src/joystick/windows/SDL_rawinputjoystick.c: In function 'RAWINPUT_HandleStatePacket':
src/joystick/windows/SDL_rawinputjoystick.c:1343:9: warning: suggest explicit braces to avoid ambiguous 'else'
multiply gyro values by sensitivity
When the hardware calibration fails, values read from sensors need to be multiplied by default sensitivity (16 for gyro, 1 for accelerometer).
Simon McVittie
When watching for hotplug events we can poll the inotify fd, but we
still need to scan /dev/input once per process, otherwise we'll fail
to detect devices that were already connected.
Flatpak[1] and pressure-vessel[2] are known to use user namespaces,
therefore udev event notification via netlink won't work reliably.
Both frameworks provide a filesystem API that libraries can use to
detect them. Do that, and automatically fall back from udev-based
device discovery to the inotify-based fallback introduced in Bug #5337.
[1] <https://flatpak.org/>
[2] <https://gitlab.steamos.cloud/steamrt/steam-runtime-tools/-/tree/master/pressure-vessel>
Signed-off-by: Simon McVittie <smcv@collabora.com>
Alex S
Evdev headers aren't actually included in the base system (well, it has a private copy), they are available through the devel/evdev-proto port instead. We also have devel/libinotify and devel/libudev-devd shims, I didn't verify whether they work with SDL.
This fixes bad report parsing for various newer Xbox controllers, and this driver is now preferred over XInput, since it handles more than 4 controllers.
C.W. Betts
This patch adds support to the GameController framework on macOS Big Sur and later, adding support for MFi controllers as well as rumble support for PS4 and Xbox One. There is some code to make sure that the IOKit joystick handler doesn't include two controllers at once.
While the GameController framework is present in earlier versions of macOS, there was no public, approved way of checking if a specific IOHIDDevice is a controller that GameController could handle. This was changed in Big Sur.
Added support for the PS4 controller gyro and accelerometer on iOS and HIDAPI drivers
Also fixed an issue with the accelerometer on iOS having inverted axes
Spooky
For some reason the Logitech Extreme 3D joystick was added to SDL_gamecontrollerdb.h in the linux section only.
This breaks the joystick in linux as it is not a gamepad. I am unable to correctly use or map the Logitech Exteme 3D joystick in games that use SDL2 in linux.
Please remove Logitech Extreme 3D from SDL_gamecontrollerdb.h Linux section. It is a joystick not a gamepad.
Bart van der Werf
When directinput fails to load, but a controlller is plugged in, an access violation happens.
This is due to IEventHandler_CRawGameControllerVtbl_InvokeAdded calling SDL_DINPUT_JoystickPresent which does not check if dinput is assigned signalling initialization of directinput.
Joel Linn
This fixes two types of MSVC compiler warnings.
- One parameter in the function signatures of two WGI event handlers had one level of indirection too much (and did not match Windows SDK headers). The indirection was cast away so it still worked.
- size_t was implicitly cast to UINT32 for a number of (constant) string lengths.
This improves SDL's ability to detect joystick hotplug in a container
environment.
We cannot reliably receive events from udev in a container, because they
are delivered as netlink events, which are authenticated by their uid
being 0. However, in a user namespace created by an unprivileged user
(for example bubblewrap, as used by Flatpak and Steam's
pressure-vessel-wrap), the kernel does not allow us to map uid 0, and
the netlink events appear to be from the kernel's overflowuid (typically
65534/nobody), meaning libudev cannot distinguish between genuine uevents
from udevd and an attack by a malicious local user.
Signed-off-by: Simon McVittie <smcv@collabora.com>
Previously we only checked for at least one button or key and at least
the X and Y absolute axes, but this has both false positives and false
negatives.
Graphics tablets, trackpads and touchscreens all have buttons and
absolute X and Y axes, but we don't want to detect those as joysticks.
On normal Linux systems ordinary users do not have access to these
device nodes, but members of the 'input' group do.
Conversely, some game controllers only have digital buttons and no
analogue axes (the Nintendo Wiimote is an example), and some have axes
and no buttons (steering wheels or flight simulator rudders might not
have buttons).
Use the more elaborate heuristic factored out from SDL's udev code path
to handle these cases.
In an ideal world we could use exactly the same heuristic as udev's
input_id builtin, but that isn't under a suitable license for inclusion
in SDL, so we have to use a parallel implementation of something
vaguely similar.
Signed-off-by: Simon McVittie <smcv@collabora.com>
This works on capability bitfields that can either come from udev or
from ioctls, so it is equally applicable to both udev and non-udev
input device detection.
