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.)
The concern is that a massive int sample, like 0x7FFFFFFF, won't fit in a
float32, which doesn't have enough bits to hold a whole number this large,
just to divide it to get a value between 0 and 1.
Previously we would convert to double, to get more bits, do the division, and
cast back to a float, but this is expensive.
Casting to double is more accurate, but it's 2x to 3x slower. Shifting out
the least significant byte of an int32, so it'll definitely fit in a float,
and dividing by 0x7FFFFF is still accurate to about 5 decimal places, and the
difference doesn't appear to be perceptable.
SDL now builds with gcc 7.2 with the following command line options:
-Wall -pedantic-errors -Wno-deprecated-declarations -Wno-overlength-strings --std=c99
XAudio2 doesn't have capture support, so WASAPI was to replace it; the holdout
was WinRT, which still needed it as its primary audio target until the WASAPI
code code be made to work.
The support matrix now looks like:
WinXP: directsound by default, winmm as a fallback for buggy drivers.
Vista+: WASAPI (directsound and winmm as fallbacks for debugging).
WinRT: WASAPI
Simon Hug
Patch that adds [-1, 1] clamping to the scalar audio type conversions.
This may come from the SDL_Convert_F32_to_X_Scalar functions. They don't clamp the float value to [-1, 1] and when they cast it to the target integer it may be too large or too small for the type and get truncated, causing horrible noise.
The attached patch throws clamping in, but I don't know if that's the preferred way to fix this. For x86 (without SSE) the compiler (I tested MSVC) seems to throw a horrible amount of x87 code in it. It's a bit better with SSE, but probably still quite the performance hit. And SSE2 uses a branchless approach with maxss and minss.
The audioqueue thread needs to keep running, and processing the CFRunLoop
until the AudioQueue is disposed of, otherwise CoreAudio will hang waiting for
final data to feed the device.
At least, I think this is how it all works. It definitely fixes the bug here!
Since AudioQueueDispose() calls AudioQueueStop() internally, there's no need
for our thread to handle this, either, which is good because the AudioQueue
would be disposed by this point. So now the AudioQueue is disposed first, and
then our thread is joined, and everything works out okay.
Just in case, we mark the device "paused" before setting everything in motion,
so any further callbacks from CoreAudio will write silence and not fire the
app's audio callback again.
Fixes Bugzilla #3868.
(I thought padding size ranged from 5 frames to ~30 frames (based around
RESAMPLER_ZERO_CROSSINGS, which is 5), but it's actually between 512 and
several thousands (based on RESAMPLER_SAMPLES_PER_ZERO_CROSSING)). It gets
big fast when downsampling.
Previously, the padding was silence, which was a problem when streaming since
you would sample a little bit of this silence between each buffer.
We still need a means to get padding data for the right hand side, but this
patch makes the resampler output more correct.
This time it's using real math from a real whitepaper instead of my previous
amateur, fast-but-low-quality attempt. The new resampler does "bandlimited
interpolation," as described here: https://ccrma.stanford.edu/~jos/resample/
The output appears to sound cleaner, especially at high frequencies, and of
course works with non-power-of-two rate conversions.
There are some obvious optimizations to be done to this still, and there is
other fallout: this doesn't resample a buffer in-place, the 2-channels-Sint16
fast path is gone because this resampler does a _lot_ of floating point math.
There is a nasty hack to make it work with SDL_AudioCVT.
It's possible these issues are solvable, but they aren't solved as of yet.
Still, I hope this effort is slouching in the right direction.
This would cause playback problems in certain situations, such as on the
Raspberry Pi. The device that the wait was added for seems to not benefit from
it in modern times, and standard desktop Linux seems to do the right thing
when a USB device is unplugged now, without this patch.
Fixes Bugzilla #3599.
Simon Hug
This issue actually raises the question if this API change (requirement of initialized audio subsystem) is breaking backwards compatibility. I don't see the documentation saying it is needed in 2.0.5.
"Major changes, roughly in order of appearance:
- Use float math everywhere, instead of promoting to double and casting back
all the time.
- Conserve sound energy when downmixing any channel into two other channels.
- Add a QuadToStereo filter. (The previous technique of reusing StereoToMono
never worked, since it assumed an incorrect channel layout for 4.0.)
- Add a 71to51 filter. This removes just under half of the cases the previous
code would silently break in.
- Add a QuadTo51 filter. More silent breakage fixed.
- Add a 51to71 filter, removing another almost-half of the silently broken
cases.
- Add 8 to the list of values SDL_SupportedChannelCount will accept.
- Change SDL_BuildAudioCVT's channel-related logic to handle every case, and
to actually fail if it fails instead of silently corrupting sound data and/or
crashing down the road."
(Note that SDL doesn't otherwise support 7.1 audio yet, but hopefully it will
soon and the 7.1 converters are an important piece of that. --ryan.)
Fixes Bugzilla #3727.
David Ludwig
I've created a new set of patches. I am happy to create more, if it would help.
One version only copies 'size'.
A second version copies both 'size' and 'silence'. When looking over the documentation for SDL_OpenAudio in SDL_audio.h, it mentioned that both 'size' and 'silence' were things that SDL_OpenAudio would calculate.
