The assumption that the format enum is ordered by size is not valid for quite
some time, since 24bit formats were appended to format enum later than 32bit
formats. This causes resampler to produce properly aligned memblock of size
larger than maximum mempool block size if input format is 24bit and output
format is 32bit.
Oversized block is getting split by `pa_pstream_send_memblock()` into parts of
size not exceeding maximum mempool block size. This usually works well but for
32ch 32bit 48000Hz stream the frame alignment is 128 bytes and maximum mempool
block size value is multiple of 64 but not 128 bytes, therefore resulting parts
are misaligned.
On receiving side this causes extra allocation of 128 byte chunk while `mcalign`
helper reassembles properly aligned frame out of second block of misaligned
size. While first and second properly aligned frames are retrieved successfully
from `mcalign` helper, third retrieved frame would end up with properly aligned
size but misaligned memblock index (in this example, that would be 64 bytes.)
Attempt to push a chunk with misaligned memblock index causes assertion failure
Assertion 'uchunk->index % bq->base == 0' failed at memblockq.c:289,
function pa_memblockq_push(). Aborting.
Fix oversized block issue by checking proper size of format instead of enum
value.
Fixes: a67c21f09 ("merge 'lennart' branch back into trunk.")
Part-of: <https://gitlab.freedesktop.org/pulseaudio/pulseaudio/-/merge_requests/778>
This patch uses the two previous patches to implemnt pseudo-rewinding for the
resamplers by feeding some old data into the resampler after a reset. This is
necessary because PA is using external resamplers that do not implement
rewinding.
To get exactly the same output data from the resampler after a rewind if possible,
the matching period is calculated. This is the number of input samples that produces
an integral number of output samples. After the matching period, the resampler state
repeats. If the matching period is not too large, feeding history into the resampler
will start at a point that is a multiple of the matching period back in time. Then
the resampler will produce exactly the same samples.
The PA_RESAMPLER_MAX_HISTORY value has been replaced by PA_RESAMPLER_MAX_DELAY_USEC
and the required number of history samples is calculated from the sink input sample
rate. The number of history samples can be as large as about 12500.
This fixes glitches during volume changes when the sink runs on a rate different
from the sink input rate.
Part-of: <https://gitlab.freedesktop.org/pulseaudio/pulseaudio/-/merge_requests/120>
The pa_resampler_get_delay() function allows to retrieve the current resampler
delay in input samples for all supported resamplers. The return value is a double
to maintain precision when using variable rate resamplers. Because in many places
the delay is needed in usec, pa_resampler_get_delay_usec() was also supplied.
The speex resampler now skips leading zero samples to provide meaningful delay values.
In the next patch, the pa_resampler_prepare() function will be used to train the
resampler after a rewind. It takes data from a history memblockq and runs it through
the resampler. The output data is discarded.
To make this logic possible, the soxr resampler had to be converted to use variable
rate. The fixed rate version has a variable delay, therefore the logic above could
not be applied. Additionally, with fixed rate, the delay is larger than 150ms in
some situations, while with variable rate the delay is fixed and comparable to the
other resamplers.
Part-of: <https://gitlab.freedesktop.org/pulseaudio/pulseaudio/-/merge_requests/120>
A new memblockq is added to the sink input code to keep some history of the
input data. The queue is kept in sync with the render memblockq. The old input
data will be used to prepare the resampler after a rewind.
pa_resampler_request() and pa_resampler_result() have been changed to round
as good as possible to avoid loosing or duplicating samples during rewinds.
Part-of: <https://gitlab.freedesktop.org/pulseaudio/pulseaudio/-/merge_requests/120>
remixing-produce-lfe controls upmixing, and remixing-consume-lfe
controls downmixing. The motivation is that a user might want to
synthesize LFE while playing stereo audio on his/her 5.1 speakers,
but at the same time follow the industry recommendation to omit
the LFE channel when producting a stereo downmix (e.g. for headphones)
from 5.1 content. Or the other way round.
Fixes: #753.
So far PulseAudio only supported two different work formats: S16NE if
it's sufficient to represent the input and output formats without loss
of precision and FLOAT32NE in all other cases. For systems that use
S32NE exclusively, this results in unnecessary conversions from S32NE to
FLOAT32NE and back again.
Add S32NE remap operations and make use of them (for the COPY and
TRIVIAL resamplers) if both input and output format are S32NE. This
avoids the back and forth conversions between S32NE and FLOAT32NE,
significantly improving performance for those cases.
mpv and vlc play "normal" 5.1 AC3 and DTS files as if they had a
"5.1 (Side)" layout. Which is fine and consistent with ITU
recommendations if the user has a proper 7.1 system. But if the user
actually has a 5.1 system, PulseAudio will try to remap, poorly, between
the "5.1 (Side)" and "5.1" layouts, sending either an average between
front and rear channels, or an attenuated version of that average,
depending on the remixing-use-all-sink-channels setting.
This is not desired, the "Side" channels should be sent to "Rear", it is
only an unfortunate nomenclature confusion.
This patch does not fix 5.1 <-> 7.1 remixing.
Signed-off-by: Alexander E. Patrakov <patrakov@gmail.com>
Correct spelling of 'through' in a comment helps to fix a warning :)
also drop some unrelated comments
Signed-off-by: Peter Meerwald-Stadler <pmeerw@pmeerw.net>
Fix memory leak in pa_resampler_new() in resampler.c, Deallocating
memory of r->lfe_filter in case of fail.