Signed-off-by: Simon McVittie <smcv@collabora.com>
Device enumeration via libudev can fail in a container for two reasons:
* the netlink protocol between udevd and libudev is considered private,
so there is no API guarantee that the version of libudev in a container
will understand netlink messages from a dissimilar version of udevd
on the host system;
* the netlink protocol between udevd and libudev relies for security on
being able to check the uid of each message, but in a container with
a user namespace where host uid 0 is not mapped, the libudev client
cannot distinguish between messages from host uid 0 and messages from
a different, malicious user on the host
To make this easier to experiment with, always compile the fallback
code path even if libudev is disabled. libudev remains the default if
enabled at compile time, but the fallback code path can be forced.
Signed-off-by: Simon McVittie <smcv@collabora.com>
Xbox Elite controllers use AUX1-AUX4 to represent the paddle buttons when using the HIDAPI driver
PS4 and PS5 controllers use AUX1 to represent the touchpad button
Nintendo Switch Pro controllers use AUX1 to represent the capture button
Anthony Pesch
I was looking into my own input bug and noticed an issue in the HIDAPI code while looking over it. I don't have a controller that goes down this path to test and try to provoke the issue, but it looks pretty straight forward.
The memmove to shift the joystick id array on disconnect isn't scaling the size by sizeof(SDL_JoystickID), likely corrupting the ids on disconnect.
Jan Bujak
I wrote a new driver for my gamepad on Linux. I'd like SDL to support it out-of-box, as currently it just treats it as a generic joystick instead of a gamepad. From what I can see the only way to do that is to either 1) pick one of the already supported controllers' PID, VID and button layouts and have my driver send that (effectively lying that it's something else), or 2) submit a preconfigured, hardcoded mapping to SDL.
Both of those, in my opinion, are silly when we already have the Linux Gamepad Specification which standarizes this:
https://www.kernel.org/doc/html/v4.15/input/gamepad.html
Unfortunately SDL doesn't make use of it currently. So I've took it upon myself to add it; patch is in the attachments.
Basically what the patch does is that if SDL finds no built-it controller mappings for a given joystick it then asks the joystick backend to autodetect it, and that uses the relevant evdev bits to figure out which button/axis is which. (See the specs for more details.)
With this patch applied my own driver for my controller works out-of-box with SDL with no extra configuration and is correctly recognized as a gamepad; this is also going to be the case for any other driver which follows the Linux Gamepad Specification.
* 8BitDo N30 Pro 2
* 8BitDo SN30 Gamepad
* 8BitDo SN30 Pro+
* 8BitDo Zero 2
* SZMY-POWER PC Gamepad
* ThrustMaster eSwap PRO Controller
* ZEROPLUS P4 Wired Gamepad
In additional, all 8BitDo controllers use SDL_HINT_GAMECONTROLLER_USE_BUTTON_LABELS to have the correct mapping based on user preferences.
The joystick layer can't necessarily give us perfect centering, but we know
that the game controller level has logical absolute idle positions that have
nothing to do with the physical device.
So send game controller events to make it look like the device is completely
untouched before sending the final removal event.
This driver supports the Razer Atrox Arcade Stick
Some of the quirks of this driver, inherent in Windows Gaming Input:
* There will never appear to be controllers connected at startup. You must support hot-plugging in order to see these controllers.
* You can't read the state of the guide button
* You can't get controller events in the background
meyraud705
On line 220 of SDL_hidapi_xbox360.c https://hg.libsdl.org/SDL/file/4608f0e6e8e3/src/joystick/hidapi/SDL_hidapi_xbox360.c#l220
if (!XINPUTGETSTATE(user_index, &xinput_state[user_index].state) == ERROR_SUCCESS) {
logical not is only applied to the left hand side of this comparison.
I think you mean:
if (XINPUTGETSTATE(user_index, &xinput_state[user_index].state) != ERROR_SUCCESS) {
bluenaxela+sdl
I've noticed that the Switch Pro Controller hidapi driver does not report battery levels when connected via Bluetooth, despite having code for setting joystick->epowerlevel.
This is caused by the driver always using k_eSwitchInputReportIDs_SimpleControllerState via Bluetooth. Using that mode means that the state reports you get back from the controller do not include battery state. Not using the full controller state over Bluetooth effectively makes this driver's support for setting joystick->epowerlevel entirely pointless, only ever reporting SDL_JOYSTICK_POWER_WIRED.
Is there a reason this was set to only use SimpleControllerState via Bluetooth?
I've attached a patch I'm using to allow getting battery level for the Switch Pro Controller.
