fix building with Mesa 19.2
With Mesa 19.2 building fails with:
/include/GLES/gl.h:63:25: error: conflicting types for 'GLsizeiptr'
The same type is defined in include/SDL_opengl.h for OpenGL and the two
headers should not be included at the same time.
This was just never noticed because the same header guard '__gl_h_' was
used. This was changed in Mesa. The result is this error.
Fix this the same way GLES2 already handles this: Don't include the GLES
header when the OpenGL header was already included.
(https://hg.libsdl.org/SDL/rev/6a3670d6108d)
The Offscreen video driver is intended to be used for headless rendering
as well as allows for multiple GPUs to be used for headless rendering
Currently only supports EGL (OpenGL / ES) or Framebuffers
Adds a hint to specifiy which EGL device to use: SDL_HINT_EGL_DEVICE
Adds testoffscreen.c which can be used to test the backend out
Disabled by default for now
The LE transformation for vec_perm has an implicit assumption that the
permutation is being used to reorder vector elements (in this case 4-byte
integer word elements), not to reorder bytes within those elements. Although
this is legal behavior, it is not anticipated by the transformation performed
by the compilers.
This causes pygame-1.9.1 test failure on PPC64LE because blitted pixmaps are
corrupted there due to how SDL uses vec_perm().
From RedHat / Fedora: https://bugzilla.redhat.com/show_bug.cgi?id=1392465
Original patch was provided by: Menanteau Guy <menantea@linux.vnet.ibm.com>
If KMSDRM_drmModeSetCrtc is called when the swap interval is
set to 0, the driver behaves as though vertical sync is engaged by
limiting framerate to the refresh rate, but performance is much worse
than with vertical sync enabled.
Resolve this issue by ensuring that the Crtc is only set up once,
and KMSDRM_drmModePageFlip is called, albeit without any followup
queueing or waiting for flips.
Ozkan Sezer
As for the issue: This bmp reports bpp=0, therefore SDL_CalculatePitch()
returns pitch==0, which is then fed to SDL_malloc() (which is malloc())
and malloc(0) returns _something_ which is not NULL but not someting
that we expect.. Then testsprite.c:LoadSprite() accesses the pixels
as *(Uint8*)pixels which valrind reports as:
==15533== Invalid read of size 1
==15533== at 0x8048C08: LoadSprite (testsprite.c:45)
==15533== by 0x80492FC: main (testsprite.c:224)
==15533== Address 0x449e588 is 0 bytes after a block of size 0 alloc'd
==15533== at 0x40072B2: malloc (vg_replace_malloc.c:270)
==15533== by 0x4045719: SDL_CreateRGBSurface (SDL_surface.c:126)
==15533== by 0x40403C1: SDL_LoadBMP_RW (SDL_bmp.c:237)
==15533== by 0x8048BB2: LoadSprite (testsprite.c:36)
==15533== by 0x80492FC: main (testsprite.c:224)
Besides, valrind also reports this:
==15533== Conditional jump or move depends on uninitialised value(s)
==15533== at 0x40403F3: SDL_LoadBMP_RW (SDL_bmp.c:247)
==15533== by 0x8048BB2: LoadSprite (testsprite.c:36)
==15533== by 0x80492FC: main (testsprite.c:224)
Easy/quick solution would be early-rejecting a bmp with 0 bpp from SDL_bmp.c:SDL_LoadBMP_RW()
Caleb Cornett
SDL_ShowMessageBox on UIKit doesn't do anything special with buttons that are marked with the flag SDL_MESSAGEBOX_BUTTON_RETURNKEY_DEFAULT. According to Apple's documentation on UIAlertController, a button can respond to a return key if it's marked as the preferredAction of the controller. SDL doesn't set a preferredAction currently, so I've attached a patch to fix that.
M Stoeckl
To reproduce:
1. Run any SDL-based program with a Wayland compositor, orphaning it so that it doesn't have an immediate parent process. (For example, from a terminal, running `supertux2 & disown`.) The program should use the wayland backend, i.e. by setting environment variable SDL_VIDEODRIVER=wayland.
2. Kill the compositor process.
Results:
- The SDL program will keep running.
Expected results:
- The SDL program should close. (What close should mean here, I'm not sure - is injecting an SDL_Quit the appropriate action when a video driver disconnects?)
Build data:
2019-06-22, hg tip (12901:bf8d9d29cbf1), Linux, can reproduce with sway, weston, and other Wayland oompositors.
superfury
I notice that, somehow, when locking the mouse into place(using SDL_SetRelativeMouseMode), somehow at least the movement information gets through to both mouse movement and touch movement events?
