ericonr
/
sndio-htdocs
mirror of https://github.com/ericonr/sndio-htdocs.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
sndio-htdocs/tips.html

1154 lines
28 KiB
HTML
Raw Permalink Blame History

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<!--
/*
* Copyright (c) 2008-2011 Alexandre Ratchov <alex@caoua.org>
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
-->
<html>
<head>
<title>sndio - hints on writing &amp; porting audio code</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta name="language" content="en">
<style type="text/css">
@media screen, print {
h1 {
text-align: center
}
table {
border-style: solid;
border-width: thin;
border-collapse: collapse;
margin-left: auto;
margin-right: auto;
}
th {
border-style: solid;
border-width: thin;
padding: 0.5em;
}
td {
border-style: solid;
border-width: thin;
padding: 0.5em;
}
pre {
padding-top: 1em;
padding-bottom: 1em;
padding-left: 4em;
padding-right: 4em;
}
dt {
font-weight: bold;
}
dd {
padding: 0.5em;
}
}
@media screen {
body {
margin-top: 1em;
margin-bottom: 1em;
margin-left: auto;
margin-right: auto;
padding-top: 1em;
padding-bottom: 0em;
padding-left: 1em;
padding-right: 1em;
border-style: dashed;
border-top-width: thin;
border-bottom-width: thin;
border-left-width: thin;
border-right-width: thin;
position: relative;
max-width: 44em;
border-color: #ccc;
background-color: #fff;
}
th {
background-color: #eee
}
pre {
background-color: #eee
}
}
@media print {
pre {
border-style: dotted;
border-width: thin;
}
}
</style>
</head>
<body>
<h1>sndio<br>hints on writing &amp; porting audio code</h1>
<!-- toc_begin - automatically generated, don't modify -->
<h2>Table of contents</h2>
<ul>
<li><a href="#section_1">1 Introduction</a>
<ul>
<li><a href="#section_1_1">1.1 Aim of this document</a>
<li><a href="#section_1_2">1.2 Device model overview</a>
</ul>
<li><a href="#section_2">2 Parameter negotiation</a>
<ul>
<li><a href="#section_2_1">2.1 Selecting formats and encodings</a>
</ul>
<li><a href="#section_3">3 Choosing the buffer size</a>
<li><a href="#section_4">4 Synchronizing stuff on audio playback</a>
<ul>
<li><a href="#section_4_1">4.1 Absolute play and record positions</a>
<li><a href="#section_4_2">4.2 Playback latency à la GET_ODELAY</a>
<li><a href="#section_4_3">4.3 Playback buffer usage à la GET_OSPACE</a>
<li><a href="#section_4_4">4.4 Record buffer usage à la GET_ISPACE</a>
<li><a href="#section_4_5">4.5 Sleeping until there's space for one block in the play buffer</a>
</ul>
<li><a href="#section_5">5 Choosing and using the block size</a>
<ul>
<li><a href="#section_5_1">5.1 Getting the block size to optimize I/O</a>
<li><a href="#section_5_2">5.2 Using a small block size for low latency</a>
<li><a href="#section_5_3">5.3 Getting higher clock resolution for synchronization</a>
</ul>
<li><a href="#section_6">6 Adjusting the volume, mute knob</a>
<ul>
<li><a href="#section_6_1">6.1 Setting the volume</a>
<li><a href="#section_6_2">6.2 Volume feedback (reading the current volume 1)</a>
<li><a href="#section_6_3">6.3 Volume getter (reading the current volume 2)</a>
</ul>
<li><a href="#section_7">7 Pausing and resuming</a>
<li><a href="#section_8">8 Using sndio in multi-threaded programs</a>
<li><a href="#section_9">9 Windows-style callbacks</a>
<li><a href="#section_10">10 Pitfalls</a>
<ul>
<li><a href="#section_10_1">10.1 Playback and record aren't independent</a>
<li><a href="#section_10_2">10.2 Using appbufsz instead of bufsz</a>
<li><a href="#section_10_3">10.3 How many bytes to store a 24-bit sample</a>
<li><a href="#section_10_4">10.4 poll(2) not called fast enough</a>
</ul>
<li><a href="#section_11">11 Glossary</a>
</ul>
<!-- toc_end -->
<h2><a name="section_1">1 Introduction</a></h2>
<h3><a name="section_1_1">1.1 Aim of this document</a></h3>
<p>
This document contains simple tips on how to write new code for the
sndio API as well as how to port existing code to it.
