RAWINPUT_DeviceContext *ctx)
{
wgi_state.need_device_list_update = SDL_TRUE;
--wgi_state.ref_count;
if (!wgi_state.ref_count && wgi_state.initialized) {
for (int ii = 0; ii < wgi_state.per_gamepad_count; ii++) {
__x_ABI_CWindows_CGaming_CInput_CIGamepad_Release(wgi_state.per_gamepad[ii]->gamepad);
}
if (wgi_state.per_gamepad) {
SDL_free(wgi_state.per_gamepad);
wgi_state.per_gamepad = NULL;
}
wgi_state.per_gamepad_count = 0;
if (wgi_state.gamepad_statics) {
__x_ABI_CWindows_CGaming_CInput_CIGamepadStatics_Release(wgi_state.gamepad_statics);
wgi_state.gamepad_statics = NULL;
}
WIN_CoUninitialize();
wgi_state.initialized = SDL_FALSE;
}
}
#endif /* SDL_JOYSTICK_RAWINPUT_GAMING_INPUT */
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
static int
RAWINPUT_JoystickInit(void)
{
int ii;
RAWINPUTDEVICE rid[SDL_arraysize(subscribed_devices)];
SDL_assert(!SDL_RAWINPUT_inited);
SDL_assert(SDL_HelperWindow);
if (!SDL_GetHintBoolean(SDL_HINT_JOYSTICK_RAWINPUT, SDL_TRUE)) {
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
return -1;
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
if (WIN_LoadHIDDLL() < 0) {
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
return -1;
}
for (ii = 0; ii < SDL_arraysize(subscribed_devices); ii++) {
rid[ii].usUsagePage = USB_USAGEPAGE_GENERIC_DESKTOP;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
rid[ii].usUsage = subscribed_devices[ii];
rid[ii].dwFlags = RIDEV_DEVNOTIFY | RIDEV_INPUTSINK; /* Receive messages when in background, including device add/remove */
rid[ii].hwndTarget = SDL_HelperWindow;
}
if (!RegisterRawInputDevices(rid, SDL_arraysize(rid), sizeof(RAWINPUTDEVICE))) {
SDL_SetError("Couldn't initialize RAWINPUT");
WIN_UnloadHIDDLL();
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
return -1;
}
SDL_RAWINPUT_inited = SDL_TRUE;
RAWINPUT_JoystickDetect();
RAWINPUT_PumpMessages();
return 0;
}
static int
RAWINPUT_JoystickGetCount(void)
{
return SDL_RAWINPUT_numjoysticks;
}
static SDL_RAWINPUT_Device *
RAWINPUT_AcquireDevice(SDL_RAWINPUT_Device *device)
{
SDL_AtomicIncRef(&device->refcount);
return device;
}
static void
RAWINPUT_ReleaseDevice(SDL_RAWINPUT_Device *device)
{
if (SDL_AtomicDecRef(&device->refcount)) {
if (device->preparsed_data) {
SDL_HidD_FreePreparsedData(device->preparsed_data);
}
SDL_free(device->name);
SDL_free(device);
}
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
static SDL_RAWINPUT_Device *
RAWINPUT_DeviceFromHandle(HANDLE hDevice)
{
SDL_RAWINPUT_Device *curr;
for (curr = SDL_RAWINPUT_devices; curr; curr = curr->next) {
if (curr->hDevice == hDevice)
return curr;
}
return NULL;
}
static void
RAWINPUT_AddDevice(HANDLE hDevice)
{
#define CHECK(exp) { if(!(exp)) goto err; }
SDL_RAWINPUT_Device *device = NULL;
SDL_RAWINPUT_Device *curr, *last;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
RID_DEVICE_INFO rdi;
UINT rdi_size = sizeof(rdi);
char dev_name[MAX_PATH];
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
UINT name_size = SDL_arraysize(dev_name);
HANDLE hFile = INVALID_HANDLE_VALUE;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
/* Make sure we're not trying to add the same device twice */
if (RAWINPUT_DeviceFromHandle(hDevice)) {
return;
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
/* Figure out what kind of device it is */
CHECK(GetRawInputDeviceInfoA(hDevice, RIDI_DEVICEINFO, &rdi, &rdi_size) != (UINT)-1);
CHECK(rdi.dwType == RIM_TYPEHID);
/* Get the device "name" (HID Path) */
CHECK(GetRawInputDeviceInfoA(hDevice, RIDI_DEVICENAME, dev_name, &name_size) != (UINT)-1);
/* Only take XInput-capable devices */
CHECK(SDL_strstr(dev_name, "IG_") != NULL);
#ifdef SDL_JOYSTICK_HIDAPI
/* Don't take devices handled by HIDAPI */
CHECK(!HIDAPI_IsDevicePresent((Uint16)rdi.hid.dwVendorId, (Uint16)rdi.hid.dwProductId, (Uint16)rdi.hid.dwVersionNumber, ""));
#endif
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
CHECK(device = (SDL_RAWINPUT_Device *)SDL_calloc(1, sizeof(SDL_RAWINPUT_Device)));
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
device->hDevice = hDevice;
device->vendor_id = (Uint16)rdi.hid.dwVendorId;
device->product_id = (Uint16)rdi.hid.dwProductId;
device->version = (Uint16)rdi.hid.dwVersionNumber;
device->is_xinput = SDL_TRUE;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
{
const Uint16 vendor = device->vendor_id;
const Uint16 product = device->product_id;
const Uint16 version = device->version;
Uint16 *guid16 = (Uint16 *)device->guid.data;
*guid16++ = SDL_SwapLE16(SDL_HARDWARE_BUS_USB);
*guid16++ = 0;
*guid16++ = SDL_SwapLE16(vendor);
*guid16++ = 0;
*guid16++ = SDL_SwapLE16(product);
*guid16++ = 0;
*guid16++ = SDL_SwapLE16(version);
*guid16++ = 0;
/* Note that this is a RAWINPUT device for special handling elsewhere */
device->guid.data[14] = 'r';
device->guid.data[15] = 0;
}
hFile = CreateFileA(dev_name, GENERIC_READ, FILE_SHARE_READ|FILE_SHARE_WRITE, NULL, OPEN_EXISTING, 0, NULL);
CHECK(hFile != INVALID_HANDLE_VALUE);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
{
char *manufacturer_string = NULL;
