I have no idea if this works (or if it ever worked, having now examined this
code), as I have no way to compile or test this.
If it's broken, send patches. :)
This means it doesn't have to block while the current frame finishes using the
vertex buffer; it just moves on to the next, probably-not-in-use buffer.
- high-level filters out duplicate render commands from the queue so
backends don't have to.
- Setting draw color is now a render command, so backends can put color
information into the vertex buffer to upload with everything else instead
of setting it with slower dynamic data later.
- backends can request that they always batch, even for legacy programs,
since the lowlevel API can deal with it (Metal, and eventually Vulkan
and such...)
- high-level makes sure the queue has at least one setdrawcolor and
setviewport command before any draw calls, so the backends don't ever have
to manage cases where this hasn't been explicitly set yet.
- backends allocating vertex buffer space can specify alignment, and the
high-level will keep track of gaps in the buffer between the last used
positions and the aligned data that can be used for later allocations
(Metal and such need to specify some constant data on 256 byte boundaries,
but we don't want to waste all that space we had to skip to meet alignment
requirements).
It now uses a growable linked list that keeps a pool of allocated items for
reuse, and reallocs the vertex array as necessary. Testsprite2 can scale to
20,000 (or more!) draws now without drama.
This moves all the rendering to a command list that is flushed to the GL as
necessary, making most common activities upload a single vertex buffer per
frame and dramatically reducing state changes. In pathological cases,
like Emscripten running on iOS's Safari, performance can go from a dozen
draw calls killing your performance to 1000 draw calls running smoothly.
This is work in progress, and not ready to ship. Among other things, it has
a hardcoded array that isn't checked for overflow. But the basic idea is
sound!
src/core/linux/SDL_evdev.c:104:1: warning: "_THIS" redefined
In file included from src/core/linux/../../events/SDL_events_c.h:26,
from src/core/linux/SDL_evdev.c:45:
src/core/linux/../../events/../video/SDL_sysvideo.h:146:1: warning: this is the location of the previous definition
Alexandre
DirectFB supports 32-bit ABGR pixel format via DSPF_ABGR, but SDL doesn't map SDL_PIXELFORMAT_ABGR8888 to DSPF_ABGR.
A patch is attached and should add support for ABGR pixel format devices.
The only place angle is activated and causes effect is RenderCopyEx. All other
methods which use vertex shader, leave angle disabled and cause useless sin/cos
calculation in shader.
To get around shader's interface is changed to a vector that contains results
of sin and cos. To behave properly when disabled, cos value is set with offset
-1.0 making 0.0 default when deactivated.
As nice side effect it simplifies GLES2_UpdateVertexBuffer: All attributes are
vectors now.
Additional background:
* On RaspberryPi it gives a performace win for operations. Tested with
[1] numbers go down for 5-10% (not easy to estimate due to huge variation).
* SDL_RenderCopyEx was tested with [2]
* It works around left rotated display caused by low accuracy sin implemetation
in RaspberryPi/VC4 [3]
[1] https://github.com/schnitzeltony/sdl2box
[2] https://github.com/schnitzeltony/sdl2rendercopyex
[3] https://github.com/anholt/mesa/issues/110
Signed-off-by: Andreas M?ller <schnitzeltony@gmail.com>
This prevents us from clearing the clip rect globally when another application has set it.
There's also an experimental change to regularly update the clip rect for a window defensively, in case someone else has reset it. It works well, but I don't know if it's cheap enough to call as frequently as it would be called now, and might have other undesirable side effects.
Also fixed whitespace and SDL coding style
Polling without wl_display_flush() never responds to ping requests.
In that case ping-pong works only on other events, such as user input
or on frame swapped.
From https://git.merproject.org/mer-core/libsdl/merge_requests/3
Original author: Alexander Akulich <a.akulich@omprussia.ru>
Otherwise if SDL_POWER_DISABLED is disabled (eg with --disable-power):
... with clang -pedantic:
src/power/SDL_power.c:48:50: warning: use of GNU empty initializer extension [-Wgnu-empty-initializer]
static SDL_GetPowerInfo_Impl implementations[] = {
^
src/power/SDL_power.c:48:50: warning: zero size arrays are an extension [-Wzero-length-array]
2 warnings generated.
