Commit Graph

7294 Commits

Author SHA1 Message Date
Vladislav Dmitrievich Turbanov 31e8a4627a * Fixed closing the test window. 2021-02-11 14:20:01 -08:00
Sam Lantinga ef52560a24 Fixed bug 5539 - Clang 11 fails to compile a CMake build with conflicting types for _m_prefetchw
vladius

In SDL_cpuinfo.h it seems like <intrin.h> is not included when __clang__ is defined, as the comment in the file explicitly reads:
"Many of the intrinsics SDL uses are not implemented by clang with Visual Studio"

However, the SDL_endian.h header does include <intrin.h> without any precautions like:
>#ifdef _MSC_VER
>#include <intrin.h>
>#endif

Maybe it should be changed to something like:
>#ifdef _MSC_VER
>#ifndef __clang__
>#include <intrin.h>
>#endif
>#endif
2021-02-10 10:22:20 -05:00
Sam Lantinga 629334f283 Fixed bug 5543 - Wayland: Fix waylandvideo.h warnings
wahil1976

This patch fixes the warnings seen when compiling the Wayland backend. This will also be required in the future to avoid issues with compilation.
2021-02-10 10:22:20 -05:00
Cameron Gutman f70e197363 Fix waiting on condition variables with the SRW lock implmentation
When SleepConditionVariableSRW() releases the SRW lock internally, it causes
our SDL_mutex_srw state to become inconsistent. The lock is unowned yet inside,
the owner is still the sleeping thread and more importantly the owner count is
still 1.

The next time someone acquires the lock, they will bump the owner count from 1
to 2. At that point, the lock is hosed. From the internal lock state, it looks
to us like that owner has acquired the lock recursively, even though they have
not. When they call SDL_UnlockMutex(), it will see the owner count > 0 and not
call ReleaseSRWLockExclusive().

Now when someone calls SDL_CondSignal(), SleepConditionVariableSRW() will start
the wakeup process by attempting to re-acquire the SRW lock. This will deadlock
because the lock was never released after the other thread had used it. The
thread waiting on the condition variable will never be able to wake up, even if
the SDL_CondWaitTimeout() function is used and the timeout expires.
2021-02-10 10:22:20 -05:00
Sam Lantinga d9ba20442e Backed out changeset 852a7bdbdf4b
This causes a use-after-free memory error
2021-02-10 10:22:20 -05:00
Christian Rauch 4183211a54 free the 'display' after it was added to global list 2021-02-10 10:22:20 -05:00
Manuel Alfayate Corchete 005e2c59a7 [KMS/DRM] Bugfix number #5535: Improve reliability, by wahil1976. 2021-02-10 10:22:20 -05:00
Christian Rauch 2170c7bfde make AddressSanitizer optional and disabled by default 2021-02-10 10:22:20 -05:00
Christian Rauch 5045d299a6 enable AddressSanitizer only for GCC 5 onwards (bug #5533) 2021-02-10 10:22:20 -05:00
Christian Rauch 066960c207 add '-shared-libasan' to debug flags (bug #5533) 2021-02-10 10:22:20 -05:00
Ozkan Sezer dc45a228b9 avoid some pedantic warnings in array initializers 2021-02-10 10:22:20 -05:00
Sam Lantinga 9c3aa7f055 SDL: fix packet handling for original version of Stadia FW 2021-02-10 10:22:20 -05:00
Ozkan Sezer bb9e049d1d minor updates to libc function checks 2021-02-10 10:22:19 -05:00
Manuel Alfayate Corchete 19fa85d8af [KMS/DRM] Fix build warning. 2021-02-10 10:22:19 -05:00
Manuel Alfayate Corchete 6ee53258cd [KMS/DRM] Replace indent tabs with spaces, as intended. 2021-02-10 10:22:19 -05:00
Manuel Alfayate Corchete 59cd46c24a [KMS/DRM] Merge patch for bug #5532: No need to correct cursor position now that all windows are fullscreen. Link: https://bugzilla.libsdl.org/show_bug.cgi?id=5519. 2021-02-10 10:22:19 -05:00
Manuel Alfayate Corchete 538f7ad6b7 [KMS/DRM] Remove redundant SDL_SendWindowEvent() call. 2021-02-10 10:22:19 -05:00
Manuel Alfayate Corchete 7beba05013 [KMS/DRM] Restore all-windows-are-fullscreen functionality, since there is no window manager in KMSDRM. 2021-02-10 10:22:19 -05:00
Cameron Gutman f4d58689e0 Fix Xbox Series X controller on macOS
There were two different implementations of IsBluetoothXboxOneController(), one
in SDL_hidapi_xbox360.c and one in SDL_hidapi_xboxone.c. The latter had been
updated to include USB_PRODUCT_XBOX_ONE_SERIES_X_BLUETOOTH while the former had
not.

This mismatch led to the Xbox Series X failing on macOS only. We have special
code for handling the 360Controller driver for macOS which requires us to use
the Xbox 360 driver for wired Xbox One controllers, and the SDL_hidapi_xbox360
version of IsBluetoothXboxOneController() was used to determine which devices
were wired.