Regarding *both* patches, I did notice that SDL 1.2 appears to have always modified desired's size and silence fields. The SDL wiki, at https://wiki.libsdl.org/SDL_OpenAudio#Remarks , does note:
manuel.montezelo
Original bug report (note that it was against 2.0.0, it might have been fixed in between): http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=733015
--------------------------------------------------------
Package: libsdl2-2.0-0
Version: 2.0.0+dfsg1-3
Severity: normal
Tags: patch
I have occasional crashes here caused by the X11 backend of SDL2. It seems to
be caused by the X11_Pending function trying to add a high number (> 1024)
file descriptor to a fd_set before doing a select on it to avoid busy waiting
on X11 events. This causes a buffer overflow because the file descriptor is
larger (or equal) than the limit FD_SETSIZE.
Attached is a possible workaround patch.
Please also keep in mind that fd_set are also used in following files which
may have similar problems.
src/audio/bsd/SDL_bsdaudio.c
src/audio/paudio/SDL_paudio.c
src/audio/qsa/SDL_qsa_audio.c
src/audio/sun/SDL_sunaudio.c
src/joystick/linux/SDL_sysjoystick.c
--------------------------------------------------------
On Tuesday 24 December 2013 00:43:13 Sven Eckelmann wrote:
> I have occasional crashes here caused by the X11 backend of SDL2. It seems
> to be caused by the X11_Pending function trying to add a high number (>
> 1024) file descriptor to a fd_set before doing a select on it to avoid busy
> waiting on X11 events. This causes a buffer overflow because the file
> descriptor is larger (or equal) than the limit FD_SETSIZE.
I personally experienced this problem while hacking on the python bindings
package for SDL2 [1] (while doing make runtest). But it easier to reproduce in
a smaller, synthetic testcase.
Simon Hug
Some code in SDL loads libraries with SDL_LoadObject to get more information or use newer APIs. SDL_LoadObject may fail, set an error message and SDL will continue with some fallback code. Since SDL will overwrite the error or exit the function with a return value that indicates success, the error form SDL_LoadObject for the optional stuff might as well be cleared right away.
kdrakehp
The attached patch adds capture support to the sndio backend.
The patch also allows the `OpenDevice' function to accept arbitrary device names.
Simon Hug
There's a chance that an audio conversion from many channels to a few can use more than 9 audio filters. SDL_AudioCVT has 10 SDL_AudioFilter pointers of which one has to be the terminating NULL pointer. The SDL code has no checks for this limit. If it overflows there can be stack or heap corruption or a call to 0xa.
Attached patch adds a function that checks for this limit and throws an error if it is reached. Also adds some documentation.
Test parameters that trigger this issue:
AUDIO_U16MSB with 224 channels at 46359 Hz
V
AUDIO_S16MSB with 6 channels at 27463 Hz
The fuzzer program I uploaded in bug 3667 has more of them.
We don't fill buffers just to throw them away during shutdown now, we let the
AudioQueue free its own buffers during disposal (which fixes possible warnings
getting printed to stderr by CoreAudio), and we stop the queue after running
any queued audio during shutdown, which prevents dropping the end of the
audio playback if you opened the device with an enormous sample buffer.
Fixes Bugzilla #3555.
We need more than two buffers to flip between if they are small, or CoreAudio
won't make any sound; apparently it needs X milliseconds of audio queued when
it needs to play more or it drops any queued buffers. We are currently
guessing 50 milliseconds as a minimum, but there's probably a more proper
way to get the minimum time period from the system.
Fixes Bugzilla #3656.
This gracefully recovers when a device format is changed, and will switch
to the new default device if the current one is unplugged, etc.
This does not handle when a new default device is added; it only notices
if the current default goes away. That will be fixed by implementing the
stubbed-out MMNotificationClient_OnDefaultDeviceChanged() function.
* alsa hotplug thread is low priority
* give a chance for other threads to catch up when audio playback is not progressing
* use nonblocking for alsa audio capture
There is a bug with SDL hanging when an audio capture USB device is removed, because poll never returns
We will throw away the data anyhow, but some apps depend on the callback
firing to make progress; testmultiaudio.c, if nothing else, is an example
of this.
Capture also will now fire the callback in these conditions, offering nothing
but silence.
Apps can check SDL_GetAudioDeviceStatus() or listen for the
SDL_AUDIODEVICEREMOVED event if they want to gracefully deal with
an opened audio device that has been unexpectedly lost.
This is just enough to get you through a file that just used the extended
header for float or int data. It doesn't handle all the other things that
you expect from this header, like 24-bit samples inside a 32-bit container
or speaker masks.
This should remain binary compatible with Windows XP, as we dynamically
load anything we need and fall back to DirectSound/WinMM/XAudio2 if not
available.
Walter van Niftrik
We have found that since SDL 2.0.5 the audio callback thread is created with a very small stack size. In our application this is leading to stack overflows.
We believe there is a bug at http://hg.libsdl.org/SDL/file/391fd532f79e/src/audio/SDL_audio.c#l1132, where the is_internal_thread flag appears to be inverted.
This defaults to the internal SDL resampler, since that's the likely default
without a system-wide install of libsamplerate, but those that need more can
tweak this.
This currently favors libsamplerate over the fast path (quality over speed),
but I'm not sure that's the correct approach, as there may be surprising
changes in performance metrics depending on what packages are available on
a user's system. That being said, currently, the only thing with access to
SDL_AudioStream is an SDL audio device's thread, and it might be mostly idle
otherwise, so maybe this is generally good.
Turns out that iterating from 0 to channels-1 was a serious performance hit!
These cases now tend to match or beat the original audio resampler's speed!
This allows us to avoid an extra copy, allocate less memory and reduce cache
pressure. On the downside: we have to do a lot of tapdancing to resample the
buffer in reverse when the output is growing.