This document doesn't explain how to invoke sndio functions, which are
already described in the sio_open(3) manual page.
<p>
Remember to keep things as simple as possible; the sndio API is
designed to make this possible.
If something looks complicated, the approach may be wrong.
In some cases, it may be better to drop some complicated feature rather
than adding hackish code that may hurt the overall correctness and
robustness of the application.
<h3><a name="section_1_2">1.2 Device model overview</a></h3>
<p>
The sndio device model is as follows:
<ul>
<li>
A bidirectional data stream exists between the program and the
sound-card, for the <i>play</i> and <i>record</i> directions,
respectively.
<li>
Data is a sequence of frames, where each frame corresponds to a
sample for all channels of the stream.
It is submitted and retrieved using functions similar to the read(2)
and write(2) syscalls.
<li>
A wall clock ticks when samples are processed by the hardware; i.e.
the <i>n</i>-th frame of the stream corresponds to the <i>n</i>-th
clock tick.
The clock is exposed through a callback mechanism: a function
registered by the program is called periodically, which takes as argument
the number of clock ticks elapsed since the last call.
</ul>
In other words, the <i>n</i>-th sample read is recorded exactly when
the <i>n</i>-th written sample is played.
This means that samples to play must be written before
recorded samples can be read, otherwise a deadlock will occur.
<h2><a name="section_2">2 Parameter negotiation</a></h2>
To minimize mistakes, the following approach can
be used:
<ol>
<li>
call sio_setpar(3) using the application's native parameters.
<li>
call sio_getpar(3) and verify whether returned parameters
are usable.
</ol>
<p>
Certain applications support multiple parameters sets, so if the above
steps failed, you may want to retry with another set. However, that's
unlikely to work in real life for two reasons:
<ul>
<li>
apps often support "common" formats that "common" hardware
supports, so if it didn't work, it's probably because the
hardware is not so "common". Therefore, trying another "common"
format the application supports has little chance of working
<li>
that's more code, so greater chance of introducing a bug. Why? To allow
the app to emulate the format one particular piece of
hardware supports. Is it worth the effort given that sndiod(8)
already does emulation and supports <b>any</b> hardware
and is always enabled by default?
</ul>
<h3><a name="section_2_1">2.1 Selecting formats and encodings</a></h3>
A typical example is a simple player that tries to play 2-channel, s16
at 44.1kHz. If the audio subsystem doesn't support this format it should
just fail, so the following is OK:
<pre>
...
par.pchan = 2;
par.sig = 1;
par.bits = 16;
par.le = SIO_LE_NATIVE;
par.rate = 44100;
if (!sio_setpar(hdl, &amp;par))
errx(1, "internal error, sio_setpar() failed");
if (!sio_getpar(hdl, &amp;par))
errx(1, "internal error, sio_getpar() failed");
if (par.pchan != 2)
errx(1, "couldn't set number of channels");
if (!par.sig || par.bits != 16 || par.le != SIO_LE_NATIVE)
errx(1, "couldn't set format");
if (par.bits != 16 || par.bps != 2)
errx(1, "couldn't set precision");
if (par.rate &lt; 44100 * 995 / 1000 ||
par.rate &gt; 44100 * 1005 / 1000)
errx(1, "couldn't set rate");
...
</pre>
As sndiod(8) is used by default, sio_setpar(3) will always use the requested
parameters.
If the user has requested direct access to the hardware, then
sio_setpar(3) may configure the device to other parameters, so the new ones
must be checked with sio_getpar(3).
<h2><a name="section_3">3 Choosing the buffer size</a></h2>
The buffer size represents the amount of time given to the
application to produce data to play (or to consume recorded data).
If the application doesn't respect this constraint, xruns will occur.