My app handles both, so when moving a touched finger accross the app(using RDP from an Android device) I see the mouse moving inside the app when it shouldn't(meaning that the touch movement is ignored properly by the app(press-location dependant) but the mouse movement is still performed due to the mouse movement events)?
With multitouch, register to receive XI_Motion (which desctivates MotionNotify),
so that we can distinguish real mouse motions from synthetic one.
(bug 4690)
Windows generates fake raw mouse events for touchscreens for compatibility
with legacy apps that predate touch support in Windows. We already handle
touch events explicitly, so drop the synthetic events to avoid duplicates.
This time, we make anything we think is a MacBook trackpad report its touches
as SDL_MOUSE_TOUCHID, even though they're not _actually_ synthesized events,
and let all mouse input--even if the OS synthesized it from a multitouch
trackpad on our behalf--look like physical input. This is backwards from
reality, but produces the results most apps will expect.
Note that if you have a real touch device that doesn't appear to be the
trackpad, it'll produce real touch events with unique device ids, so it's
not a total loss here, but also note that the way we decide if it was the
trackpad is an imperfect heuristic; it happens to work out right now, but
it's not impossible that a real touchscreen could come to the Mac at some
point and (incorrectly?) call it a "mouse" input, etc.
But for now, good enough.
Fixes Bugzilla #4690.
"Starting with changeset 12433, the mouse cursor is not displayed on the
Raspberry Pi platform, due to a bug in the handling of the new
"global_cursor" in RPI_ShowCursor(). Currently, if cursor == global_cursor,
the function immediately returns 0. The function should not return here.
Instead, if cursor == global_cursor, it shouldn't try to hide the current
cursor and update global_cursor = cursor. However, it *should* still continue
through the rest of the function."
Fixes Bugzilla #4699.
Using IOKit for this pops up a warning at startup on macOS 10.15 ("Catalina"),
asking the user to authorize the app to listen to all keyboard input in the
system, which is unacceptable.
I _think_ we were using IOKit under incorrect presumptions here; the Stack
Overflow link mentioned in it was complaining about not being able to use
flagsChanged to differentiate between left and right mod keys, but that's not
an issue for capslock.
It's also possible this code was trying to deal with capslock changing when
the window didn't have focus, but we handle this elsewhere now, if we didn't
at the time.
GetWindowText() wants you to tell it the size of the buffer--including the
terminating NULL char--but we weren't counting that last char, losing the
last char of the string in the process. This was only seen with the special
case of SDL_CreateWindowFrom() to use an existing native window, not
the usual SDL_CreateWindow() codepath.
Fixes Bugzilla #4696.
Jan Martin Mikkelsen
Patch to scan /dev/dri based on names rather than file type
Loading KMS/DRM on FreeBSD fails because the "available" code in the driver checks for character device nodes under /dev/dri and the /dev/dri/card* files are symlinks rather than device nodes nodes on FreeBSD. The symlink points to /dev/drm/0.
The attached patch counts /dev/dri/card* entries rather than directory entries which are character devices.
It was originally added to work around an input event problem in the code of a specific app which mixed SDL and native UIViews, but that app solved its problems in a better manner since then.
This was to deal with broken vsync support in macOS 10.14, which we assumed
would remain broken indefinitely, but a later 10.14 released fixed it.
This is a loss of late-swap support, but there are several subtle problems
in our CVDiplayLink code that are also evaporating, to be fair.
Fixes Bugzilla #4575.
(Backed out changeset 8760fed23001)
Dzmitry Malyshau
Current code, search paths, and error messages are written to only consider MoltenVK on macOS as a Vulkan Portability implementation. It's not the only implementation available to the users. gfx-portability [1] has been shown to run a number of titles well, including Dota2, Dolphin Emulator, and vkQuake3, often out-performing MoltenVK in frame rate and stability (see Dolphin benchmark [2]).
There is no reason for SDL to be that specific, it's not using any MVK-specific functions other than the WSI initialization ("VK_MVK_macos_surface"). gfx-portability exposes this extension as well, and a more generic WSI extension is in process. It would be good if SDL was written in a more generic way that expect a Vulkan Portability library as opposed to MoltenVK specifically.