char *product_string = NULL;
WCHAR string[128];
if (SDL_HidD_GetManufacturerString(hFile, string, sizeof(string))) {
manufacturer_string = WIN_StringToUTF8(string);
}
if (SDL_HidD_GetProductString(hFile, string, sizeof(string))) {
product_string = WIN_StringToUTF8(string);
}
device->name = SDL_CreateJoystickName(device->vendor_id, device->product_id, manufacturer_string, product_string);
if (manufacturer_string) {
SDL_free(manufacturer_string);
}
if (product_string) {
SDL_free(product_string);
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
CHECK(SDL_HidD_GetPreparsedData(hFile, &device->preparsed_data));
CloseHandle(hFile);
hFile = INVALID_HANDLE_VALUE;
device->joystick_id = SDL_GetNextJoystickInstanceID();
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
#ifdef DEBUG_RAWINPUT
SDL_Log("Adding RAWINPUT device '%s' VID 0x%.4x, PID 0x%.4x, version %d, handle 0x%.8x\n", device->name, device->vendor_id, device->product_id, device->version, device->hDevice);
#endif
/* Add it to the list */
RAWINPUT_AcquireDevice(device);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
for (curr = SDL_RAWINPUT_devices, last = NULL; curr; last = curr, curr = curr->next) {
continue;
}
if (last) {
last->next = device;
} else {
SDL_RAWINPUT_devices = device;
}
++SDL_RAWINPUT_numjoysticks;
SDL_PrivateJoystickAdded(device->joystick_id);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
return;
err:
if (hFile != INVALID_HANDLE_VALUE) {
CloseHandle(hFile);
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
if (device) {
if (device->name)
SDL_free(device->name);
SDL_free(device);
}
#undef CHECK
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
static void
RAWINPUT_DelDevice(SDL_RAWINPUT_Device *device, SDL_bool send_event)
{
SDL_RAWINPUT_Device *curr, *last;
for (curr = SDL_RAWINPUT_devices, last = NULL; curr; last = curr, curr = curr->next) {
if (curr == device) {
if (last) {
last->next = curr->next;
} else {
SDL_RAWINPUT_devices = curr->next;
}
--SDL_RAWINPUT_numjoysticks;
SDL_PrivateJoystickRemoved(device->joystick_id);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
#ifdef DEBUG_RAWINPUT
SDL_Log("Removing RAWINPUT device '%s' VID 0x%.4x, PID 0x%.4x, version %d, handle 0x%.8x\n", device->name, device->vendor_id, device->product_id, device->version, device->hDevice);
#endif
RAWINPUT_ReleaseDevice(device);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
return;
}
}
}
static void
RAWINPUT_PumpMessages(void)
{
if (SDL_RAWINPUT_need_pump) {
MSG msg;
while (PeekMessage(&msg, SDL_HelperWindow, WM_INPUT, WM_INPUT, PM_REMOVE)) {
TranslateMessage(&msg);
DispatchMessage(&msg);
}
SDL_RAWINPUT_need_pump = SDL_FALSE;
}
}
static void
RAWINPUT_UpdateDeviceList(void)
{
MSG msg;
/* In theory, want only WM_INPUT_DEVICE_CHANGE messages here, but PeekMessage returns nothing unless you also ask
for WM_INPUT */
while (PeekMessage(&msg, SDL_HelperWindow, WM_INPUT_DEVICE_CHANGE, WM_INPUT, PM_REMOVE)) {
TranslateMessage(&msg);
DispatchMessage(&msg);
}
}
static void
RAWINPUT_PostUpdate(void)
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
{
#ifdef SDL_JOYSTICK_RAWINPUT_MATCHING
SDL_bool unmapped_guide_pressed = SDL_FALSE;
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
if (!wgi_state.dirty) {
int ii;
for (ii = 0; ii < wgi_state.per_gamepad_count; ii++) {
WindowsGamingInputGamepadState *gamepad_state = wgi_state.per_gamepad[ii];
if (!gamepad_state->used && (gamepad_state->state.Buttons & GamepadButtons_GUIDE)) {
unmapped_guide_pressed = SDL_TRUE;
break;
}
}
}
wgi_state.dirty = SDL_TRUE;
#endif
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
if (!xinput_state_dirty) {
int ii;
for (ii = 0; ii < SDL_arraysize(xinput_state); ii++) {
if (xinput_state[ii].connected && !xinput_state[ii].used && (xinput_state[ii].state.Gamepad.wButtons & XINPUT_GAMEPAD_GUIDE)) {
unmapped_guide_pressed = SDL_TRUE;
break;
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
}
xinput_state_dirty = SDL_TRUE;
#endif
if (unmapped_guide_pressed) {
if (guide_button_candidate.joystick && !guide_button_candidate.last_joystick) {
SDL_Joystick *joystick = guide_button_candidate.joystick;
RAWINPUT_DeviceContext *ctx = joystick->hwdata;
if (ctx->guide_hack) {
int guide_button = joystick->nbuttons - 1;
SDL_PrivateJoystickButton(guide_button_candidate.joystick, guide_button, SDL_PRESSED);
}
guide_button_candidate.last_joystick = guide_button_candidate.joystick;
}
} else if (guide_button_candidate.last_joystick) {
SDL_Joystick *joystick = guide_button_candidate.last_joystick;
RAWINPUT_DeviceContext *ctx = joystick->hwdata;
if (ctx->guide_hack) {
int guide_button = joystick->nbuttons - 1;
SDL_PrivateJoystickButton(joystick, guide_button, SDL_RELEASED);
}
guide_button_candidate.last_joystick = NULL;
}
guide_button_candidate.joystick = NULL;
#endif /* SDL_JOYSTICK_RAWINPUT_MATCHING */
}
SDL_bool
RAWINPUT_IsEnabled()
{
return SDL_RAWINPUT_inited;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
SDL_bool
RAWINPUT_IsDevicePresent(Uint16 vendor_id, Uint16 product_id, Uint16 version)