<p>
Therefore, we must estimate the maximum time it will take to
prepare the data and to fill the buffer, and then choose a slightly larger
buffer size by setting the appbufsz parameter in the sio_par
structure.
<p>
On a multitasking system, the delay estimate must take into account
the other processes hogging the system.
On a typical Unix-like system, a margin of around
~5-10ms seems OK. If the buffer size is not set, the audio subsystem
will choose a reasonable value, something around 50ms.
<p>
For example, consider a file player. It's organized as follows:
<pre>
for (;;) {
read_file_to_fifo();
play_from_fifo();
}
</pre>
The maximum time it takes for the application to call play_from_fifo()
is roughly equal to the maximum time read_file_to_fifo() takes to
complete. Reading from a file may block for around 50ms, so around
100ms of buffer is mostly OK. If the file uses a 44.1kHz sampling
rate, then the buffer size is:
<pre>
0.1s * 44100Hz = 4410 frames
</pre>
The orders of magnitudes of the maximum delay for different operations,
measured on a slow i386 system with ~2 users doing simple stuff
(editors, basic X11, compilations), can be seen below:
<p align="center">
<table>
<tr>
<th>operation
<th>max delay
<tr>
<td>extract a block from a CD
<td>300ms
<tr>
<td>read less than 64kB from hard disk
<td>50ms
<tr>
<td>read from a pipe + pair of context switches
<td>10ms
</table>
<p>
<b>Note</b>: the device may choose a different buffer size that the one the
application requested.
In any case, the application must use sio_getpar(3) and take into account the
actual buffer size.
<h2><a name="section_4">4 Synchronizing stuff on audio playback</a></h2>
<p>
Timing information is available by setting up a callback
with the sio_onmove(3) function:
<pre>
struct sio_par par;
long long writecnt; /* frames written (in bytes) */
long long readcnt; /* frames read (in bytes) */
long long realpos; /* frame number Joe is hearing */
void
cb(void *addr, int delta)
{
realpos += delta;
}
int
main(void)
{
sio_hdl *hdl;
sio_par par;
...
writecnt = readcnt = realpos = 0;
sio_onmove(hdl, cb, NULL);
...
for (;;) {
...
writepos += sio_write(hdl, buf, count);
...
readpos += sio_read(hdl, buf, count);
...
}
...
}
</pre>
The callback is invoked every time a block is processed by the hardware.
It's called from one of the following functions:
<ul>
<li>sio_revents(3) after poll(2)
<li>blocking sio_write(3) and sio_read(3)
</ul>
<h3><a name="section_4_1">4.1 Absolute play and record positions</a></h3>
<p>
The absolute play position is given by realpos, from the above example.
If the application needs this expressed in seconds:
<pre>
realpos_sec = realpos / par.rate;
</pre>
Note that in earlier versions of sndio, ``realpos'' could be
negative, but that feature was removed.
<h3><a name="section_4_2">4.2 Playback latency à la GET_ODELAY</a></h3>
The playback latency is the delay (expressed in number of frames) that
it will take until the last frame that was written becomes audible.
This is exactly the buffer usage:
<pre>
writepos = writecnt / (par.pchan * par.bps); /* convert to frames */
bufused = writepos - realpos;
</pre>
The recording latency is generally zero, since the application is
waiting and consuming the data immediately.
<h3><a name="section_4_3">4.3 Playback buffer usage à la GET_OSPACE</a></h3>
<p>
Certain applications ask for the number of bytes left in the playback
buffer, assuming that sio_write(3) will not block if the program writes
less than the space available in the buffer.
<b>This is wrong</b>, but sometimes it's not desirable to change the
application, so the available buffer space could be calculated as follows:
<pre>
space_avail = par.bufsz - bufused;
</pre>
<h3><a name="section_4_4">4.4 Record buffer usage à la GET_ISPACE</a></h3>
Using this for non-blocking I/O is wrong too,
nevertheless the buffer usage is:
<pre>
readpos = readcnt / (par.rchan * par.bps);
bufused = realpos - readpos;
</pre>
<h3><a name="section_4_5">4.5 Sleeping until there's space for one block in the play buffer</a></h3>
<p>
Certain applications want to sleep until there's space
for at least one block in the play buffer.
