`Wayland_CreateAnimatedCursor` calls `SDL_qsort` on an array of
type `**SDL_Surface`, not on an array of `*SDL_Surface`. As such the void
pointers in the callback need to be casted to `SDL_Surface **` not
`SDL_Surface *`.
This was likely not caught since it's very unlikely that reading a few
bytes past the end of this array results in reading unreadable memory,
so the only side effect was invalid sort results, which is a bit subtle.
I caught this because I build SDL with UBSAN enabled in my debug builds,
and it trapped when the multiplication of the garbage `SDL_Surface*`
width and heights overflowed. I validated that the change is correct by
adding logs (that have then since been removed) demonstrating that it
was previously comparing garbage, and is now comparing the actual cursor
surfaces.
These functions already return immediately for NULL objects in normal builds, but it was done through CHECK_PARAM. When building with SDL_DISABLE_INVALID_PARAMS, these checks were compiled out, and NULL resources could reach backend release functions and cause segfault.
Always return on NULL so API behaves consistently regardless of SDL_DISABLE_INVALID_PARAMS.
The event coordinates returned by GetMessagePos() for WM_NCACTIVATE are out of date if the cursor moved while an overlay was active, and may indicate that the cursor is still in the window when it is not. Always use the current cursor coordinates when processing this message to avoid incorrectly setting mouse focus if the cursor is no longer within the window.
Add AndGAMER/Void Gaming USB IDs for Void GENESIS.
Recognize the Void GENESIS SInput VID/PID pair and set its device name.
Add DirectInput gamepad mappings for the wired and dongle DirectInput modes.
Allow Android embedders to use SDL without the full video/audio Java
layer by gating subsystem-specific code behind SDL_VIDEO_DISABLED and
SDL_AUDIO_DISABLED preprocessor flags.
This enables applications that only need joystick/gamepad support
(e.g. Qt-based apps like QGroundControl) to build SDL without shipping
stub Java classes for unused subsystems.
Changes:
- Split SDLActivity JNI method table into core (lifecycle, hints,
permissions) and video (surface, input, clipboard, orientation)
- Gate SDLAudioManager and SDLInputConnection JNI registration
- Make checkJNIReady() subsystem-aware: no longer requires
mAudioManagerClass when SDL_AUDIO_DISABLED
- Group method ID resolution by subsystem in nativeSetupJNI()
- Guard all video/audio function implementations and declarations
- Keep display orientation accessors always available (needed by camera)
- Add subsystem-selective SDL.setupJNI(int)/initialize(int) to SDL.java
with backwards-compatible zero-arg overloads
- Guard SDL_VIDEO_DRIVER_ANDROID and related defines in
SDL_build_config_android.h
SDL has no dummy fallback for the RTC functions, and the PS2 implementation is just an empty dummy implementation, so rename it and build it if no system-specific time functionality is found.
If wireplumber is not running or if there is no audio device,
PIPEWIRE_OpenDevice could remain stuck indefinityely on PIPEWIRE_pw_thread_loop_wait
because priv->stream_init_status is never equal to PW_READY_FLAG_ALL_PREOPEN_BITS.
Use PIPEWIRE_pw_thread_loop_timed_wait instead with a 2 seconds timeout and bail
out with an error on timeout.
A 2 seconds timeout seems plenty enough: in my observations, when there is an audio device,
the wait for the device to be ready is just a few milliseconds.
Related: 65091012b4
After the aforementioned commit, pinching in any iOS app using SDL
crashes with
*** Terminating app due to uncaught exception 'NSRangeException', reason: '-[UIPinchGestureRecognizer locationOfTouch:inView:]: index (1) beyond bounds (1).'
*** First throw call stack:
(0x19d31a23c 0x199de9224 0x19d37fec4 0x1a3f99bf0 0x1a3fa3c68 0x103277370 0x1a33b8b7c 0x1a33b848c 0x1a33b8238 0x1a2efc9a4 0x1a2ec1310 0x1a8ab8754 0x1a8af6c1c 0x1a8abc324 0x1a8ac79f4 0x1a2f010e8 0x1a2ee9550 0x1a2efd638 0x1a2eea040 0x1a2f0318c 0x1a2eecd48 0x1a2f062a8 0x1a2f03834 0x2b210f56c 0x19d2a5358 0x19d2a52cc 0x19d26a5a8 0x19d2341a0 0x19d23354c 0x1032bcdd8 0x1032ee2a0 0x1032ee3fc 0x1032ee360 0x102280644 0x10228042c 0x1033cd0cc 0x1032a8198 0x19a61f5dc 0x19d273960 0x19d273658 0x19d2731cc 0x19d234584 0x19d23354c 0x24299f498 0x1a2f2c244 0x1a2e97158 0x1032a5780 0x102280030 0x199e41c1c)
libc++abi: terminating due to uncaught exception of type NSException
As it turns out, `UIPinchGestureRecognizer` does not guarantee the
presence of two touches throughout the entire duration of the pinch.
When the user is releasing the touches that constitute the pinch
gesture, more often than not they will release one touch before the
other. When this happens, `UIPinchGestureRecognizer` *does not terminate
the pinch*, but continues updating it, only with one touch remaining in
the gesture. The gesture is only marked concluded when the user releases
*both* touches.
To prevent the crash, check that the second touch is actually available
via the `numberOfTouches` property. If it is not, avoid emitting
`SDL_EVENT_PINCH_UPDATE` events to prevent sending unexpected payloads
downstream; there is no second touch to use anymore to calculate
`(focus|span)_(x|y)`.
In some cases, conflicts can occur at the platform or driver level if a window initially configured for OpenGL is then used with a Vulkan-based renderer. Try to reconfigure or recreate the window in the unlikely event that it has the OpenGL flag set when initializing the Vulkan or GPU renderer to remove any platform or driver-specific OpenGL objects and properties.
This change also makes it so relative mode doesn't kick in until the mouse enters the window client area. This prevents relative mode from kicking in while clicking and dragging on the title bar, etc.
This adds support for:
* System theme
* Sandbox detection
* Device form factor detection
Many things aren't properly supported yet, but changes and upgrades will happen on the Ubuntu Touch side, so SDL should automatically support more Ubuntu Touch features as time goes.
This doesn't _suppress_ the warnings, it annotates our memory access pattern
so ThreadSanitizer knows that threads aren't in conflict, since TSan doesn't
know about io_uring's mechanisms.
Fixes#15083.
External window surfaces can't be destroyed and recreated, so try our best to reconfigure them when switching between GL profiles, or between GL and Vulkan.
- Addresses issue #15115
- mach_absolute_time can be misused for fingerprinting. This is bad.
- Apple docs prefer clock_gettime_nsec_np(CLOCK_UPTIME_RAW) in time.h
- Since HAVE_NANOSLEEP is defined, time.h is included
Before, MetalFence was implemented as simply a busy loop on an atomic
int on metal, meaning the cpu would busy wait on the gpu to finish
taking power from it and decreasing battery life. This was the only kind
of cpu-gpu syncing (apart from requesting a swapchain)