{
SDL_RAWINPUT_Device *device;
/* Make sure the device list is completely up to date when we check for device presence */
RAWINPUT_UpdateDeviceList();
device = SDL_RAWINPUT_devices;
while (device) {
if (device->vendor_id == vendor_id && device->product_id == product_id) {
return SDL_TRUE;
}
device = device->next;
}
return SDL_FALSE;
}
static void
RAWINPUT_JoystickDetect(void)
{
/* Just ensure the window's add/remove messages have been pumped */
RAWINPUT_UpdateDeviceList();
RAWINPUT_PostUpdate();
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
SDL_RAWINPUT_need_pump = SDL_TRUE;
}
static SDL_RAWINPUT_Device *
RAWINPUT_GetDeviceByIndex(int device_index)
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
{
SDL_RAWINPUT_Device *device = SDL_RAWINPUT_devices;
while (device) {
if (device_index == 0) {
break;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
--device_index;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
device = device->next;
}
return device;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
static const char *
RAWINPUT_JoystickGetDeviceName(int device_index)
{
return RAWINPUT_GetDeviceByIndex(device_index)->name;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
static int
RAWINPUT_JoystickGetDevicePlayerIndex(int device_index)
{
return -1;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
static void
RAWINPUT_JoystickSetDevicePlayerIndex(int device_index, int player_index)
{
}
static SDL_JoystickGUID
RAWINPUT_JoystickGetDeviceGUID(int device_index)
{
return RAWINPUT_GetDeviceByIndex(device_index)->guid;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
static SDL_JoystickID
RAWINPUT_JoystickGetDeviceInstanceID(int device_index)
{
return RAWINPUT_GetDeviceByIndex(device_index)->joystick_id;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
static int
RAWINPUT_JoystickOpen(SDL_Joystick *joystick, int device_index)
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
{
SDL_RAWINPUT_Device *device = RAWINPUT_GetDeviceByIndex(device_index);
RAWINPUT_DeviceContext *ctx;
HIDP_CAPS caps;
HIDP_BUTTON_CAPS *button_caps;
HIDP_VALUE_CAPS *value_caps;
ULONG i;
ctx = (RAWINPUT_DeviceContext *)SDL_calloc(1, sizeof(RAWINPUT_DeviceContext));
if (!ctx) {
return SDL_OutOfMemory();
}
joystick->hwdata = ctx;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
ctx->device = RAWINPUT_AcquireDevice(device);
device->joystick = joystick;
if (device->is_xinput) {
/* We'll try to get guide button and trigger axes from XInput */
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
xinput_device_change = SDL_TRUE;
ctx->xinput_enabled = SDL_GetHintBoolean(SDL_HINT_JOYSTICK_HIDAPI_CORRELATE_XINPUT, SDL_TRUE);
if (ctx->xinput_enabled && (WIN_LoadXInputDLL() < 0 || !XINPUTGETSTATE)) {
ctx->xinput_enabled = SDL_FALSE;
}
ctx->xinput_slot = XUSER_INDEX_ANY;
#endif
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
RAWINPUT_InitWindowsGamingInput(ctx);
#endif
}
ctx->is_xinput = device->is_xinput;
ctx->preparsed_data = device->preparsed_data;
ctx->max_data_length = SDL_HidP_MaxDataListLength(HidP_Input, ctx->preparsed_data);
ctx->data = (HIDP_DATA *)SDL_malloc(ctx->max_data_length * sizeof(*ctx->data));
if (!ctx->data) {
RAWINPUT_JoystickClose(joystick);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
return SDL_OutOfMemory();
}
if (SDL_HidP_GetCaps(ctx->preparsed_data, &caps) != HIDP_STATUS_SUCCESS) {
RAWINPUT_JoystickClose(joystick);
return SDL_SetError("Couldn't get device capabilities");
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
button_caps = SDL_stack_alloc(HIDP_BUTTON_CAPS, caps.NumberInputButtonCaps);
if (SDL_HidP_GetButtonCaps(HidP_Input, button_caps, &caps.NumberInputButtonCaps, ctx->preparsed_data) != HIDP_STATUS_SUCCESS) {
RAWINPUT_JoystickClose(joystick);
return SDL_SetError("Couldn't get device button capabilities");
}
value_caps = SDL_stack_alloc(HIDP_VALUE_CAPS, caps.NumberInputValueCaps);
if (SDL_HidP_GetValueCaps(HidP_Input, value_caps, &caps.NumberInputValueCaps, ctx->preparsed_data) != HIDP_STATUS_SUCCESS) {
RAWINPUT_JoystickClose(joystick);
return SDL_SetError("Couldn't get device value capabilities");
}
for (i = 0; i < caps.NumberInputButtonCaps; ++i) {
HIDP_BUTTON_CAPS *cap = &button_caps[i];
if (cap->UsagePage == USB_USAGEPAGE_BUTTON) {
int count;
if (cap->IsRange) {
count = 1 + (cap->Range.DataIndexMax - cap->Range.DataIndexMin);
} else {
count = 1;
}
joystick->nbuttons += count;
}
}
if (joystick->nbuttons > 0) {
int button_index = 0;
ctx->button_indices = (USHORT *)SDL_malloc(joystick->nbuttons * sizeof(*ctx->button_indices));
if (!ctx->button_indices) {
RAWINPUT_JoystickClose(joystick);
return SDL_OutOfMemory();
}
for (i = 0; i < caps.NumberInputButtonCaps; ++i) {
HIDP_BUTTON_CAPS *cap = &button_caps[i];
if (cap->UsagePage == USB_USAGEPAGE_BUTTON) {
if (cap->IsRange) {
int j, count = 1 + (cap->Range.DataIndexMax - cap->Range.DataIndexMin);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
for (j = 0; j < count; ++j) {