There's no way to wait for such an event, and that's not compatible
with Unix file semantics.
<p>
The best approach is to change the application to use poll(2).
If that's not possible, wait until the stream is writable as follows:
<pre>
void
wait_space_avail(void)
{
int nfds, revents;
struct pollfd *pfds = malloc(sio_nfds(hdl) * sizeof(*pfds));
do {
nfds = sio_pollfd(hdl, pfds, POLLOUT);
if (nfds &gt; 0) {
if (poll(pfds, nfds, -1) &lt; 0)
err(1, "poll failed");
}
revents = sio_revents(hdl, pfds);
if (revents &amp; POLLHUP)
errx(1, "device disappeared");
} while (!(revents &amp; POLLOUT));
}
</pre>
Other approaches would probably lead to stuttering or to a busy loop,
which, in turn, may lead to stuttering.
<p>
Note, however, that if poll(2) is called with no file descriptors
and non-zero timeout, it will hang, and if timeout is negative, it
will hang forever. That means we need to check if nfds is positive.
<h2><a name="section_5">5 Choosing and using the block size</a></h2>
<h3><a name="section_5_1">5.1 Getting the block size to optimize I/O</a></h3>
<p>
Audio is a continuous stream of frames, but the hardware processes
them in blocks. A typical player will have an internal ring that will
be filled by the player and consumed using sio_write(3). If the ring
size is a multiple of the hardware block size, then calls to
sio_write(3) will be optimal.
<p>
The block size is stored in the ``round'' field of the sio_par
structure, and is negotiated using sio_setpar(3) and sio_getpar(3).
Application should round their internal buffer sizes as follows:
<pre>
buf_size = desired_buf_size + par.round - 1;
buf_size -= buf_size % par.round;
</pre>
The ``round'' parameter is very constrained by the hardware, so
sio_setpar(3) only uses it as a hint.
<h3><a name="section_5_2">5.2 Using a small block size for low latency</a></h3>
The minimum latency a program can get is related to the minimum buffer
size, which is often one or two blocks. So if an application needs very
low latency, it must use a small block size too, but there's no need to
change it explicitly.
<p>
When changing the ``appbufsz'' parameter, an optimal block size is
calculated by the sio_setpar(3) function. The sio_setpar(3) function
will evolve to cope with future hardware and software constraints, so
it's expected to always do the right thing, on any hardware. Therefore,
in order to get the maximum robustness, don't change the block size.
<h3><a name="section_5_3">5.3 Getting higher clock resolution for synchronization</a></h3>
<p>
Synchronization is based on the callback set with the sio_onmove(3)
function. It's called periodically, every time a block is
processed. Basically, this provides clock ticks to the program,
which correspond to the sound card's clock.
<p>
If the block size is large, the tick rate is low, and time increases in
big steps, which may not be desirable for applications requiring higher
clock resolution.
The easiest solution is to use a smaller block size to get a higher
tick rate. This approach has the advantage of being very accurate,
but it's CPU intensive. It's also not always possible to choose the
block size (e.g. because of hardware constraints).
<p>
Example: a video player plays 25 images per second. To get a smooth
video, images must be displayed at regular time intervals. Thus, the
clock resolution must be at least twice the image rate, i.e. 50 ticks
per second. If the audio rate is 44.1kHz, the maximum block size to get
smooth video is:
<pre>
44100Hz / 50Hz = 882 frames per block
</pre>
Another solution is to use a large block size, and extrapolate the
time between clock ticks using gettimeofday(2). This is more
complicated to get right, but works in all situations, is less CPU
intensive and works even if very high clock resolution is needed.
<h2><a name="section_6">6 Adjusting the volume, mute knob</a></h2>
<h3><a name="section_6_1">6.1 Setting the volume</a></h3>
<p>
It's as simple as calling sio_setvol(3) with a value in the 0..127
range, where 0 means ``mute the stream'' and 127 is the maximum volume
(the default).