In addition to adding the missing USB_PRODUCT_XBOX_ONE_SERIES_X_BLUETOOTH, this
change moves IsBluetoothXboxOneController() into a single shared function which
will ensure this bug won't happen again.
2021-02-10 10:22:19 -05:00
Sam Lantinga 525196dd8b Updated runtime dependency on libudev 2021-02-10 10:22:19 -05:00
Sam Lantinga ff297753db Fixed compiler warning 2021-02-10 10:22:19 -05:00
Sam Lantinga a60af1e4d0 Backed out changeset 1cde3dd0f44d - this breaks windows which are created and then set to FULLSCREEN_DESKTOP 2021-02-10 10:22:19 -05:00
Sylvain Becker d74779b748 Android: documention update 2021-02-10 10:22:19 -05:00
Sam Lantinga e404b525f1 KMSDRM doesn't have a window manager, so all windows are fullscreen 2021-02-10 10:22:19 -05:00
Sylvain Becker 549bc13362 SDL_vulkan_utils: minor code clean-up 2021-02-10 10:22:19 -05:00
Sylvain Becker c0166a29b7 SDL_ConvertColorkeyToAlpha: remove and clarify a FIXME
This function doesn't handle bpp 1 or 3 case, because those formats never have an alpha channel
2021-02-10 10:22:19 -05:00
Sam Lantinga 85235985fb Fixed detecting the paddles on the Xbox Elite Series 1 controller 2021-02-10 10:22:19 -05:00
Sam Lantinga 69e9b2acee Fixed bug 5471 - Creating a fullscreen desktop window goes windowed temporarily
This is caused by the Metal renderer recreating the window because by default we create an OpenGL window on macOS.

It turns out that at least on macOS 10.15, a window that has been initialized for OpenGL can also be used with Metal. So we'll skip recreating the window in that case.
2021-02-10 10:22:19 -05:00
Sam Lantinga ef2f13e125 Added test command line options to force different window types 2021-02-10 10:22:19 -05:00
Sam Lantinga 2426949a18 Removed support for clock_gettime_nsec_np()
SDL_GetTicks() was broken and it's not adding any real value here.
2021-02-10 10:22:19 -05:00
Christian Rauch 42798cec8b cmake: enable AddressSanitizer in Debug builds if supported 2021-02-10 10:22:19 -05:00
Sam Lantinga fadfa51022 Don't uncorrelate while rumble is active and stay correlated longer in case raw input messages are lagging a bit. 2021-02-10 10:22:19 -05:00
Sebastian Krzyszkowiak e862856e6f wayland: Don't crash when the properties of already existing wl_output change 2021-02-10 10:22:18 -05:00
Sam Lantinga cef198c9cb Fixed bug 5524 - Pass NSString to NSLog()
Hiroyuki Iwatsuki

If you pass the C string directly to NSLog(), it will be garbled with Japanese and probably other language strings, or no log will be output at all.

NSLog("Hello, World!"); // => "Hello, World!"
NSLog("こんにちは、世界!"); // => No output...

Therefore, you need to convert the string to an NSString before passing it to NSLog().

NSString *str = [NSString stringWithUTF8String:"こんにちは、世界!"];
NSLog(@"%@", str); // => "こんにちは、世界!"

Thank you.
2021-02-10 10:22:18 -05:00
Sam Lantinga b7f711cd76 Don't enable address sanitize flags without checking compiler first 2021-02-10 10:22:18 -05:00
Christian Rauch 0ed16ceddb free 'outputs' in 'Wayland_DestroyWindow' 2021-02-10 10:22:18 -05:00
Christian Rauch a44678465f free the 'data_device_manager' 2021-02-10 10:22:18 -05:00
Christian Rauch f8b4a6e25a cmake: enable AddressSanitizer in Debug builds 2021-02-10 10:22:18 -05:00
Sebastian Krzyszkowiak 3ae2ec34fa wayland: Fix transform and scale handling when setting display mode 2021-02-10 10:22:18 -05:00
Manuel Alfayate Corchete 2c76433151 [KMS/DRM] Merge patch for bug 5522#: Implement KMSDRM_GetWindowWMInfo(). 2021-02-10 10:22:18 -05:00
Manuel Alfayate Corchete bfa51c3845 [KMS/DRM] Fix for bug #5518: only do async pageflips when hardware supports them. 2021-02-10 10:22:18 -05:00
Cameron Gutman 79cd8cab08 Add default handler for Alt+Tab while keyboard grab is enabled
By default, we will minimize the window when we receive Alt+Tab with a
full-screen keyboard grabbed window to allow the user to escape the
full-screen application.

Some applications like remote desktop clients may want to handle Alt+Tab
themselves, so provide an opt-out via SDL_HINT_ALLOW_ALT_TAB_WHILE_GRABBED=0.
2021-02-10 10:22:18 -05:00
Sylvain Becker 9d70711342 Revert checks on destination scaling size (see bug #5510) 2021-02-10 10:22:18 -05:00
Manuel Alfayate Corchete 088fb52579 [KMS/DRM] Correct small omission on bugfix #5513: y-coord correction has to be done on WarpMouseGlobal, too. 2021-02-10 10:22:18 -05:00
Sam Lantinga 2cc70ca39e Fixed crash if the GameCube controller hasn't been opened yet 2021-02-10 10:22:18 -05:00
Ozkan Sezer 091a4649e8 added --enable-xinput switch for windows builds 2021-02-10 10:22:18 -05:00
Sylvain Becker d6a6662fa1 Remove checks on destination scaling size (see bug #5510) 2021-02-10 10:22:18 -05:00
Manuel Alfayate Corchete 8d95aba174 [KMS/DRM] Remove unused KMSDRM_SetWindowGrab prototype in header file. 2021-02-10 10:22:18 -05:00
Manuel Alfayate Corchete b17c49509b [KMS/DRM] Patch for bug #5513. KMSDRM backend can now manage and use several displays. 2021-02-10 10:22:18 -05:00
Cameron Gutman a78bce9e38 Properly handle keys already down when the hook is installed
For keys that are already down when we install the keyboard hook, we need to
allow the WM_KEYUP/WM_SYSKEYUP message to be processed normally. This ensures
that other applications see the key up, which prevents the key from being stuck
down from the perspective of other apps when our grab is released.
2021-02-10 10:22:18 -05:00