... with gcc -pedantic:
src/power/SDL_power.c:48:50: warning: ISO C forbids empty initializer braces [-Wpedantic]
src/power/SDL_power.c:48:50: warning: ISO C forbids empty initializer braces [-Wpedantic]
static SDL_GetPowerInfo_Impl implementations[] = {
^
src/power/SDL_power.c:48:30: error: zero or negative size array ?implementations?
static SDL_GetPowerInfo_Impl implementations[] = {
^~~~~~~~~~~~~~~
... with Watcom:
./src/power/SDL_power.c(85): Error! E1112: Initializer list cannot be empty
Creating a full-screen SDL renderer on Windows will keep the screensaver
suspended by DirectX, as is default for full-screen DX applications. However,
for applications that render in windowed-mode, the screensaver will
still kick in, even if SDL_DisableScreenSaver() is called or
SDL_HINT_VIDEO_ALLOW_SCREENSAVER is set to 0 (default). Implementing
a SuspendScreenSaver() function for Win32 fixes this behavior.
This means that if you have two devices named "Soundblaster Pro" in your
machine, one will be reported as "Soundblaster Pro" and the other as
"Soundblaster Pro (2)".
This makes it so you can't into a position where one of your devices can't
be opened because another is sitting on the same name.
duckgrease
SDL_RenderCopyEx blits wrong image (in some cases it's bunch of alternating horizontal lines, some cases it's image from the wrong coordinate, and in some cases it's just a bunch of garbled pixels), when the following conditions are met:
- Use software renderer.
- Enable either horizontal or vertical flip.
- source and destination rectangles must have same width and height, and must be smaller than the size of the texture.
- source rectangle's X and Y coordinates must be 0.
Icenowy Zheng
One front buffer is locked in GLES_SetupCrtc() and overrides the next_bo just locked in KMSDRM_GLES_SwapWindow, then the next_bo gets lost and is not released even when quitting the video.
It may leads to problems with GLES drivers that doesn't clean up GBM correctly if there's any bo left (e.g. the Mali Utgard r6p2 blob). In the case of Mali Utgard r6p2 blob, the DRM device file is still hold by the blob, and if you try to SDL_Quit to let another program to run (this is done by EmulationStation), the new program will fail to open DRM device.
Sylvain
I did various benches. with clang 6.0.0 on linux, and ndk-r16b on android (NDK_TOOLCHAIN_VERSION=clang).
- still see a x10 speed factor.
- with duff_loops, it does not use vectorisation (but doesn't seem to be a problem).
on linux my patch is already at full speed on -O2, whereas the duff_loops need -O3 (200 ms at -03, and 300ms at -02).
I realized that on Android, I had a slight variation which fits best.
both on linux with -O2 and -O3, and on android with 02/03 and armeabi-v7a/arm64.
Here's the patch.
suffix to the temp Pixel local in the DUFFS_LOOP.
SDL_blit.h (ASSEMBLE_RGB): add _ prefix to temp Pixel locals to avoid
any possible shadowings.
The warnings were like the following:
In file included from src/video/SDL_blit_N.c:26:0:
src/video/SDL_blit_N.c: In function 'BlitNtoNKeyCopyAlpha':
src/video/SDL_blit_N.c:2421:24: warning: declaration of 'Pixel' shadows a previous local [-Wshadow]
Uint32 Pixel = ((*src32 & rgbmask) == ckey) ? *dst32 : *src32;
^
src/video/SDL_blit.h:475:21: note: in definition of macro 'DUFFS_LOOP8'
case 0: do { pixel_copy_increment; /* fallthrough */ \
^
src/video/SDL_blit_N.c:2419:13: note: in expansion of macro 'DUFFS_LOOP'
DUFFS_LOOP(
^
src/video/SDL_blit_N.c:2399:12: warning: shadowed declaration is here [-Wshadow]
Uint32 Pixel;
^
Author: Anthony Pesch <inolen@gmail.com>
Date: Fri May 4 20:21:21 2018 -0400
Added SDL_AUDIO_ALLOW_SAMPLES_CHANGE flag enabling users of SDL_OpenAudioDevice to get
the sample size of the actual hardware buffer vs having a stream created to handle the
delta
The code is now reliant on SDL_PrivateJoystickAdded() and SDL_PrivateJoystickRemoved() being called correctly when devices are added or removed on Windows
"In release 2.0.6, when Linux evdev keyboard support has been moved to a
separate source file, a feature was added to disable normal keyboard event
processing to prevent "spilling" keystrokes to background virtual console.
This feature has one unpleasant side effect: if application fails to call
`SDL_Exit` before termination or crashes with fatal signal, console is left
in unusable state with keyboard not working and no possibility to switch
virtual console. If user has a chance, he can login remotely and restore
keyboard with `kbd_mode`, otherwise the only option is to reboot the machine.
This patch fixes that problem by intercepting fatal signals (with `sigaction`)
and process termination (with `atexit`), to restore keyboard state, if it
wasn't properly restored with `SDL_Exit`.