[1] https://github.com/gfx-rs/portability
[2] https://gfx-rs.github.io/2019/03/22/dolphin-macos-performance.html
Caleb Cornett
On iOS 12, creating a window with the SDL_WINDOW_FULLSCREEN flag does not dim the home indicator or defer system gestures. The same goes for setting the SDL_HINT_IOS_HIDE_HOME_INDICATOR to "2" -- it has no effect at all.
I've tracked down the source of this misbehavior to a timing issue. The initial `setNeedsUpdate...` calls were happening too early and getting applied to the launch screen by mistake. In the attached patch, I've added a call to those functions right after the launch screen is hidden so that they apply to the main view controller instead. This appears to fix the issue, at least on my iPhone 6s Plus.
Sylvain
On Android, if you set no attribute using SDL_GL_SetAttribute(), and try to create a SDL Render OpenGLES 1:
- it loads first by default GLESv2 libraries
- creates the rendere OpenGLES 1
- recreates the Window to have a context 1.1 ( https://hg.libsdl.org/SDL/file/4db4cfd59470/src/render/opengles/SDL_render_gles.c#l298 )
But it doesn't unload libraries, then reload GLESv1 lib. So the SDL_Renderer OpenGLES 1 is working with GLES 2 libs, which seems inconsistent.
If you, at first, set
SDL_GL_SetAttribute(SDL_GL_CONTEXT_PROFILE_MASK, SDL_GL_CONTEXT_PROFILE_ES);
SDL_GL_SetAttribute(SDL_GL_CONTEXT_MAJOR_VERSION, 1);
SDL_GL_SetAttribute(SDL_GL_CONTEXT_MINOR_VERSION, 1);
It will correctly load GLES v1 libraries.
Here's a small patch to reload egl libs when SDL_RecreateWindow() is called.
It fixes the issue, also the case from bug 4042
( SDL_RecreateWindow() is used by SDL_Renderer gl, gles1, gles2. )
bplu4t2f
When num lock is on, the scancode reported for numpad 5 is SDL_SCANCODE_KP_5, which is correct. However, when num lock is off, windows reports the VK_CLEAR virtual key code, which is incorrectly translated into SDL_SCANCODE_CLEAR inside of the VKeytoScancode(WPARAM vkey) function.
Closing the window is asynchronous, but we free the window data immediately,
so we can get an updateLayer callback before the window is really destroyed which
will cause us to access the freed memory.
Clearing the content view will cause it to be immediately released, so no further
updateLayer callbacks will occur.
Removed incorrect call to SDL_SendWindowEvent(window, SDL_WINDOWEVENT_MOVED, x, y);
If the position of the window isn't adjusted in the SetWindowPosition() call, then sending the window event would have no effect because x and y equals the window x and y. If the position of the window is adjusted in the SetWindowPosition() call, then we don't want to clobber it with values that the user passed in.
Andrei Drexler
For X11 GenericEvents, the associated data is only available between a call to XGetEventData and the matching XFreeEventData, i.e. in X11_HandleGenericEvent. Trying to call XGetEventData a second time on the same event will fail, so an application that wants to inspect XInput2 events (e.g. for stylus pressure) has no way of retrieving its data from queued SYSWM events.
The attached patch (based on SDL-2.0.7-11629) sends SYSWM messages from X11_HandleGenericEvent while the data is still available, allowing client code to register an event filter/watcher and process the event inside the callback.
Vivante drivers use the VK_KHR_display extension for rendering directly
to the frame buffer. This patch adds support to the video driver for
Vulkan rendering using that method.
- Add an utility function SDL_Vulkan_Display_CreateSurface that creates
a surface using this extension. The display to use (if there are
multiple) can be overridden using the environment variable
"SDL_VULKAN_DISPLAY".
- Use this function in a new compilation unit SDL_vivantevideo.c,
which implements the SDL_VIDEO_VULKAN methods of the driver structure.
Dan Ginsburg
I've seen this on several devices including Moto Z running Android 7 and a Snapdragon 845 running Android 9.