ctx->button_indices[button_index++] = cap->Range.DataIndexMin + j;
}
} else {
ctx->button_indices[button_index++] = cap->NotRange.DataIndex;
}
}
}
}
if (ctx->is_xinput && joystick->nbuttons == 10) {
ctx->guide_hack = SDL_TRUE;
joystick->nbuttons += 1;
}
for (i = 0; i < caps.NumberInputValueCaps; ++i) {
HIDP_VALUE_CAPS *cap = &value_caps[i];
if (cap->IsRange) {
continue;
}
if (ctx->trigger_hack && cap->NotRange.Usage == USB_USAGE_GENERIC_Z) {
continue;
}
if (cap->NotRange.Usage == USB_USAGE_GENERIC_HAT) {
joystick->nhats += 1;
continue;
}
if (ctx->is_xinput && cap->NotRange.Usage == USB_USAGE_GENERIC_Z) {
continue;
}
joystick->naxes += 1;
}
if (joystick->naxes > 0) {
int axis_index = 0;
ctx->axis_indices = (USHORT *)SDL_malloc(joystick->naxes * sizeof(*ctx->axis_indices));
if (!ctx->axis_indices) {
RAWINPUT_JoystickClose(joystick);
return SDL_OutOfMemory();
}
for (i = 0; i < caps.NumberInputValueCaps; ++i) {
HIDP_VALUE_CAPS *cap = &value_caps[i];
if (cap->IsRange) {
continue;
}
if (cap->NotRange.Usage == USB_USAGE_GENERIC_HAT) {
continue;
}
if (ctx->is_xinput && cap->NotRange.Usage == USB_USAGE_GENERIC_Z) {
ctx->trigger_hack = SDL_TRUE;
ctx->trigger_hack_index = cap->NotRange.DataIndex;
continue;
}
ctx->axis_indices[axis_index++] = cap->NotRange.DataIndex;
}
}
if (ctx->trigger_hack) {
joystick->naxes += 2;
}
if (joystick->nhats > 0) {
int hat_index = 0;
ctx->hat_indices = (USHORT *)SDL_malloc(joystick->nhats * sizeof(*ctx->hat_indices));
if (!ctx->hat_indices) {
RAWINPUT_JoystickClose(joystick);
return SDL_OutOfMemory();
}
for (i = 0; i < caps.NumberInputValueCaps; ++i) {
HIDP_VALUE_CAPS *cap = &value_caps[i];
if (cap->IsRange) {
continue;
}
if (cap->NotRange.Usage != USB_USAGE_GENERIC_HAT) {
continue;
}
ctx->hat_indices[hat_index++] = cap->NotRange.DataIndex;
}
}
joystick->epowerlevel = SDL_JOYSTICK_POWER_WIRED;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
return 0;
}
static int
RAWINPUT_JoystickRumble(SDL_Joystick *joystick, Uint16 low_frequency_rumble, Uint16 high_frequency_rumble)
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
{
#if defined(SDL_JOYSTICK_RAWINPUT_GAMING_INPUT) || defined(SDL_JOYSTICK_RAWINPUT_XINPUT)
RAWINPUT_DeviceContext *ctx = joystick->hwdata;
#endif
SDL_bool rumbled = SDL_FALSE;
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
if (!rumbled && ctx->wgi_correlated) {
WindowsGamingInputGamepadState *gamepad_state = ctx->wgi_slot;
HRESULT hr;
gamepad_state->vibration.LeftMotor = (DOUBLE)low_frequency_rumble / SDL_MAX_UINT16;
gamepad_state->vibration.RightMotor = (DOUBLE)high_frequency_rumble / SDL_MAX_UINT16;
hr = __x_ABI_CWindows_CGaming_CInput_CIGamepad_put_Vibration(gamepad_state->gamepad, gamepad_state->vibration);
if (SUCCEEDED(hr)) {
rumbled = SDL_TRUE;
}
}
#endif
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
if (!rumbled && ctx->xinput_correlated) {
XINPUT_VIBRATION XVibration;
if (!XINPUTSETSTATE) {
return SDL_Unsupported();
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
XVibration.wLeftMotorSpeed = low_frequency_rumble;
XVibration.wRightMotorSpeed = high_frequency_rumble;
if (XINPUTSETSTATE(ctx->xinput_slot, &XVibration) == ERROR_SUCCESS) {
rumbled = SDL_TRUE;
} else {
return SDL_SetError("XInputSetState() failed");
}
}
#endif /* SDL_JOYSTICK_RAWINPUT_XINPUT */
return 0;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
static int
RAWINPUT_JoystickRumbleTriggers(SDL_Joystick *joystick, Uint16 left_rumble, Uint16 right_rumble)
{
#if defined(SDL_JOYSTICK_RAWINPUT_GAMING_INPUT)
RAWINPUT_DeviceContext *ctx = joystick->hwdata;
if (ctx->wgi_correlated) {
WindowsGamingInputGamepadState *gamepad_state = ctx->wgi_slot;
HRESULT hr;
gamepad_state->vibration.LeftTrigger = (DOUBLE)left_rumble / SDL_MAX_UINT16;
gamepad_state->vibration.RightTrigger = (DOUBLE)right_rumble / SDL_MAX_UINT16;
hr = __x_ABI_CWindows_CGaming_CInput_CIGamepad_put_Vibration(gamepad_state->gamepad, gamepad_state->vibration);
if (!SUCCEEDED(hr)) {
return SDL_SetError("Setting vibration failed: 0x%x\n", hr);
}
}
return 0;
#else
return SDL_Unsupported();
#endif
}
static SDL_bool
RAWINPUT_JoystickHasLED(SDL_Joystick *joystick)
{
return SDL_FALSE;
}
static int
RAWINPUT_JoystickSetLED(SDL_Joystick *joystick, Uint8 red, Uint8 green, Uint8 blue)
{
return SDL_Unsupported();
}
static int
RAWINPUT_JoystickSetSensorsEnabled(SDL_Joystick *joystick, SDL_bool enabled)
{
return SDL_Unsupported();
}
static HIDP_DATA *GetData(USHORT index, HIDP_DATA *data, ULONG length)
{
ULONG i;
/* Check to see if the data is at the expected offset */
if (index < length && data[index].DataIndex == index) {
return &data[index];
}
/* Loop through the data to find it */
for (i = 0; i < length; ++i) {
if (data[i].DataIndex == index) {
return &data[i];
}
}
return NULL;
}
/* This is the packet format for Xbox 360 and Xbox One controllers on Windows,
however with this interface there is no rumble support, no guide button,
and the left and right triggers are tied together as a single axis.
We use XInput and Windows.Gaming.Input to make up for these shortcomings.