Certain apps use percents in the 0..100 range, in that case a conversion
must be performed as follows:
<pre>
#define PCT_TO_SIO(pct) ((127 * (pct) + 50) / 100)
#define SIO_TO_PCT(vol) ((100 * (vol) + 64) / 127)
void setvol(int p)
{
...
sio_setvol(hdl, PCT_TO_SIO(p));
}
</pre>
<h3><a name="section_6_2">6.2 Volume feedback (reading the current volume 1)</a></h3>
<p>
There's no getter for the current volume; instead the program
can install a callback to be notified about volume changes:
<pre>
void
cb(void *addr, unsigned vol)
{
redraw_volume_slider(SIO_TO_PCT(vol));
}
int
main(void)
{
...
sio_onvol(hdl, cb, NULL);
...
for (;;) {
p = mouse_event_to_pct();
setvol(p);
}
}
</pre>
<h3><a name="section_6_3">6.3 Volume getter (reading the current volume 2)</a></h3>
<p>
Certain applications require a ``get volume'' function and
work as follows:
<pre>
for (;;) {
p = volume_slider_to_pct();
setvol(p);
p = getvol();
move_volume_slider(p);
}
</pre>
One may think that it's enough to set a global ``current volume''
variable in the callback and to return it in the getter. This can't
work because the following property is required:
<pre>
x == SIO_TO_PCT(PCT_TO_SIO(x)) /* for all x */
y == PCT_TO_SIO(SIO_TO_PCT(y)) /* for all y */
</pre>
So it may lead to various weird effects like the cursor stuttering
around a given position, or ``+/- volume'' keyboard shortcuts not
working.
The correct implementation is to use feedback as in the above section.
If that's not possible, a fake getter can be implemented as follows:
<pre>
unsigned current_pct;
void
cb(void *addr, unsigned vol)
{
if (vol != PCT_TO_SIO(current_pct))
current_pct = SIO_TO_PCT(vol);
}
unsigned
getvol(int p)
{
return current_pct;
}
</pre>
<h2><a name="section_7">7 Pausing and resuming</a></h2>
<p>
Pause and resume functions do not exist, because it's hard to properly
implement on <b>any</b> hardware.
If the pause feature is required, it's easier to stop the
stream with sio_stop(3) and to later restart it with sio_start(3).
<p>
Certain programs expect a pause-resume cycle to not change the
amount of buffered data.
If so, the "resume" function must play the same amount of silence as the
amount of data the buffer contained when the "pause" function was called.
<p>
<b>Update&nbsp;: </b>Doing nothing would also work, but only in few cases.
If you just stop providing data to play, the stream will underrun and stop
automatically.
Once data is available again, the stream will resume automatically.
However, this abuse of the xrun mechanism is not desirable for two reasons:
<ul>
<li>The device will still be processing data (silence)
and will waste CPU time (which consumes more energy from laptop batteries).
<li>This doesn't work if sndiod(8) is used and the subdevice is controlled
by MMC.
Indeed, sndiod(8) will try to resynchronize after the underrun and will
drop a huge amount of samples, corresponding to the duration of the pause.
</ul>
<h2><a name="section_8">8 Using sndio in multi-threaded programs</a></h2>
<p>
The sndio library can be safely used in multi-threaded programs as long
as all calls to function using the same handle are serialized.
This is achieved either with locks or by simply running all sndio
related bits in the same thread.
In any case, using multiple threads to handle audio I/O buys nothing since
the process is I/O bound.
<h2><a name="section_9">9 Windows-style callbacks</a></h2>
<p>
Certain programs expect to register a callback that will be invoked
automatically by the audio subsystem whenever the play buffer must be
filled.
For instance, Windows, jack and portaudio APIs use such semantics;
callbacks are tipically called by a real-time thread or in an interrupt
context.
This approach is equivalent to the read/write based approach,
which is widespread on Unix.
Consider the following callback-style pseudo-code:
<pre>
void
play_cb(void *buf, size_t buflen)
{
/* fill buf with data to play */
}
int
main(void)
{
register_audio_callback(play_cb);
...
wait_forever();
}
</pre>
It could be rewritten using read/write style semantics:
<pre>
void
play_cb(void *buf, size_t buflen)
{
/* fill buf with data to play */
}
int
main(void)
{
unsigned char *buf;
unsigned buflen = par.round;
...
for (;;) {
play_cb(buf, buflen);
sio_write(hdl, buf, buflen);
}
}
</pre>
there's no fundamental difference.