A couple notes about this patch:
1) It changes LoadStickCalibration to accept the input_mode that is selected, because that's really what should determine what is used for stick extents, since stick extents differ between the modes.
2) In my patch I only use FullControllerState when the vid/pid matches the official Switch Pro Controller, as a cautionary measure in case some third-party controllers have problems with FullControllerState mode via Bluetooth (I noticed a HORI Wireless Switch Pad I had seemed to not read controller calibration correctly for stick extents. Maybe it's calibration data was uninitialized on account of having never been used with a Switch? I'm unsure, though if that guess is right maybe SDL2 should be detecting an uninitiated calibration state and using some sensible defaults)
bluenaxela+sdl
The HORI Wireless Switch Pad does not properly connect via bluetooth. I did some debugging and found that the code that tries to control the Home LED causes this controller to disconnect.
This is a multi-part fix, and is the 2nd attempt at a fix for Bug 5034. Here
are the problems being addressed:
1. On macOS 10.14.x and earlier, trying to call IOHIDDeviceUnscheduleFromRunLoop
without a prior, paired call to IOHIDDeviceScheduleWithRunLoop, appears to
lead to a crash. A per-device flag has been added to make sure that these
calls are paired.
2. DARWIN_JoystickDetect was free'ing its SDL_joystick's hwdata field
(via FreeDevice) without setting it to NULL, and DARWIN_JoystickRumble wasn't
checking for a NULL hwdata. FreeDevice will now set hwdata to NULL and
DARWIN_JoystickRumble will check for a NULL hwdata.
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.
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.
Malte Kie?ling
I get a build error in SDL_sysjoystick.c:74 for the merged patch, but its nothing to sweat about, just -Werror=declaration-after-statement doing its usual stuff.
David Ludwig
I have created a new driver for SDL's Joystick and Game-Controller subsystem: a Virtual driver. This driver allows one to create a software-based joystick, which to SDL applications will look and react like a real joystick, but whose state can be set programmatically. A primary use case for this is to help enable developers to add touch-screen joysticks to their apps.
The driver comes with a set of new, public APIs, with functions to attach and detach joysticks, set virtual-joystick state, and to determine if a joystick is a virtual-one.
Use of virtual joysticks goes as such:
1. Attach one or more virtual joysticks by calling SDL_JoystickAttachVirtual. If successful, this returns the virtual-device's joystick-index.
2. Open the virtual joysticks (using indicies returned by SDL_JoystickAttachVirtual).
3. Call any of the SDL_JoystickSetVirtual* functions when joystick-state changes. Please note that virtual-joystick state will only get applied on the next call to SDL_JoystickUpdate, or when pumping or polling for SDL events (via SDL_PumpEvents or SDL_PollEvent).
Here is a listing of the new, public APIs, at present and subject to change:
------------------------------------------------------------
/**
* Attaches a new virtual joystick.
* Returns the joystick's device index, or -1 if an error occurred.
*/
extern DECLSPEC int SDLCALL SDL_JoystickAttachVirtual(SDL_JoystickType type, int naxes, int nballs, int nbuttons, int nhats);
/**
* Detaches a virtual joystick
* Returns 0 on success, or -1 if an error occurred.
*/
extern DECLSPEC int SDLCALL SDL_JoystickDetachVirtual(int device_index);
/**
* Indicates whether or not a virtual-joystick is at a given device index.
*/
extern DECLSPEC SDL_bool SDLCALL SDL_JoystickIsVirtual(int device_index);
/**
* Set values on an opened, virtual-joystick's controls.
* Returns 0 on success, -1 on error.
*/
extern DECLSPEC int SDLCALL SDL_JoystickSetVirtualAxis(SDL_Joystick * joystick, int axis, Sint16 value);
extern DECLSPEC int SDLCALL SDL_JoystickSetVirtualBall(SDL_Joystick * joystick, int ball, Sint16 xrel, Sint16 yrel);
extern DECLSPEC int SDLCALL SDL_JoystickSetVirtualButton(SDL_Joystick * joystick, int button, Uint8 value);
extern DECLSPEC int SDLCALL SDL_JoystickSetVirtualHat(SDL_Joystick * joystick, int hat, Uint8 value);
------------------------------------------------------------
Miscellaneous notes on the initial patch, which are also subject to change:
1. no test code is present in SDL, yet. This should, perhaps, change. Initial development was done with an ImGui-based app, which potentially is too thick for use in SDL-official. If tests are to be added, what kind of tests? Automated? Graphical?