*/
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
static void
RAWINPUT_HandleStatePacket(SDL_Joystick *joystick, Uint8 *data, int size)
{
RAWINPUT_DeviceContext *ctx = joystick->hwdata;
#ifdef SDL_JOYSTICK_RAWINPUT_MATCHING
/* Map new buttons and axes into game controller controls */
static int axis_map[] = {
SDL_CONTROLLER_AXIS_LEFTY,
SDL_CONTROLLER_AXIS_LEFTX,
SDL_CONTROLLER_AXIS_RIGHTY,
SDL_CONTROLLER_AXIS_RIGHTX
};
static int button_map[] = {
SDL_CONTROLLER_BUTTON_A,
SDL_CONTROLLER_BUTTON_B,
SDL_CONTROLLER_BUTTON_X,
SDL_CONTROLLER_BUTTON_Y,
SDL_CONTROLLER_BUTTON_LEFTSHOULDER,
SDL_CONTROLLER_BUTTON_RIGHTSHOULDER,
SDL_CONTROLLER_BUTTON_BACK,
SDL_CONTROLLER_BUTTON_START,
SDL_CONTROLLER_BUTTON_LEFTSTICK,
SDL_CONTROLLER_BUTTON_RIGHTSTICK
};
Uint32 match_state = ctx->match_state;
/* Update match_state with button bit, then fall through */
#define SDL_PrivateJoystickButton(joystick, button, state) if (button < SDL_arraysize(button_map)) if (state) match_state |= 1 << button_map[button]; else match_state &= ~(1 << button_map[button]); SDL_PrivateJoystickButton(joystick, button, state)
/* Grab high 4 bits of value, then fall through */
#define SDL_PrivateJoystickAxis(joystick, axis, value) if (axis < SDL_arraysize(axis_map)) match_state = (match_state & ~(0xF << (4 * axis_map[axis] + 16))) | ((value) & 0xF000) << (4 * axis_map[axis] + 4); SDL_PrivateJoystickAxis(joystick, axis, value)
#endif
ULONG data_length = ctx->max_data_length;
int i;
int nbuttons = joystick->nbuttons - (ctx->guide_hack * 1);
int naxes = joystick->naxes - (ctx->trigger_hack * 2);
int nhats = joystick->nhats;
Uint32 button_mask = 0;
if (SDL_HidP_GetData(HidP_Input, ctx->data, &data_length, ctx->preparsed_data, (PCHAR)data, size) != HIDP_STATUS_SUCCESS) {
return;
}
for (i = 0; i < nbuttons; ++i) {
HIDP_DATA *item = GetData(ctx->button_indices[i], ctx->data, data_length);
if (item && item->On) {
button_mask |= (1 << i);
}
}
for (i = 0; i < nbuttons; ++i) {
SDL_PrivateJoystickButton(joystick, i, (button_mask & (1 << i)) ? SDL_PRESSED : SDL_RELEASED);
}
for (i = 0; i < naxes; ++i) {
HIDP_DATA *item = GetData(ctx->axis_indices[i], ctx->data, data_length);
if (item) {
Sint16 axis = (int)(Uint16)item->RawValue - 0x8000;
SDL_PrivateJoystickAxis(joystick, i, axis);
}
}
for (i = 0; i < nhats; ++i) {
HIDP_DATA *item = GetData(ctx->hat_indices[i], ctx->data, data_length);
if (item) {
const Uint8 hat_states[] = {
SDL_HAT_CENTERED,
SDL_HAT_UP,
SDL_HAT_UP | SDL_HAT_RIGHT,
SDL_HAT_RIGHT,
SDL_HAT_DOWN | SDL_HAT_RIGHT,
SDL_HAT_DOWN,
SDL_HAT_DOWN | SDL_HAT_LEFT,
SDL_HAT_LEFT,
SDL_HAT_UP | SDL_HAT_LEFT,
};
ULONG state = item->RawValue;
if (state < SDL_arraysize(hat_states)) {
SDL_PrivateJoystickHat(joystick, i, hat_states[state]);
}
}
}
#ifdef SDL_JOYSTICK_RAWINPUT_MATCHING
#undef SDL_PrivateJoystickButton
#undef SDL_PrivateJoystickAxis
#endif
if (ctx->trigger_hack) {
SDL_bool has_trigger_data = SDL_FALSE;
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
/* Prefer XInput over WindowsGamingInput, it continues to provide data in the background */
if (!has_trigger_data && ctx->xinput_enabled && ctx->xinput_correlated) {
has_trigger_data = SDL_TRUE;
}
#endif /* SDL_JOYSTICK_RAWINPUT_XINPUT */
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
if (!has_trigger_data && ctx->wgi_correlated) {
has_trigger_data = SDL_TRUE;
}
#endif /* SDL_JOYSTICK_RAWINPUT_GAMING_INPUT */
if (!has_trigger_data) {
HIDP_DATA *item = GetData(ctx->trigger_hack_index, ctx->data, data_length);
if (item) {
int left_trigger = joystick->naxes - 2;
int right_trigger = joystick->naxes - 1;
Sint16 value = (int)(Uint16)item->RawValue - 0x8000;
if (value < 0) {
value = -value * 2 - 32769;
SDL_PrivateJoystickAxis(joystick, left_trigger, SDL_MIN_SINT16);
SDL_PrivateJoystickAxis(joystick, right_trigger, value);
} else if (value > 0) {
value = value * 2 - 32767;
SDL_PrivateJoystickAxis(joystick, left_trigger, value);
SDL_PrivateJoystickAxis(joystick, right_trigger, SDL_MIN_SINT16);
} else {
SDL_PrivateJoystickAxis(joystick, left_trigger, SDL_MIN_SINT16);
SDL_PrivateJoystickAxis(joystick, right_trigger, SDL_MIN_SINT16);
}
}
}
}
#ifdef SDL_JOYSTICK_RAWINPUT_MATCHING
if (ctx->is_xinput) {
ctx->match_state = match_state;
ctx->last_state_packet = SDL_GetTicks();
}
#endif
}
static void
RAWINPUT_UpdateOtherAPIs(SDL_Joystick *joystick)
{
#ifdef SDL_JOYSTICK_RAWINPUT_MATCHING
RAWINPUT_DeviceContext *ctx = joystick->hwdata;
SDL_bool has_trigger_data = SDL_FALSE;
SDL_bool correlated = SDL_FALSE;
WindowsMatchState match_state_xinput;
int guide_button = joystick->nbuttons - 1;
int left_trigger = joystick->naxes - 2;
int right_trigger = joystick->naxes - 1;
RAWINPUT_FillMatchState(&match_state_xinput, ctx->match_state);
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