The function registered with `atexit` also restores original signal handlers,
to prevent leaving invalid handlers after SDL library is unloaded, if it was
loaded dynamically with `dlopen`.
No signal handlers or `atexit` function are installed if SDL boolean hint
`SDL_HINT_NO_SIGNAL_HANDLERS` is `SDL_TRUE`.
Additionally, if environment variable `SDL_INPUT_LINUX_KEEP_KBD` exists,
keyboard initialization function completely skips disabling keyboard. This
can be useful for debugging."
Fixes Bugzilla #4193.
This would cause problems in various ways, but specifically triggers an
assert when you close a WASAPI capture device in an app running over RDP.
Related to (but not the actual bug) in Bugzilla #3924.
SDL_UDEV_Scan must be called during SDL_SYS_HapticInit to ensure devices
outside of the 0-31 range are added to the list of haptic devices.
Fixes Bugzilla #3923.
First: disable d'n'd events by default; most apps don't need these at all, and
if an app doesn't explicitly handle these, each drop on the window will cause
a memory leak if the events are enabled. This follows the guidelines we have
for SDL_TEXTINPUT events already.
Second: when events are enabled or disabled, signal the video layer, as it
might be able to inform the OS, causing UI changes or optimizations (for
example, dropping a file icon on a Cocoa app that isn't accepting drops will
cause macOS to show a rejection animation instead of the drop operation just
vanishing into the ether, X11 might show a different cursor when dragging
onto an accepting window, etc).
Third: fill in the drop event details in the test library and enable the
events in testwm.c for making sure this all works as expected.
Parse out a copy of the varargs ourselves to get to the reply portion, since
the original passed to D-Bus might modify or not modify the caller's copy,
depending on system ABI.
At the HG state abdd17144682, 64-bit assemblies are using SSE2-based resampler, produces junk sound when converting the S32 -> Float32 -> S16 chain. The `NEED_SCALAR_CONVERTER_FALLBACKS` thing works perfectly.
If I will find a reason that caused this mistake, I'll send a patch by myself.
This is just in parity with the existing zxdg-shell-unstable-v6 code. Making
the Wayland target robust (and uh, with title bars) is going to take a lot
of work on top of this.
it used to place zeroes between the sign and the number. (space-padding
from within SDL_PrintString() seems OK: spaces are added before sign.)
also fixed the maxlen handling if the number has a sign.
Sylvain
Patch a few warnings when using:
-Wmissing-prototypes -Wdocumentation -Wdocumentation-unknown-command
They are automatically enabled with -Wall
- remove force-enabling of pad_zeroes for %u for compatibility
(was added in https://hg.libsdl.org/SDL/rev/701f4a25df89)
- ignore pad_zeroes for %s and %S
- ignore pad_zeroes for %d, %i and %u if a precision is given
Alexei
On WM_WINDOWPOSCHANGED event, WIN_UpdateClipCursor() is called. SDL_WINDOW_INPUT_FOCUS is set even when the mouse pointer is not inside the SDL window and therefore ClipCursor(&rect) is called. When dragging the window and rect.bottom=800 (i.e. the bottom edge of the screen) the SDL window is clipped to the bottom of the screen and it is not possible to move it back to the center of the screen.
Anthony @ POW Games
SDL_CreateTextureFromSurface makes an internal call to SDL_GetColorKey which can return an error and spams the error log with "Surface doesn't have a colorkey" even though the original function didn't return an error.
src/joystick/hidapi/SDL_hidapijoystick.c: In function 'HIDAPI_InitializeDiscovery':
src/joystick/hidapi/SDL_hidapijoystick.c:281: error: 'true' undeclared (first use in this function)
src/joystick/hidapi/SDL_hidapijoystick.c:281: error: (Each undeclared identifier is reported only once
src/joystick/hidapi/SDL_hidapijoystick.c:281: error: for each function it appears in.)
src/joystick/hidapi/SDL_hidapijoystick.c: In function 'HIDAPI_UpdateDiscovery':
src/joystick/hidapi/SDL_hidapijoystick.c:339: error: 'true' undeclared (first use in this function)
src/joystick/hidapi/SDL_hidapijoystick.c:341: error: ISO C90 forbids mixed declarations and code
The previous code attempted to use set_buffer_size / set_period_size
discretely, favoring the parameters which generated a buffer size that was
exactly 2x the requested buffer size. This solution ultimately prioritizes
only the buffer size, which comes at a large performance cost on some machines
where this results in an excessive number of periods. In my case, for a 4096
sample buffer, this configured the device to use 37 periods with a period size
of 221 samples and a buffer size of 8192 samples. With 37 periods, the SDL
Audio thread was consuming 25% of the CPU.
