* ThreadableJob: Move from AtomicInt to std::atomic
This is the first in a series of commits that migrates away from
AtomicInt towards the C++11 standard library types.
This would allow the removal of AtomicInt and related Qt
version-specific code.
While we're at it, also make ProcessingState an `enum class` so that
it's not implicitly convertible to integers.
* LocklessAllocator: Switch from AtomicInt to std::atomic_int
If it looks like some assignments around the Compare & Swap loops went
missing, it's because compare_exchange_weak() updates the first argument
to the current value when it fails (i.e., returns false).
* BufferManager: Remove extra AtomicInt include
* MixerWorkerThread: AtomicInt to std::atomic_int
* NotePlayHandle: AtomicInt to std::atomic_int
* Remove AtomicInt.h
* Move from QAtomicPointer to std::atomic<T*>
This removes some #ifdef trickery that was being used to get
load-acquire and store-release from Qt5 and "plain" (presumably
sequentially-consistent) loads and stores from Qt4.
In the course of 32b7e04, I removed the channel lock from FxChannel because
I was under the impression that it was only needed to prevent crashes on
channel delete. However, at least two people experience crackling audio
after it was removed (#2708). Therefore, this commit reinstates it.
Lock the mixer before performing a channel delete to prevent any race
conditions causing a crash. Also, update the audioport FX channel when
an InstrumentTrack's FX channel is changed to prevent the audioport
mixing to a nonexistent channel.
Well, this commit got a bit out of hand, what with 26 files changed. Oh well.
Basically, we're using the buffermanager to dispense temporary buffers for playhandles and audioports to use.
This allows us to change the way playhandles work. Earlier, playhandles of the same track were waiting in line
to push their output to the audioport. This was of course inefficient, so now they just register themselves to the port,
then the port handles mixing the buffers.
Caveat: this is still a work in progress, the vol/pan knobs on instruments are temporarily non-functional - will be fixed in
the next commit, but I have to get some sleep now.
Use dynamic building of jobqueues with dependency counting:
- At the start, each channel that has no dependencies is added automatically to the queue
- Then, after each channel is processed, it increments the dep.counter of all its recipients
- When a channel's dep.counter hits the amount of its dependencies (senders), it gets automatically added to the queue
- The queue is finished when the master channel has been processed
- Muted channels are automatically processed at the start regardless dependencies, because they don't have to care about senders, being muted
Hopefully this will improve Fx Mixer performance.
There once have been huge efforts to implement FX send support in the
master branch. In order to make it available on a stable base here's
a backport which is non-trivial as there have been major rewrites of
the mixer's worker thread architecture.
There still seem to be bugs which we have to fix before merging into
stable branch.
Thanks to Andrew Kelley for the original work.
Most files and most of the core classes and their methods have been
renamed to match new coding style conventions:
391 files changed, 25400 insertions(+), 25598 deletions(-)
Furthermore splitted some files where model and view classes were
declared or implemented together in the same file.
Should be tested thoroughly as I might have missed renaming some virtual
methods or SIGNAL/SLOT parameters.
(cherry picked from commit 8c9a9dd14c)
Lukas W