/* Parallel logic to WINDOWS_XINPUT below */
RAWINPUT_UpdateWindowsGamingInput();
if (ctx->wgi_correlated) {
/* We have been previously correlated, ensure we are still matching, see comments in XINPUT section */
if (RAWINPUT_WindowsGamingInputSlotMatches(&match_state_xinput, ctx->wgi_slot)) {
ctx->wgi_uncorrelate_count = 0;
} else {
++ctx->wgi_uncorrelate_count;
/* Only un-correlate if this is consistent over multiple Update() calls - the timing of polling/event
pumping can easily cause this to uncorrelate for a frame. 2 seemed reliable in my testing, but
let's set it to 3 to be safe. An incorrect un-correlation will simply result in lower precision
triggers for a frame. */
if (ctx->wgi_uncorrelate_count >= 3) {
#ifdef DEBUG_JOYSTICK
SDL_Log("UN-Correlated joystick %d to WindowsGamingInput device #%d\n", joystick->instance_id, ctx->wgi_slot);
#endif
RAWINPUT_MarkWindowsGamingInputSlotFree(ctx->wgi_slot);
ctx->wgi_correlated = SDL_FALSE;
ctx->wgi_correlation_count = 0;
/* Force release of Guide button, it can't possibly be down on this device now. */
/* It gets left down if we were actually correlated incorrectly and it was released on the WindowsGamingInput
device but we didn't get a state packet. */
if (ctx->guide_hack) {
SDL_PrivateJoystickButton(joystick, guide_button, SDL_RELEASED);
}
}
}
}
if (!ctx->wgi_correlated) {
SDL_bool new_correlation_count = 0;
if (RAWINPUT_MissingWindowsGamingInputSlot()) {
Uint8 correlation_id;
WindowsGamingInputGamepadState *slot_idx;
if (RAWINPUT_GuessWindowsGamingInputSlot(&match_state_xinput, &correlation_id, &slot_idx)) {
/* we match exactly one WindowsGamingInput device */
/* Probably can do without wgi_correlation_count, just check and clear wgi_slot to NULL, unless we need
even more frames to be sure. */
if (ctx->wgi_correlation_count && ctx->wgi_slot == slot_idx) {
/* was correlated previously, and still the same device */
if (ctx->wgi_correlation_id + 1 == correlation_id) {
/* no one else was correlated in the meantime */
new_correlation_count = ctx->wgi_correlation_count + 1;
if (new_correlation_count == 2) {
/* correlation stayed steady and uncontested across multiple frames, guaranteed match */
ctx->wgi_correlated = SDL_TRUE;
#ifdef DEBUG_JOYSTICK
SDL_Log("Correlated joystick %d to WindowsGamingInput device #%d\n", joystick->instance_id, slot_idx);
#endif
correlated = SDL_TRUE;
RAWINPUT_MarkWindowsGamingInputSlotUsed(ctx->wgi_slot, ctx);
/* If the generalized Guide button was using us, it doesn't need to anymore */
if (guide_button_candidate.joystick == joystick)
guide_button_candidate.joystick = NULL;
if (guide_button_candidate.last_joystick == joystick)
guide_button_candidate.last_joystick = NULL;
}
} else {
/* someone else also possibly correlated to this device, start over */
new_correlation_count = 1;
}
} else {
/* new possible correlation */
new_correlation_count = 1;
ctx->wgi_slot = slot_idx;
}
ctx->wgi_correlation_id = correlation_id;
} else {
/* Match multiple WindowsGamingInput devices, or none (possibly due to no buttons pressed) */
}
}
ctx->wgi_correlation_count = new_correlation_count;
} else {
correlated = SDL_TRUE;
}
#endif /* SDL_JOYSTICK_RAWINPUT_GAMING_INPUT */
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
/* Parallel logic to WINDOWS_GAMING_INPUT above */
if (ctx->xinput_enabled) {
RAWINPUT_UpdateXInput();
if (ctx->xinput_correlated) {
/* We have been previously correlated, ensure we are still matching */
/* This is required to deal with two (mostly) un-preventable mis-correlation situations:
A) Since the HID data stream does not provide an initial state (but polling XInput does), if we open
5 controllers (#1-4 XInput mapped, #5 is not), and controller 1 had the A button down (and we don't
know), and the user presses A on controller #5, we'll see exactly 1 controller with A down (#5) and
exactly 1 XInput device with A down (#1), and incorrectly correlate. This code will then un-correlate
when A is released from either controller #1 or #5.
B) Since the app may not open all controllers, we could have a similar situation where only controller #5
is opened, and the user holds A on controllers #1 and #5 simultaneously - again we see only 1 controller
with A down and 1 XInput device with A down, and incorrectly correlate. This should be very unusual
(only when apps do not open all controllers, yet are listening to Guide button presses, yet
for some reason want to ignore guide button presses on the un-opened controllers, yet users are
pressing buttons on the unopened controllers), and will resolve itself when either button is released
and we un-correlate. We could prevent this by processing the state packets for *all* controllers,
even un-opened ones, as that would allow more precise correlation.
