Now we sleep the thread in WaitDevice until ALSA reawakens it because it
needs more data, and we feed it exactly as much as it can take at that
point.
Like the recent PulseAudio changes, this is both more efficient, reliable,
and consistent.
In practice, this seems to buffer a little upfront and then gives a pretty
consistent request flow after that of 1/4 of the requested buffer size without
variation, which is significantly better than the previous code that would
vary a little each frame.
Plus, as long as the device asks for _anything_, we won't block forever, and
if it asks for more than our expected buffer size, we'll run multiple times
to satisfy it, so this is likely more robust against dropouts in general, too.
Previously, if acting on a surface with less than 32 bits per pixel,
this code was placing the pixel value from the surface in the first
few bytes of the Uint32 to be decoded, and unrelated data from a
subsequent pixel in the remaining bytes.
Because SDL_GetRGBA takes the bits to be decoded from the
least-significant bits of the given value, ignoring the higher-order
bits if any, this happened to be correct on little-endian platforms,
where the first few bytes store the least-significant bits of an
integer.
However, it was incorrect on big-endian, where the first few bytes are
the most-significant bits of an integer.
The previous implementation also assumed that unaligned access to a
32-bit quantity is possible, which is not the case on all CPUs (but
happens to be true on x86).
These issues were not discovered until now because
SDLTest_CompareSurfaces() is only used in testautomation, which until
recently was not being run routinely at build-time, because it contained
other assumptions that can fail in an autobuilder or CI environment.
Resolves: https://github.com/libsdl-org/SDL/issues/8315
Signed-off-by: Simon McVittie <smcv@collabora.com>
This reverts commit 6fd0613ac8.
Turns out that the Steam Runtime is still on PulseAudio 1.1, and the only
thing we (currently) need a newer Pulse for is pa_threaded_mainloop_set_name,
so let's just go back to treating that symbol as optional.
We might need to force a higher version at some point, but it's not worth it
over this.
Otherwise, we get into situations where all bound streams need to change
their output formats when a device pauses...and it makes the fast case
slow: when pausing a single input, it needs to silence and then convert a
silent buffer, instead of just zeroing out the device buffer and being done.
Since these get proxied to a different thread, if we wait for that thread
to finish while holding the lock, and the management thread _also_ requests
the lock, we're screwed.
WaitDevice never holds the lock by design, so just mark devices as failed
and clean up or recover them in there.
The audio processing thread isn't scheduled in lock-step with the audio callback so sometimes the callback would consume the same data twice and sometimes the audio processing thread would write to the same buffer twice.
Also handle variable sizes in the audio callback so the Android audio system doesn't have to do additional buffering to match our buffer size requirements.
Querying the drag offer with every pointer movement would require refactoring to work with the portal implementation, however, there is little point, as the event layer just discards the file name. Remove the existing code and note that a new implementation is needed if the name ever starts to be passed though.
In the event that this is reimplemented, it should cache the filenames, as otherwise, this could potentially hammer the DBus interface hundreds or even thousands of times per second.
Saves locks and copies during audio thread iteration. We've added asserts
that can evaporate out in release mode to make sure everything stays in sync.
Libdecor creates subsurfaces of the primary SDL surface, but events from these surfaces should be ignored, or applications will get drag & drop events when dragged over drop shadows and such.
Presumably this is an accidental character due to the copyright symbol and conversion to/from different encodings. The *.c file does not have this character.
On my laptop, the battery is configured to stop charging at around 80%
most of the time, to increase the overall useful lifetime of the battery.
When in that state, upower reports UP_DEVICE_STATE_PENDING_CHARGE
(numeric value 5), which SDL previously mapped to SDL_POWERSTATE_UNKNOWN.
This made the platform_testGetPowerInfo automated test fail, because
it assumes that SDL_POWERSTATE_UNKNOWN means no battery is connected,
and does not expect to see a percentage.
Map UP_DEVICE_STATE_PENDING_CHARGE (5) to SDL_POWERSTATE_CHARGED, which
seems close enough.
Also map UP_DEVICE_STATE_PENDING_DISCHARGE (6) to
SDL_POWERSTATE_ON_BATTERY, which matches how at least GNOME presents it.
Signed-off-by: Simon McVittie <smcv@collabora.com>
This fires if an opened device changes formats (which it can on Windows,
if the user changes this in the system control panel, and WASAPI can
report), or if a default device migrates to new hardware and the format
doesn't match.
This will fire for all logical devices on a physical device (and if it's
a format change and not a default device change, it'll fire for the
physical device too, but that's honestly not that useful and might change).
Fixes#8267.
("preconverted bytes" makes it sounds like we already converted them before
the call instead of "bytes that haven't yet hit the stage where we convert
them. Just dump the wording completely.)