Signed-off-by: Arun Raghavan <arun@arunraghavan.net>
When crossover_freq is set to 0, this restores the old behaviour
of letting the LFE channel be the average of the source channels,
without additional processing. This can be useful e g in case the
user already has a hardware crossover.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
The resampler framework just forwards the request to the lfe filter.
There are no resampler impl that can rewind yet, so just reset the
resampler impl instead of properly rewinding yet.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Store current filter state at every normal block process.
When a rewind happens, rewind back to the nearest saved state,
then calculate forward to the actual sample position.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Add a user defined parameter lfe-crossover-freq for the lfe-filter,
to pass this parameter to the lfe-filter, we need to change the
pa_resampler_new() API as well.
Signed-off-by: Hui Wang <hui.wang@canonical.com>
When enable-lfe-remixing is set, an LFE channel is present in the
resampler's destination channel map but not in the source channel map,
we insert a low-pass filter instead of just averaging the channels.
Other channels will get a high-pass filter.
In this patch, the crossover frequency is hardcoded to 120Hz (to be fixed
in later patches).
Note that in current state the LFE filter is
- not very optimised
- not rewind friendly (rewinding can cause audible artifacts)
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Added ID and names for the resampler presets and also updated the working sample rate deduction to take the new resampler into account. The initial libsoxr backend version does not variable rate resampling, so it is disabled in this case.
FSF addresses used in PA sources are no longer valid and rpmlint
generates numerous warnings during packaging because of this.
This patch changes all FSF addresses to FSF web page according to
the GPL how-to: https://www.gnu.org/licenses/gpl-howto.en.html
Done automatically by sed-ing through sources.
speex_resample_float() does not work with speex compiled with
--enable-fixed-point, because speex expects its float input
to be normalized to ±32768 instead of the more usual ±1.
It is possible to fix speex_resample_float(), as demonstrated at
http://lists.freedesktop.org/archives/pulseaudio-discuss/2014-May/020617.html
However, a better idea is to avoid using the speex-float resampler and
the associated s16 <-> float conversions that speex will immediately undo
internally if it is known that speex has been compiled with FIXED_POINT.
So, transparently change speex-float-* to speex-fixed-* in that case.
Signed-off-by: Alexander E. Patrakov <patrakov@gmail.com>
Reported-by: Fahad Arslan <fahad_arslan@mentor.com>
Cc: Damir Jelić <poljarinho@gmail.com>
Cc: Peter Meerwald <pmeerw@pmeerw.net>
FIXED_POINT detection is based on code by Peter Meerwald.
State can be used by remap function implementations to
speed up the remapping, e.g. by precomputing things or
even by generating specialized code for a specific channel
remapping task
Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
Initialization of the remap structure now happens in one place
Rename calc_map_table() to setup_remap(), copy sample format and
channel specs; the remap structure is initialized when we know the
work sample format of the resampler
Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
pa_init_remap_func() only sets the appropriate remapping function, it
does not initialize the pa_remap struct
Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
Reintroduces a cleaned-up version of commit 30ce3a14e5 which
was reverted by 1ce71cbd82; for more information see
http://thread.gmane.org/gmane.comp.audio.pulseaudio.general/17479/focus=17487
The patch intends to reduce computational load when resampling AND remapping. The PA
resampler performs the following steps:
sample format conversion -> remapping -> resampling -> sample format conversion
In case the number of output channels is higher than the number of input channels, the
resampler has to be run more often than necessary. E.g. in case of mono to 4-channel remapping,
the resampler runs on 4 channels separately.
To ímprove this, the PA resampler pipeline is made adaptive:
if out-channels <= in-channels:
sample format conversion -> remapping -> resampling -> sample format conversion
if out-channels > in-channels:
sample format conversion -> resampling -> remapping -> sample format conversion
Signed-off-by: Peter Meerwald <p.meerwald@bct-electronic.com>
the patch changes the interface of the (internal) fit_buf() function:
fit_buf() manages the memblock of the buf chunk, it reallocates the memblock
if the requested number of bytes ('len') if larger than the memblock's size ('size')
and optionally preserves 'copy' bytes
the code should be in line with the comment now
Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
I think this makes the code a bit nicer to read and write. This also
reduces the chances of off-by-one errors when checking the bounds of
sample rate values.
I think this makes the code a bit nicer to read and write. This also
reduces the chances of off-by-one errors when checking the bounds of
the sample format value.
buf in struct ffmpeg_data is reset() initially and freed, but never
actually used
when a new block is allocated ffmpeg_data->buf[c].length is used
(which is always 0) to compute the new block size
so, drop buf
Signed-off-by: Peter Meerwald <p.meerwald@bct-electronic.com>
leftover_buf points to the output buffer of a stage containing leftover
data; similar for leftover_buf_size and have_leftover
Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
some resampler implementations (e.g. libsamplerate and ffmpeg) do not consume
the entire input buffer; the impl_resample() function now has a return value
returning the number of frames in the input buffer not processed
these frames must be saved in appropriate buffer and presented together with
new input data
also change the parameter names from in_samples, out_samples to in_n_frames,
out_n_frames, respectively (n_frames = samples / channels)
Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
This patch fixes a small mistake where we actually log that we are
reverting to the auto resampler if we can't use the 'copy' resampler but
never do the revert.
This would lead to a crash if the user chooses the 'copy' resampler and
then tries to play something that needs to be resampled.
With very low input sample rates the memory pool max block size may
not be big enough, in which case we should return the size of one
frame. Returning zero caused crashing.
BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=68616
This adds asserts to check if the implementation has an update rate
function defined for the unlikely event that some implementation forgets
to assign a update rate function we can simply bail.
It is expected from the resampling implementations to have such a
function even if the state of the resampler is completely reset.