*/
if (RAWINPUT_XInputSlotMatches(&match_state_xinput, ctx->xinput_slot)) {
ctx->xinput_uncorrelate_count = 0;
} else {
++ctx->xinput_uncorrelate_count;
/* Only un-correlate if this is consistent over multiple Update() calls - the timing of polling/event
pumping can easily cause this to uncorrelate for a frame. 2 seemed reliable in my testing, but
let's set it to 3 to be safe. An incorrect un-correlation will simply result in lower precision
triggers for a frame. */
if (ctx->xinput_uncorrelate_count >= 3) {
#ifdef DEBUG_JOYSTICK
SDL_Log("UN-Correlated joystick %d to XInput device #%d\n", joystick->instance_id, ctx->xinput_slot);
#endif
RAWINPUT_MarkXInputSlotFree(ctx->xinput_slot);
ctx->xinput_correlated = SDL_FALSE;
ctx->xinput_correlation_count = 0;
/* Force release of Guide button, it can't possibly be down on this device now. */
/* It gets left down if we were actually correlated incorrectly and it was released on the XInput
device but we didn't get a state packet. */
if (ctx->guide_hack) {
SDL_PrivateJoystickButton(joystick, guide_button, SDL_RELEASED);
}
}
}
}
if (!ctx->xinput_correlated) {
Uint8 new_correlation_count = 0;
if (RAWINPUT_MissingXInputSlot()) {
Uint8 correlation_id = 0;
Uint8 slot_idx = 0;
if (RAWINPUT_GuessXInputSlot(&match_state_xinput, &correlation_id, &slot_idx)) {
/* we match exactly one XInput device */
/* Probably can do without xinput_correlation_count, just check and clear xinput_slot to ANY, unless
we need even more frames to be sure */
if (ctx->xinput_correlation_count && ctx->xinput_slot == slot_idx) {
/* was correlated previously, and still the same device */
if (ctx->xinput_correlation_id + 1 == correlation_id) {
/* no one else was correlated in the meantime */
new_correlation_count = ctx->xinput_correlation_count + 1;
if (new_correlation_count == 2) {
/* correlation stayed steady and uncontested across multiple frames, guaranteed match */
ctx->xinput_correlated = SDL_TRUE;
#ifdef DEBUG_JOYSTICK
SDL_Log("Correlated joystick %d to XInput device #%d\n", joystick->instance_id, slot_idx);
#endif
correlated = SDL_TRUE;
RAWINPUT_MarkXInputSlotUsed(ctx->xinput_slot);
/* If the generalized Guide button was using us, it doesn't need to anymore */
if (guide_button_candidate.joystick == joystick)
guide_button_candidate.joystick = NULL;
if (guide_button_candidate.last_joystick == joystick)
guide_button_candidate.last_joystick = NULL;
}
} else {
/* someone else also possibly correlated to this device, start over */
new_correlation_count = 1;
}
} else {
/* new possible correlation */
new_correlation_count = 1;
ctx->xinput_slot = slot_idx;
}
ctx->xinput_correlation_id = correlation_id;
} else {
/* Match multiple XInput devices, or none (possibly due to no buttons pressed) */
}
}
ctx->xinput_correlation_count = new_correlation_count;
} else {
correlated = SDL_TRUE;
}
}
#endif /* SDL_JOYSTICK_RAWINPUT_XINPUT */
/* Poll for trigger data once (not per-state-packet) */
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
/* Prefer XInput over WindowsGamingInput, it continues to provide data in the background */
if (!has_trigger_data && ctx->xinput_enabled && ctx->xinput_correlated) {
RAWINPUT_UpdateXInput();
if (xinput_state[ctx->xinput_slot].connected) {
if (ctx->guide_hack) {
SDL_PrivateJoystickButton(joystick, guide_button, (xinput_state[ctx->xinput_slot].state.Gamepad.wButtons & XINPUT_GAMEPAD_GUIDE) ? SDL_PRESSED : SDL_RELEASED);
}
if (ctx->trigger_hack) {
SDL_PrivateJoystickAxis(joystick, left_trigger, ((int)xinput_state[ctx->xinput_slot].state.Gamepad.bLeftTrigger * 257) - 32768);
SDL_PrivateJoystickAxis(joystick, right_trigger, ((int)xinput_state[ctx->xinput_slot].state.Gamepad.bRightTrigger * 257) - 32768);
}
has_trigger_data = SDL_TRUE;
}
}
#endif /* SDL_JOYSTICK_RAWINPUT_XINPUT */
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
if (!has_trigger_data && ctx->wgi_correlated) {
RAWINPUT_UpdateWindowsGamingInput(); /* May detect disconnect / cause uncorrelation */
if (ctx->wgi_correlated) { /* Still connected */
struct __x_ABI_CWindows_CGaming_CInput_CGamepadReading *state = &ctx->wgi_slot->state;
if (ctx->guide_hack) {
SDL_PrivateJoystickButton(joystick, guide_button, (state->Buttons & GamepadButtons_GUIDE) ? SDL_PRESSED : SDL_RELEASED);
}
if (ctx->trigger_hack) {
SDL_PrivateJoystickAxis(joystick, left_trigger, ((int)(state->LeftTrigger * SDL_MAX_UINT16)) - 32768);
SDL_PrivateJoystickAxis(joystick, right_trigger, ((int)(state->RightTrigger * SDL_MAX_UINT16)) - 32768);
}
has_trigger_data = SDL_TRUE;
}
}
#endif /* SDL_JOYSTICK_RAWINPUT_GAMING_INPUT */
if (!correlated) {
if (!guide_button_candidate.joystick ||
(ctx->last_state_packet && (
!guide_button_candidate.last_state_packet ||
SDL_TICKS_PASSED(ctx->last_state_packet, guide_button_candidate.last_state_packet)
))
) {
guide_button_candidate.joystick = joystick;
guide_button_candidate.last_state_packet = ctx->last_state_packet;
}
}
#endif /* SDL_JOYSTICK_RAWINPUT_MATCHING */
}
static void
RAWINPUT_JoystickUpdate(SDL_Joystick *joystick)
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
{
/* Ensure data messages have been pumped */
RAWINPUT_PumpMessages();
RAWINPUT_UpdateOtherAPIs(joystick);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
static void
RAWINPUT_JoystickClose(SDL_Joystick *joystick)
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
{
RAWINPUT_DeviceContext *ctx = joystick->hwdata;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
#ifdef SDL_JOYSTICK_RAWINPUT_MATCHING
if (guide_button_candidate.joystick == joystick)
guide_button_candidate.joystick = NULL;