What happens is as follows:
SDLActivity.onWindowFocusChanged(true) happens pretty early on, but it's before we've done SDL_CreateWindow and so Android_Window is 0x0 thus this message does not get sent:
JNIEXPORT void JNICALL SDL_JAVA_INTERFACE(nativeFocusChanged)(
JNIEnv *env, jclass cls, jboolean hasFocus)
{
SDL_LockMutex(Android_ActivityMutex);
if (Android_Window) {
__android_log_print(ANDROID_LOG_VERBOSE, "SDL", "nativeFocusChanged()");
SDL_SendWindowEvent(Android_Window, (hasFocus ? SDL_WINDOWEVENT_FOCUS_GAINED : SDL_WINDOWEVENT_FOCUS_LOST), 0, 0);
}
SDL_UnlockMutex(Android_ActivityMutex);
}
When the window does get created, in Android_CreateWindow it does this:
window->flags &= ~SDL_WINDOW_RESIZABLE; /* window is NEVER resizeable */
window->flags &= ~SDL_WINDOW_HIDDEN;
window->flags |= SDL_WINDOW_SHOWN; /* only one window on Android */
window->flags |= SDL_WINDOW_INPUT_FOCUS; /* always has input focus */
/* One window, it always has focus */
SDL_SetMouseFocus(window);
SDL_SetKeyboardFocus(window);
The SDL_SetKeyboardFocus does send an SDL_WINDOWEVENT_FOCUS_GAINED message, but it gets eaten in SDL_SendWindowEvent because we've forced SDL_WINDOW_INPUT_FOCUS beforehand:
case SDL_WINDOWEVENT_FOCUS_GAINED:
if (window->flags & SDL_WINDOW_INPUT_FOCUS) {
return 0;
}
window->flags |= SDL_WINDOW_INPUT_FOCUS;
SDL_OnWindowFocusGained(window);
break;
I can fix the problem if I comment out this line from Android_CreateWindow:
window->flags |= SDL_WINDOW_INPUT_FOCUS; /* always has input focus */
I would propose that as a fix unless there is a reason not to.
SDL_GetWindowDisplayMode was returning an incorrect result on iPhone Plus devices (tested on iOS 12.1/12.2). The problem was that the value returned by UIScreenMode was assumed to be the physical pixels on the display, rather than the scaled retina pixels. The fix is to use the scale returned by UIScreen.scale rather than the nativeScale.
java layer runs as if separate mouse and touch was 1,
Use SDL_HINT_MOUSE_TOUCH_EVENTS and SDL_HINT_TOUCH_MOUSE_EVENTS
for generating synthetic touch/mouse events
Petr Pisar
The root cause is that the POC BMP file declares 3 colors used and 4 bpp palette, but pixel at line 28 and column 1 (counted from 0) has color number 3. Then when the image loaded into a surface is passed to SDL_DisplayFormat(), in order to convert it to a video format, a used bliting function looks up a color number 3 in a 3-element long color bliting map. (The map obviously has the same number entries as the surface format has colors.)
Proper fix should refuse broken BMP images that have a pixel with a color index higher than declared number of "used" colors. Possibly more advanced fix could try to relocate the out-of-range color index into a vacant index (if such exists).
This lets apps see and choose between both an HDMI and DSI-connected display,
such as a television and the Pi Foundation's official touchscreen. It only
exposes the second display if the hardware reports that it is connected.
Petr Pisar
The reproducer has these data in BITMAPINFOHEADER:
biSize = 40
biBitCount = 8
biClrUsed = 131075
SDL_LoadBMP_RW() function passes biBitCount as a color depth to SDL_CreateRGBSurface(), thus 256-color pallete is allocated. But then biClrUsed colors are read from a file and stored into the palette. SDL_LoadBMP_RW should report an error if biClrUsed is greater than 2^biBitCount.
Surfaces are allocated using SDL_SIMDAlloc()
They are marked with SDL_SIMD_ALIGNED flag to appropriatly free them with SDL_SIMDFree()
(Flag is cleared when pixels is free'd in RLE, in case user would hijack the pixels ptr)
When providing its own memory pointer (SDL_CreateRGBSurfaceFrom()) and clearing
SDL_PREALLOC to delegate to SDL the memory free, it's the responsability of the user
to add SDL_SIMD_ALIGNED or not, whether the pointer has been allocated with SDL_malloc() or
SDL_SIMDAlloc().
Some blit combination are not supported (eg ARGB8888 -> SDL_PIXELFORMAT_INDEX1MSB)
So prevent SDL_ConvertSurface from creating a broken surface, which cannot be blitted
It was calling glClear without a context. The issue it was trying to
solve was actually that after destroying a window and creating a new one
, the contents of the old window were preserved. This no longer happens
since we resize the window to nothing on destroy.
If the browser left fullscreen mode by the user pressing ESC, the next
call to SDL_SetWindowFullscreen(1) will fail as it thinks the window is
already fullscreen. (#65)
by checking Android_Window validity
- SDLThread: user application is exiting:
SDL_VideoQuit() and clearing SDL_GetVideoDevice()
- ActivityThread is changing orientation/size
surfaceChanged() > Android_SetScreenResolution() > SDL_GetVideoDevice()
- Separate function into Android_SetScreenResolution() and Android_SendResize(),
formating, and mark Android_DeviceWidth/Heigh as static
- If you call onPause() before CreateWindow(), SDLThread will run in infinite loop in background.