2. virtual game controllers can be created by calling SDL_JoystickAttachVirtual with a joystick-type of SDL_JOYSTICK_TYPE_GAME_CONTROLLER, with naxes (num axes) set to SDL_CONTROLLER_AXIS_MAX, and with nbuttons (num buttons) set to SDL_CONTROLLER_BUTTON_MAX. When updating their state, values of type SDL_GameControllerAxis or SDL_GameControllerButton can be casted to an int and used for the control-index (in calls to SDL_JoystickSetVirtual* functions).
3. virtual joysticks' guids are mostly all-zeros with the exception of the last two bytes, the first of which is a 'v', to indicate that the guid is a virtual one, and the second of which is a SDL_JoystickType that has been converted into a Uint8.
4. virtual joysticks are ONLY turned into virtual game-controllers if and when their joystick-type is set to SDL_JOYSTICK_TYPE_GAMECONTROLLER. This is controlled by having SDL's default list of game-controllers have a single entry for a virtual game controller (of guid, "00000000000000000000000000007601", which is subject to the guid-encoding described above).
5. regarding having to call SDL_JoystickUpdate, either directly or indirectly via SDL_PumpEvents or SDL_PollEvents, before new virtual-joystick state becomes active (as specified via SDL_JoystickSetVirtual* function-calls), this was done to match behavior found in SDL's other joystick drivers, almost all of which will only update SDL-state during SDL_JoystickUpdate.
6. the initial patch is based off of SDL 2.0.12
7. the virtual joystick subsystem is disabled by default. It should be possible to enable it by building with SDL_JOYSTICK_VIRTUAL=1
Questions, comments, suggestions, or bug reports very welcome!
Ethan Lee
Basically replicating the solution of the Switch Controller's button label issue. Physical layout should take priority unless it's explicitly requested by the user or application!
Prior to this fix, we would hit the existing_instance >= 0 case and move the joystick
again to a different index than the one requested by the caller. It also breaks the assumption
that a SDL_JoystickID is only present in SDL_joystick_players at one location.
The 8BitDo SF30 Pro Gamepad will generate a single motor pulse for each rumble message, so we need to do this frequently to have continous rumble on this device.
Jake Breen
When I run SDL_INIT with SDL_INIT_JOYSTICK it stalls for about 10 seconds (last report was 10,615ms), but only if I'm currently playing audio. (Like in Spotify for example.)
querying something related to device access (last dll loaded)
'BabbysFirst64.exe' (Win32): Loaded 'C:\Windows\SysWOW64\deviceaccess.dll'.
I use a USB DAC because my mobo's audio out is pretty not great. And I've noticed unplugging it seems to solve the issue. I haven't noticed any other issues that are caused by my DAC.
My DAC is the Sound BlasterX G1 https://us.creative.com/p/gaming-headsets/sound-blasterx-g1
Vid = 041E
PID = 3249
My system specs:
- Windows 10 Pro
- Ryzen 2700x
- 16GB Ram
- Nvidia 2070 RTX
Additional USB devices plugged in:
- Valve Index
- Xbox One Elite Controller
rofferom
I have an annoying issue on MacOS about XBoxOne Bluetooth rumble (Vendor: 0x045e, Product: 0x02fd).
When 360controller is installed, rumble is working correctly. However, Bluetooth rumble isn't working at all, with or without 360controller installed (although it is working with Chrome + https://html5gamepad.com).
I looked at the code, and it seems that XBox controllers are managed in MacOS in this file: SDL_hidapi_xbox360.c. The XBoxOne file is disabled for MacOS in SDL_hidjoystick_c.h.
The function HIDAPI_DriverXbox360_Rumble() is called correctly, and hid_write() returns no error.
I have tried a stupid test. I took the rumble packet from 360controller: ec4e88eb2d/XBOBTFF/FFDriver.cpp (L620). With the patch I have attached, I manage to have rumble working on Bluetooth (with some stupid vibration level, but it proves it can if the packet is changed).
But it breaks the USB rumble with 360controller. A comment in the function makes an explicit reference to 360controller, I think that's why I have broken this specific usecase.
I don't know what is the correct way to fix this, but it seems that the current implementation has a missing case for Bluetooth support.
Note that I also tested master this morning, and I have another issue:
if (!device->ffservice) {
return SDL_Unsupported();
}
test fails in DARWIN_JoystickRumble(). This test has been done quickly, I'm not totaly confident about its accuracy.
meyraud705
In SDL_hidapi_switch.c
line 443: Function BTrySetupUSB call WriteProprietary with pBuf=NULL and ucLen=0
line 376: WriteProprietary check its input (!pBuf && ucLen > 0) || ucLen > sizeof(packet.rgucProprietaryData): ucLen is 0 so it passes
line 382: WriteProprietary call memcpy with pBuf=NULL
meyraud705
Dualshock4 on bluetooth need 78 bytes for the rumble data while SDL_HIDAPI_RumbleRequest can only hold 64 bytes.