if (guide_button_candidate.last_joystick == joystick)
guide_button_candidate.last_joystick = NULL;
#endif
if (ctx) {
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
SDL_RAWINPUT_Device *device;
#ifdef SDL_JOYSTICK_RAWINPUT_XINPUT
xinput_device_change = SDL_TRUE;
if (ctx->xinput_enabled) {
if (ctx->xinput_correlated) {
RAWINPUT_MarkXInputSlotFree(ctx->xinput_slot);
}
WIN_UnloadXInputDLL();
}
#endif
#ifdef SDL_JOYSTICK_RAWINPUT_GAMING_INPUT
RAWINPUT_QuitWindowsGamingInput(ctx);
#endif
device = ctx->device;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
if (device) {
SDL_assert(device->joystick == joystick);
device->joystick = NULL;
RAWINPUT_ReleaseDevice(device);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
SDL_free(ctx->data);
SDL_free(ctx->button_indices);
SDL_free(ctx->axis_indices);
SDL_free(ctx->hat_indices);
SDL_free(ctx);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
joystick->hwdata = NULL;
}
}
LRESULT RAWINPUT_WindowProc(HWND hWnd, UINT msg, WPARAM wParam, LPARAM lParam)
{
if (!SDL_RAWINPUT_inited)
return -1;
switch (msg)
{
case WM_INPUT_DEVICE_CHANGE:
{
HANDLE hDevice = (HANDLE)lParam;
switch (wParam) {
case GIDC_ARRIVAL:
RAWINPUT_AddDevice(hDevice);
break;
case GIDC_REMOVAL: {
SDL_RAWINPUT_Device *device;
device = RAWINPUT_DeviceFromHandle(hDevice);
if (device) {
RAWINPUT_DelDevice(device, SDL_TRUE);
}
} break;
default:
return 0;
}
}
return 0;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
case WM_INPUT:
{
Uint8 data[sizeof(RAWINPUTHEADER) + sizeof(RAWHID) + USB_PACKET_LENGTH];
UINT buffer_size = SDL_arraysize(data);
if ((int)GetRawInputData((HRAWINPUT)lParam, RID_INPUT, data, &buffer_size, sizeof(RAWINPUTHEADER)) > 0) {
PRAWINPUT raw_input = (PRAWINPUT)data;
SDL_RAWINPUT_Device *device = RAWINPUT_DeviceFromHandle(raw_input->header.hDevice);
if (device) {
SDL_Joystick *joystick = device->joystick;
if (joystick) {
RAWINPUT_HandleStatePacket(joystick, raw_input->data.hid.bRawData, raw_input->data.hid.dwSizeHid);
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
}
}
}
}
return 0;
}
return -1;
}
static void
RAWINPUT_JoystickQuit(void)
{
int ii;
RAWINPUTDEVICE rid[SDL_arraysize(subscribed_devices)];
if (!SDL_RAWINPUT_inited)
return;
for (ii = 0; ii < SDL_arraysize(subscribed_devices); ii++) {
rid[ii].usUsagePage = USB_USAGEPAGE_GENERIC_DESKTOP;
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
rid[ii].usUsage = subscribed_devices[ii];
rid[ii].dwFlags = RIDEV_REMOVE;
rid[ii].hwndTarget = NULL;
}
if (!RegisterRawInputDevices(rid, SDL_arraysize(rid), sizeof(RAWINPUTDEVICE))) {
SDL_Log("Couldn't un-register RAWINPUT");
}
while (SDL_RAWINPUT_devices) {
RAWINPUT_DelDevice(SDL_RAWINPUT_devices, SDL_FALSE);
}
WIN_UnloadHIDDLL();
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
SDL_RAWINPUT_numjoysticks = 0;
SDL_RAWINPUT_inited = SDL_FALSE;
}
static SDL_bool
RAWINPUT_JoystickGetGamepadMapping(int device_index, SDL_GamepadMapping *out)
{
return SDL_FALSE;
}
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
SDL_JoystickDriver SDL_RAWINPUT_JoystickDriver =
{
RAWINPUT_JoystickInit,
RAWINPUT_JoystickGetCount,
RAWINPUT_JoystickDetect,
RAWINPUT_JoystickGetDeviceName,
RAWINPUT_JoystickGetDevicePlayerIndex,
RAWINPUT_JoystickSetDevicePlayerIndex,
RAWINPUT_JoystickGetDeviceGUID,
RAWINPUT_JoystickGetDeviceInstanceID,
RAWINPUT_JoystickOpen,
RAWINPUT_JoystickRumble,
RAWINPUT_JoystickRumbleTriggers,
RAWINPUT_JoystickHasLED,
RAWINPUT_JoystickSetLED,
RAWINPUT_JoystickSetSensorsEnabled,
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
RAWINPUT_JoystickUpdate,
RAWINPUT_JoystickClose,
RAWINPUT_JoystickQuit,
RAWINPUT_JoystickGetGamepadMapping
Fixed bug 4477 - Support more than 4 XInput-capable devices on Windows Jimb Esser Add new RawInput controller API, and improved correlation with XInput/WGI Reorder joystick init so drivers can ask the others if they handle a device reliably Do not poll disconnected XInput devices (major perf issue) Fix various cases where incorrect correlation could happen Simple mechanism for propagating unhandled Guide button presses even before guaranteed correlation Correlate by axis motion as well as button presses Fix failing to zero other trigger Fix SDL_HINT_JOYSTICK_HIDAPI not working if set before calling SDL_Init() Add missing device to device names Disable RawInput if we have a mismatch of XInput-capable but not RawInput-capable devices Updated to SDL 2.0.13 code with the following notes: New HID driver: xbox360w - no idea what that is, hopefully urelated SDL_hidapijoystick.c had been refactored to couple data handling logic with device opening logic and device lists caused some problems, yields slightly uglier integration than previously when the 360 HID device driver was just handling the data. SDL_hidapijoystick.c now often pulls the device off of the joystick_hwdata structure for some rumble logic, but it appears that code path is never reached, so probably not a problem. Looks like joystick_hwdata was refactored to not include a mutex in other drivers, maintainers may want to do the same refactor here if that's useful for some reason. Something changed in how devices get names, so getting generic names. Had to fix a (new?) bug where removing an XInput controller caused existing controllers (that moved to a new XInput index) to get identified as 0x045e/0x02fd ("it's probably Bluetooth" in code), rendering the existing HIDAPI_IsDevicePresent and new RAWINPUT_IsDevicePresent unreliable.
2020-03-16 20:25:02 +01:00
};
#endif /* SDL_JOYSTICK_RAWINPUT */
/* vi: set ts=4 sw=4 expandtab: */