In other words, any callback style API could be exposed using sndio.
The only remaining problem is where to put the sndio loop.
<p>
If the program is single-threaded, then it uses a poll(2)-based event
loop, in which case non-blocking I/O should be used and the sndio bits
should be hooked somewhere in the poll(2) loop.
However, such programs probably come from the Unix world and don't use a
callback-style API.
<p>
If the program is multi-threaded, then it is simpler to spawn
a thread and run the simple loop from above in it.
The thread could be spawned when sio_start(3) is called and
terminated when sio_stop(3) is called; if so, the thread contains
real-time code paths only, and its scheduling priority could be cranked.
<p>
Multi-threaded programs use locks for synchronization, and
we don't want a thread to sleep while holding a lock.
To avoid holding a lock while a blocking sio_write(3) call is sleeping,
one can use non-blocking I/O and sleep in poll(2) without holding the
lock. In other words, sio_write(2) could be expanded as follows:
<pre>
unsigned char *p = buf;
struct pollfds pfds[MAXFDS];
...
pthread_mutex_lock(&amp;hdl_mtx);
...
for (;;) {
if (p - buf == buflen) {
play_cb(buf, buflen);
p = buf;
}
n = sio_pollfds(hdl, pfds);
pthread_mutex_unlock(&amp;hdl_mtx);
if (n &gt; 0 &amp;&amp; poll(pfds, n, -1) &lt; 0) {
pthread_mutex_lock(&amp;hdl_mtx);
continue;
}
pthread_mutex_lock(&amp;hdl_mtx);
if (sio_revents(hdl, pfds) &amp; POLLOUT) {
n = sio_write(hdl, p, buflen - (p - buf));
p += n;
}
}
</pre>
<h2><a name="section_10">10 Pitfalls</a></h2>
<h3><a name="section_10_1">10.1 Playback and record aren't independent</a></h3>
<p>
If for any reason a full-duplex program stops consuming recorded
data, there's a buffer overrun and recording stops. But since playback
and record direction are synchronous, this will also stop playback.
For instance, waiting for playback to drain without consuming recorded
data will never complete, because the record direction will pause
the stream because of the overrun. Deadlock occurs.
<h3><a name="section_10_2">10.2 Using appbufsz instead of bufsz</a></h3>
<p>
The ``appbufsz'' parameter is the size of the buffer the application
is responsible for keeping non-empty (playback) or non-full (record).
It should never be used for latency or buffer usage calculations.
<p>
The ``bufsz'' parameter is read-only and gives the total buffering
between the application and Joe's ears, i.e. it's the actual latency.
It takes into account any buffering including uncontrolled buffering
of network sockets.
<h3><a name="section_10_3">10.3 How many bytes to store a 24-bit sample</a></h3>
<p>
Short answer: four. Hardware, as most of the software, stores 24-bit
samples in 4-byte words. This format is often referred to as ``s24le''
or ``s24be'', and it's the default when 24-bit encodings are requested.
<p>
However, that's not always the case: .wav and .aiff files store 24-bit
samples in 3-byte words to save space. This encoding is often
referred to as ``s24le3'' or ``s24be3''.
