This interface is very simple for the time being and can be used to
provide encoding functionality in a generalized way. Initialization and
parameter setting are intentionally not abstracted for now, since this
is usually very format-specific. We just need a general interface for
writing samples and errorable finalization.
WavWriter and the shot utility open files with this mode and never
truncate the files, which might leave some contents of a previous file
during overwriting.
This will ensure that we don't leak any memory while playing back
audio.
There is an expectation value in the test that is only set to true when
PulseAudio is present for the moment. When any new implementation is
added for other libraries/platforms, we should hopefully get a CI
failure due to unexpected success in creating the `PlaybackStream`.
To ensure that we clean up our PulseAudio connection whenever audio
output is not needed, add `PulseAudioContext::weak_instance()` to allow
us to check whether an instance exists without creating one.
If we don't clear the callbacks, they may be called after our functions
are deleted.
Disconnecting the stream also doesn't appear to be done automatically
when calling `pa_stream_unref()` for the last time, so let's do that.
We don't want to pull the stream out from under our PulseAudio main
loop, so call these with the lock to ensure that nothing is touching
them.
The `pa_threaded_mainloop_stop()` call does not require lock as it sets
a flag to tell the main loop to exit.
The mutex used to protect from multiple threads creating PulseAudio
contexts simultaneously could remain locked when an application exited.
The static variables' destructors could be called on the main thread
while another thread is running `PulseAudioContext::instance()` and
synchronously connecting to a PulseAudio daemon. This would cause an
assertion in Mutex that it is unlocked upon its destructor being
called.
By creating a static `ScopeGuard` that locks and immediately unlocks
the mutex, we can ensure that the main thread waits for the connection
to succeed or fail. In most cases, this will not take long, but if the
connection is timing out, it could take a matter of seconds.
Now that `Thread` keeps itself alive when it is running detached, we do
not need to hold onto it in the PulseAudio playback stream's internal
state object. This was a hack that did not work correctly because the
`Thread` object and its action `Function` would be deleted before the
action had exited and cause a crash.
This adds an abstract `Audio::PlaybackStream` class to allow cross-
platform audio playback to be done in an opaque manner by applications
in both Serenity and Lagom.
Currently, the only supported audio API is PulseAudio, but a Serenity
implementation should be added shortly as well.
Removes the Sample struct inside Piano and replaces it with the struct
from LibDSP.
It automatically scales the height of the wave depending on the maximum
amplitude, as the Samples now contain floats and not integers.
A previous commit made it so that SeekTable doesn't provide a seek
point from `seek_point_before()` if there is not a seek point before
the requested sample index. However, MP3Loader was only setting a seek
point after the first 10 frames, meaning that it would do nothing when
seeking back to 0.
To fix this, add a seek point at byte 0 for the first sample, so that
`seek_point_before()` will never fail.
Bytes will implicitly cast to StringView, but not to ReadonlyBytes. That
means that if you call
`Audio::Loader::create_plugin(mapped_file->bytes())`
it will silently use the `create_plugin(StringView path)` overload.
Reading audio data does not require that data to be writable, so let's
use ReadonlyBytes for it and avoid the footgun.
- Pre-allocate and reuse sample decompression buffers. In many FLAC
files, the amount of samples per frame is either constant or the
largest frame will be hit within the first couple of frames. Also,
during audio output, we need to move and combine the samples from the
decompression buffers into the final output buffers anyways. Avoiding
the reallocation of these large buffers provides an improvement from
16x to 18x decode speed on strongly compressed but otherwise usual
input.
- Leave a FIXME for a similar improvement that can be made in the
residual decoder.
- Pre-allocate audio chunks if frame size is known.
- Use reasonable inline capacities in several places where we know the
maximum or usual capacity needed.
Instead of using a seek tolerance value to get close enough to the
target, we can skip frames forward until we pass the target, then seek
back to the previous frame. That puts us in a position to immediately
decode the frame containing the target sample.
Previously, the FLAC loader would not skip samples to reach its seek
target if it saw that the current sample in the loader is closer to the
target than the seek point it finds. This prevents seeking forward when
there are no seek points past the current position.
Previously, the calculation of the distance to the previous seekpoint
would always behave as if a seek point existed at sample 0, meaning
that it would never place a seek point there. If we instead treat it as
the maximum distance if no sample is present, a seek point will be
placed.
If the seek table was incomplete, without any seek points available
before the target point, `SeekTable::seek_point_before()` would instead
return the first seek point after the target. Check whether the seek
point is before the target before returning it.
We downsample multi-channel files into stereo for now, which at least
makes the other channels listenable. The new multi-channel downmix
helper is intended to be used for other formats with the same or similar
channel arrangement, such as QOA.
Especially FLAC had an issue here before, but the loader infrastructure
itself wouldn't handle end of stream properly if the "available samples"
information didn't match up.
It's no longer needed now that this code uses ErrorOr instead of Result.
Ran:
rg -lw LOADER_TRY Userland/Libraries/LibAudio \
| xargs sed -i '' 's/LOADER_TRY/TRY/g'
...and then manually fixed up Userland/Libraries/LibAudio/LoaderError.h
to not redefine TRY but instead remove the now-unused LOADER_TRY,
and ran clang-format.