- If you call onPause() between a DestroyWindow() and a new CreateWindow(), semaphores are invalids.
SDLActivity.java: the first resume() starts the SDLThread, don't call
nativeResume() as it would post ResumeSem. And the first pause would
automatically be resumed.
First error could happen if Android_SetWindowFullscreen somehow gets
called between SurfaceDestroyed() and SurfaceCreated()
Second error should not happen has native_window validity is guaranteed.
(It would happens previously with error -19)
It accesses data->native_window, which can be changed by onNativeSurfacedChanged().
Currently, Android_SetWindowFullscreen() may access data->native_window after it
has been released, and before a new reference is acquired.
(can be reproduced by adding some SDL_Delay() in onNativeSurfacedChanged and
Android_SetWindowFullscreen() ).
It may happen to have the sequence pause()/resume()/pause(), before polling
any events.
Before, it ends in 'resumed' state because as the check is greedy.
Now, always increase the Pause semaphore, and stop at each pause.
It ends in 'paused' state.
Related to bug 3250: set up a reconfiguration of SurfaceView holder.
Turn the screen off manually before the app starts
(repro rate is not 100%..)
"Pause" transition will add events:
SDL_WINDOWEVENT_ENTER
SDL_WINDOWEVENT_FOCUS_LOST
SDL_WINDOWEVENT_MINIMIZED
SDL_APP_WILL ENTER BACKGROUND
SDL_APP_DID ENTER BACKGROUND
"Resume" transition will add events:
SDL_APP_WILL ENTER FOREGROUND
SDL_APP_DID ENTER FOREGROUND
SDL_WINDOWEVENT_FOCUS_GAINED
SDL_WINDOWEVENT_RESTORED
If Android application doesn't empty the event loop in between, it enters in
"paused" state when SDL_WINDOWEVENT_RESTORED is fetched.
See bug 3250 for pratical case.
Occurs when application goes to background:
- Java activity is destroying SurfaceView holder and "egl_surface" (in onNativeSurfaceDestroyed())
- While native thread is in Android_GLES_SwapWindow(), prepared to call SDL_EGL_SwapBuffers()
The error is "call to eglSwapBuffers failed, reporting an error of EGL_BAD_SURFACE"
It an be reproduced easily by adding a SDL_Delay(100) at the begining of SDL_EGL_SwapBuffers(),
and putting the application into background.
Use a semaphore to prevent concurrency issues between Java Activity and Native thread code, when using 'Android_Window'.
(Eg. Java sending Touch events, while native code is destroying the main SDL_Window. )
Not only does this fix macOS 10.14 ("Mojave")'s broken NSOpenGLCPSwapInterval
support, it also lets us implement "adaptive vsync" on macOS!
CVDisplayLink is supported back to macOS 10.4 ("Tiger"), so we just use it
universally without version checks and dump NSOpenGLCPSwapInterval, Mojave or
not.
Switching between renderers "software -> opengl -> opengles2 -> software" fails.
"opengl -> opengles2" calls SDL_RecreateWindow() and frees "window->surface"
without marking it as "surface_invalid".
foo.null
I'm on macOS 10.14 and I think I'm using or around SDL 2.0.9. This is about the menu bar that SDL sets up which looks like:
<App Name> <Window> <View>
1. View menu never proceeds after the Window menu in any Mac application (it is always before).
2. For SDL, the only purpose of the View menu is for a single fullscreen menu item, which is not justifiable enough to reserve space for a menu. The View menu should thus be removed, and the full screen menu item should be added at the end inside of Window's menu. See built in apps like Dictionary, Chess, App Store (on 10.14) that do this.
3. SDL should add a "Close" menu item to the Window's submenu, and it should be the first item. Its key equivalent should map to command w. Without this, you cannot close the game window via this shortcut, and you cannot close the app's About window via this shortcut.
4. Apps typically use "Enter Full Screen" or "Exit Full Screen" depending on context, not "Toggle Full Screen" which is less user friendly -- I personally care about this point the least.
This happens if you try to lock the pointer and (caps & WL_SEAT_CAPABILITY_POINTER) is false
Leading to input->pointer being NULL which ends up bringing the wayland client down (at lease on weston)