If a program just reads and plays such files without any processing,
it's likely it will try to send the file contents on the audio stream
as-is. If so, the parameters should be set as follows:
<pre>
par.bits = 24;
par.bps = 3;
</pre>
<h3><a name="section_10_4">10.4 poll(2) not called fast enough</a></h3>
<p>
A (wrong) program may use the following approach. Consider the
following function to wait for the play buffer to become ready:
<pre>
void
wait_ready(void)
{
/*
* wait buffer to be consumed, sleep not to hog the CPU
*/
while (bufused &gt; threshold)
usleep(5);
}
</pre>
where the ``bufused'' variable is updated asynchronously by the
callback set with sio_onmove(3). Suppose it's then called as
follows:
<pre>
for (;;) {
prepare_data(some_data);
wait_ready();
sio_write(hdl, some_data, count);
}
</pre>
This will deadlock. The callback is invoked from sio_write(3), but
sio_write(3) is not called until ``bufused'' is updated by the
callback. The correct implementation uses poll(2) as follows; it's
also more efficient:
<pre>
void
wait_ready(void)
{
int nfds, revents;
struct pollfd pfds[1];
do {
nfds = sio_pollfd(hdl, pfds, POLLOUT);
if (nfds &gt; 0) {
if (poll(pfds, nfds, -1) &lt; 0)
err(1, "poll failed");
}
revents = sio_revents(hdl, pfds);
} while (!(revents &amp; POLLOUT));
}
</pre>
<h2><a name="section_11">11 Glossary</a></h2>
<p>
<dl>
<dt>channel
<dd>
that's a mono signal. Multiple channels form an audio stream. For example,
a stereo stream has two channels: left and right. Channels are
identified by small integers rather than names; so ``channel 0'' means
the ``left channel''.
<p>
channels numbers start from zero and are ordered as follows:
<p align="center">
<table>
<tr>
<th>channel number
<th>physical meaning
<tr>
<td>0
<td>main left
<tr>
<td>1
<td>main right
<tr>
<td>2
<td>rear left
<tr>
<td>3
<td>rear right
<tr>
<td>4
<td>center
<tr>
<td>5
<td>lfe
</table>
<p>
above, 0 is the origin, but that's arbitrary. The important
point is that ``main left'' is just before ``main right''.
This allows, for example, for the rear speakers to be viewed
as a stereo substream.
<dt>sample
<dd>
it's a scalar value representing ``the voltage'' on a given
channel. The signal is a sequence of samples. We represent
them as integers.
<dt>
frame
<dd>
the set of one sample for each channel of the stream. I.e a
sample for channel 0 followed by a sample for channel 1,
followed by ... a sample for the last channel. Frames are
numbered.
<dt>
rate
<dd>
the number of frames per second the streams carries,
e.g. 44.1kHz, 48kHz
<dt>
encoding
<dd>
The format with which samples are stored in memory. Example:
16-bit signed integer with little endian byte order. We
use the following abbreviations:
<ul>
<li>
``s'' or ``u'' character for signed or unsigned
<li>
followed by the number of bits
<li>
followed by ``le'' or ``be'' for little or big endian.
<li>
followed by the number of bytes in which the bits are stored.
<li>
followed by ``msb'' or ``lsb'' indicating how the significant
bits are aligned in the bytes.
</ul>
Example: s16, s24le, s24le3, s24le3lsb
<dt>
format
<dd>
The number of channels, the encoding, and the rate of the stream.
Example: stereo, s16le at 44.1kHz
<dt>
stream
<dd>
a (possibly bidirectional) connection between two endpoints like the
application and the sound card. Basically this is a sequence of
frames. If the stream is bidirectional, the two sequences are
synchronous: the played frame number N is played simultaneously while
the recorded frame number N is captured.
<dt>
full-duplex stream:
<dd>
bidirectional stream as described above.
<dt>
underrun
<dd>
played frames are buffered. If the producer (eg. the
application) doesn't provide frames fast enough, the
consumer (eg. the sound card) may end up without frames to
play. Thus it will play something else (because it can't
stop); often, it plays silence.
<dt>
overrun
<dd>
the recorded frames are buffered. If the consumer (eg. the
application) doesn't consume them fast enough, the producer
(eg. the sound card) may not be able to store newly recorded
frames in the buffer, thus it will discard them (because it
can't stop recording).
<dt>
xrun
<dd>
overrun or underrun. Note that on bidirectional streams, since
both directions are synchronous, if one of the directions xruns,
the error is present in the other direction as well.
For instance, if the play buffer underruns, recorded frames
during the underrun are lost.
</dl>
<p>
<small><i>
Copyright (c) 2008-2012 Alexandre Ratchov<br>
Last updated jan 8, 2016
</i></small>
</body>
</html>
Reference in New Issue View Git Blame Copy Permalink