For very large seekpoint indices, the casts necessary for the "simple"
subtraction comparison will yield wrong and overflowing results.
Therefore, we perform the seekpoint comparison manually instead.
This specialized UTF-8 decoder is more powerful than a normal UTF-8
decoder anyways, but it couldn't account for the never spec-compliant
0xff start byte. This commit makes that byte behave as expected if
taking UTF-8 to its extreme, even if it is a little silly and likely not
relevant for real applications.
The bit magic for two's complement sign extension was only sign
extending to 32-bit signed. This issue was exposed by the last commit,
where now we actually use the 64-bit return value.
Since we can have up to 32 bits of input data, multiplications may need
up to 63 bits. This was accounted for in some places, but by far not in
all, and oss-fuzz found multiple integer overflows. We now use i64 in
all of the decoding, since we need to rescale samples to float later on
anyways. If a final sample value ends up out of range (and the range can
be a maximum of 32 bits), we may get samples past 1, but that then is a
non-compliant input file, and using over-range samples (and most likely
clipping audio) is considerably less weird than overflowing and
glitching audio.
The fuzzer found one heap buffer overflow here due to confusion between
u32* and u8* (the given size is for bytes, but we used it for 32-bit
elements, quadrupling it), and it looks like there's an opportunity for
several more. This commit modernizes the picture loader by using
String's built-in stream loader, and also adds several spec-compliance
checks: The MIME type must be ASCII in a specific range, and the picture
description must be UTF-8.
An LPC predictor (fixed or not) contains as many warm-up samples as its
order. Therefore, the corresponding subframe must have at least this
many samples.
This turns this fuzzer-found crash into a handleable format error.
There are at most 576 granule samples/frequency lines, but the side data
can specify that the big_values granule type take up to 1024 samples.
The spec says in 2.4.3.4.6 that we should always stop reading Huffman
data once we have 576 samples, so that is what this change does. I also
add some useful spec comments while I'm here.
This change was a long time in the making ever since we obtained sample
rate awareness in the system. Now, each client has its own sample rate,
accessible via new IPC APIs, and the device sample rate is only
accessible via the management interface. AudioServer takes care of
resampling client streams into the device sample rate. Therefore, the
main improvement introduced with this commit is full responsiveness to
sample rate changes; all open audio programs will continue to play at
correct speed with the audio resampled to the new device rate.
The immediate benefits are manifold:
- Gets rid of the legacy hardware sample rate IPC message in the
non-managing client
- Removes duplicate resampling and sample index rescaling code
everywhere
- Avoids potential sample index scaling bugs in SoundPlayer (which have
happened many times before) and fixes a sample index scaling bug in
aplay
- Removes several FIXMEs
- Reduces amount of sample copying in all applications (especially
Piano, where this is critical), improving performance
- Reduces number of resampling users, making future API changes (which
will need to happen for correct resampling to be implemented) easier
I also threw in a simple race condition fix for Piano's audio player
loop.
This removes a lot of duplicated stream creation code from the plugins,
and also simplifies the way that the appropriate plugin is found. This
mirrors the ImageDecoderPlugin design and necessitates new sniffing
methods on the loaders.
This is a sensible separation of concerns that mirrors the WindowServer
IPC split. On the one hand, there is the "normal" audio interface, used
for clients that play audio, which is the primary service of
AudioServer. On the other hand, there is the management interface,
which, like the WindowManager endpoint, provides higher-level control
over clients and the server itself.
The reasoning for this split are manifold, as mentioned we are mirroring
the WindowServer split. Another indication to the sensibility of the
split is that no single audio client used the APIs of both interfaces.
Also, useless audio queues are no longer created for managing clients
(since those don't even exist, just like there's no window backing
bitmap for window managing clients), eliminating any bugs that may occur
there as they have in the past.
Implementation-wise, we just move all the APIs and implementations from
the old AudioServer into the AudioManagerServer (and respective clients,
of course). There is one point of duplication, namely the hardware
sample rate. This will be fixed in combination with per-client sample
rate, eliminating client-side resampling and the related update bugs.
For now, we keep one legacy API to simplify the transition.
The new AudioManagerServer also gains a hardware sample rate change
callback to have exact symmetry on the main server parameters (getter,
setter, and callback).
WavWriter needs a TON of modernization work, but for now this commit
just tackles two FIXMEs by converting samples correctly into all
supported integer PCM formats. The supported formats are only signed
16-bit and unsigned 8-bit for now, but can be expanded later. At least
we don't produce horrible speaker-destroying noise when writing any
other format.
With this, the WAV loader is a completely modern LibAudio loader:
- Own type header for RIFF data structures
- custom stream read functions for the types
- Final removal of legacy I/O error checking
- clearer error messages
- clean handling of header chunks
The latter will allow proper handling of other chunks (before "data") in
the future, such as metadata :^)
That's what this class really is; in fact that's what the first line of
the comment says it is.
This commit does not rename the main files, since those will contain
other time-related classes in a little bit.