'volatile' is not meant for thread synchronization.
The list of rumble request could grow infinitely if user call SDL_JoystickRumble too much. The documentation says "Each call to this function cancels any previous rumble effect", so overwriting pending request seem like a good idea.
This fixes a crash whereby SDL could crash on macOS/Darwin, if and when a
USB game controller gets unplugged. SDL was not retaining a reference
to the controller's OS/IOKit-provided 'device object', and was capable
of trying to use it, after a device was hot-unplugged.
There is now a thread that handles all HIDAPI rumble requests and a lock that guarantees that we're not reading and writing the device at the same time.
The index and indices were swapped; Which is fine as long as there are
no gaps in the ABS_HAT* event availability but otherwise things do get confused.
Signed-off-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
This sequence works across Microsoft, PowerA, PDP, and HORI controllers.
The newer Microsoft XBox firmware requires synchronizing the rumble sequence number, when SDL sees it after the initial connect
The Razer Wildcat controller requires waiting for init responses before continuing the initialization sequence.
The PDP Battlefield 1 controller takes over a second to be ready for initialization, and if initialization is attempted before then, it will fail.
LinGao
We build SDL with Visual studio 2017 compiler on Windows Server 2016, but it failed to build due to error LNK2019: unresolved external symbol memset referenced in function SDL_SetJoystickIDForPlayerIndex with MSVC x64 on Windows on latest default branch. And we found that it can be first reproduced on 0fff06175109 changeset. Could you please help have a look about this issue? Thanks in advance!
Steps to Reproduce:
1.hg clone https://hg.libsdl.org/SDL D:\SDL\src
2.Open a VS 2017 x64 command prompt as admin and browse to D:\SDL
3.msbuild /p:Configuration=Release /p:Platform=x64 /p:WindowsTargetPlatformVersion=10.0.17134.0 VisualC\SDL.sln /t:Rebuild
Actual result:
Creating library D:\SDL\src\VisualC\x64\Release\SDL2.lib and object D:\SDL\src\VisualC\x64\Release\SDL2.exp
SDL_joystick.obj : error LNK2019: unresolved external symbol memset referenced in function SDL_SetJoystickIDForPlayerIndex [D:\SDL\src\VisualC\SDL\SDL.vcxproj]
D:\SDL\src\VisualC\x64\Release\SDL2.dll : fatal error LNK1120: 1 unresolved externals [D:\SDL\src\VisualC\SDL\SDL.vcxproj]
Done Building Project "D:\SDL\src\VisualC\SDL\SDL.vcxproj" (Rebuild target(s)) -- FAILED.
Using Wii U GameCube USB adapter with multiple controllers attached and
restarting SDL input in a game results in extra joysticks with NULL name.
HIDAPI_CleanupDeviceDriver() shut down joysticks by iterating through
device->num_joysticks but each HIDAPI_JoystickDisconnected() decreases
device->num_joysticks and shifts joysticks array down. Resulting in only
half of controllers being shutdown. It worked with only 1 controller
attached though.
Disconnect HIDAPI device joystick 0 until there are none left.
Added the functions SDL_JoystickFromPlayerIndex(), SDL_JoystickSetPlayerIndex(), SDL_GameControllerFromPlayerIndex(), and SDL_GameControllerSetPlayerIndex()
meyraud705
On a Dualshock 4 controller using hidapi driver, calling SDL_JoystickRumble with a duration too long (SDL_HAPTIC_INFINITY for example) causes the rumble to stop immediately.
This happens because of integer overflow on line 301 of SDL_hidapi_ps4.c
(https://hg.libsdl.org/SDL/file/a3077169ad23/src/joystick/hidapi/SDL_hidapi_ps4.c#l301), which sets expiration time in the past.
When we initialize the controller it has an internal rumble sequence number, and if our rumble sequence number doesn't match that, rumble won't happen. To fix that we cycle through the range of sequence numbers, and at some point we'll match up with the controller's sequence number and it'll roll forward until it matches our next rumble sequence number.
Calling open() on input devices can generate device I/O which blocks
the main thread and causes dropped frames. Using stat() we can avoid
opening anything unless /dev/input has changed since we last polled.
We could have used something fancy like inotify, but it didn't seem
worth the added complexity for this uncommon non-udev case.