This code has been refactored to use set_period_size and set_buffer_size
together. set_period_size is called first to attempt to set the period to
exactly match the requested buffer size, and set_buffer_size is called second
to further refine the parameters to attempt to use only 2 periods. The
fundamental change here is that the period size / count won't go to extreme
values if the buffer size can't be exactly matched, the buffer size should
instead just increase to the next closest multiple of the target period size
that is supported. After changing this, for a 4096 sample buffer, the device
is configured to use 3 periods with a period size of 4096 samples and a buffer
size of 12288 samples. With only 3 periods, the SDL Audio thread doesn't even
show up when profiling.
Fixes Bugzilla #4156.
"Applications (such as SDL's testgesture) do "event.tfinger.x * window_width"
to find window coord. Currently the X11 XInput2 backend expects application
to do "event.tfinger.x * (window_width-1)" instead.
X11 XInput2 touch events are normalized so x is 1.0 at "width - 1" but other
SDL backends appear to have x be 1.0 at "width". Same issue for touch event
y with regards to height."
Fixes Bugzilla #4183.
If we change the current context behind the app's back, those tracking
the current context to minimize context changes are going to get
confused.
This brings the EGL backend in line with the GLX one.
Fixes Bugzilla #4199.
This was reproducible by running an SDL app on the console from an ssh login. In this case the terminal wasn't owned by the user running the app, so we were using the default keymap, which didn't have state transitions defined for ctrl and alt, so once we entered that state keypresses would no longer transition out of that state, nor would they generate text.
As a workaround, we'll just reset to the default shift state if that happens, which means we'll get text for keys pressed while ctrl is held down, but I don't think that's a big problem.
Note that in this case we also can't mute the keyboard, so the keypresses go to the console, which probably isn't what you want...
SDL_ExitProcess(), SDL_AbortAssertion() and SDLTest_BailOut().
(Commit 303c1e0fb0cf for bug #4100 removed SDL_NORETURN from
SDL_ExitProcess() and SDL_AbortAssertion() in order to avoid
warnings from windows builds, but that's temporary I guess..)
These are entirely untested
Several USB ids refer to multiple packaged products. In those cases I tried to use the most common name, or a general name (e.g. PS3 Controller), or a completely generic name (e.g. USB gamepad) if it wasn't clear what type of controller it was.
Patches welcome!
Expand SDLActivity::SDLSurface::surfaceChanged() callback to grab the panel width and height at the same time and pass that along to the native code. Only works on API 17+. Duplicates surface dimensions whenever it fails.
Add Android_DeviceWidth/Android_DeviceHeight globals to native code.
Disambiguate Android_ScreenWidth/Android_ScreenHeight -> Android_SurfaceWidth/Android_SurfaceHeight
Use device width/height for all display mode settings.
This means we have to consider SDL_WINDOW_MINIMIZED a window creation flag, but on non-windows platforms we just remove it and let the normal FinishWindowCreation re-apply and do the minimize as I have no idea what is right on them or if anything should change.
CR: Phil
Martin ?irokov
Launching an SDL application with SDL_AUDIODRIVER=jack, and then calling SDL_OpenAudioDevice() with whatever parameters fails with an error like this one:
SDL_OpenAudioDevice: Couldn't connect JACK ports: SDL:sdl_jack_output_0 => system:midi_playback_1
This happens because JACK_OpenDevice in src/audio/jack/SDL_jackaudio.c blindly tries to connect to all input ports without checking whether they are for audio or midi.
The fix is to check port types and ignore all non audio ports. Also I removed devports field from struct SDL_PrivateAudioData, because it's never really used and removing unused ports from it would be PITA.
Jona
The following explains why this bug was happening:
This crash was caused because the audio session was being set as active [session setActive:YES error:&err] when the audio device was actually being CLOSED. Certain cases the audio session being set to active would fail and the method would return right away. Because of the way the error was handled we never removed the SDLInterruptionListener thus leaking it. Later when an interruption was received the THIS_ object would contain a pointer to an already released device causing the crash.
The fix:
When only one device remained open and it was being closed we needed to set the audio session as NOT active and completely ignore the returned error to successfully release the SDLInterruptionListener. I think the user assumed that the open_playback_devices and open_capture_devices would equal 0 when all of them where closed but the truth is that at the end of the closing process that the open devices count is decremented.
(It gets upset at the -2147483648, thinking this should be an unsigned value
because 2147483648 is too large for an int32, so the negative sign upsets the
compiler.)