xref: /illumos-gate/usr/src/man/man4i/dsp.4i (revision b8052df9f609edb713f6828c9eecc3d7be19dfb3)
1.\"  Copyright (c) 2009 Sun Microsystems, Inc. All rights reserved.
2.\" Copyright 2018, Joyent, Inc.
3.\" The contents of this file are subject to the terms of the
4.\" Common Development and Distribution License (the "License").
5.\" You may not use this file except in compliance with the License.
6.\"
7.\" You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
8.\" or http://www.opensolaris.org/os/licensing.
9.\" See the License for the specific language governing permissions
10.\" and limitations under the License.
11.\"
12.\" When distributing Covered Code, include this CDDL HEADER in each
13.\" file and include the License file at usr/src/OPENSOLARIS.LICENSE.
14.\" If applicable, add the following below this CDDL HEADER, with the
15.\" fields enclosed by brackets "[]" replaced with your own identifying
16.\" information: Portions Copyright [yyyy] [name of copyright owner]
17.Dd July 9, 2018
18.Dt DSP 4I
19.Os
20.Sh NAME
21.Nm dsp
22.Nd generic audio device interface
23.Sh SYNOPSIS
24.In sys/soundcard.h
25.Sh DESCRIPTION
26To record audio input, applications
27.Xr open 2
28the appropriate device and
29read data from it using the
30.Xr read 2
31system call.
32Similarly, sound data is queued to the audio output port by using the
33.Xr write 2
34system call.
35Device configuration is performed using the
36.Xr ioctl 2
37interface.
38.Pp
39Because some systems can contain more than one audio device, application
40writers are encouraged to open the
41.Pa /dev/mixer
42device and determine the
43physical devices present on the system using the
44.Dv SNDCTL_SYSINFO
45and
46.Dv SNDCTL_AUDIOINFO
47ioctls.
48See
49.Xr mixer 4I .
50The user should be provided
51wth the ability to select a different audio device, or alternatively, an
52environment variable such as
53.Ev AUDIODSP
54can be used.
55In the absence of any
56specific configuration from the user, the generic device file,
57.Pa /dev/dsp ,
58can be used.
59This normally points to a reasonably appropriate default audio
60device for the system.
61.Ss "Opening the Audio Device"
62The audio device is not treated as an exclusive resource.
63.Pp
64Each
65.Xr open 2
66completes as long as there are channels available to be allocated.
67If no channels are available to be allocated, the call returns
68.Sy -1
69with the
70.Va errno
71set to
72.Er EBUSY .
73.Pp
74Audio applications should explicitly set the encoding characteristics to match
75the audio data requirements after opening the device, and not depend on any
76default configuration.
77.Ss "Recording Audio Data"
78The
79.Xr read 2
80system call copies data from the system's buffers to the application.
81Ordinarily,
82.Xr read 2
83blocks until the user buffer is filled.
84The
85.Xr poll 2
86system call can be used to determine the presence of data
87that can be read without blocking.
88The device can alternatively be set to a
89non-blocking mode, in which case
90.Xr read 2
91completes immediately, but can return fewer bytes than requested.
92Refer to the
93.Xr read 2
94manual page for a complete description of this behavior.
95.Pp
96When the audio device is opened with read access, the device driver allocates
97resources for recording.
98Since this consumes system resources, processes that
99do not record audio data should open the device write-only
100.Pq Dv O_WRONLY .
101.Pp
102The recording process can be stopped by using the
103.Dv SNDCTL_DSP_HALT_INPUT
104ioctl, which also discards all pending record data in underlying device FIFOs.
105.Pp
106Before changing record parameters, the input should be stopped using the
107.Dv SNDCTL_DSP_HALT_INPUT
108ioctl, which also flushes the any underlying device input FIFOs.
109(This is not necessary if the process never started recording by calling
110.Xr read 2 .
111Otherwise, subsequent reads can return samples in the
112old format followed by samples in the new format.
113This is particularly
114important when new parameters result in a changed sample size.
115.Pp
116Input data can accumulate in device buffers very quickly.
117At a minimum, it accumulates at 8000 bytes per second for 8-bit, 8 KHz, mono,
118.Sy \(*m-Law
119data.
120If the device is configured for more channels, higher sample resolution, or
121higher sample rates, it accumulates even faster.
122If the application that consumes the data cannot keep up with this data rate,
123the underlying FIFOs can become full.
124When this occurs, any new incoming data is lost until the
125application makes room available by consuming data.
126Additionally, a record overrun is noted, which can be retrieved using the
127.Dv SNDCTL_DSP_GETERROR
128ioctl.
129.Pp
130Record volume for a stream can be adjusted by issuing the
131.Dv SNDCTL_DSP_SETRECVOL
132ioctl.
133The volume can also be retrieved using the
134.Dv SNDCTL_DSP_GETRECVOL .
135.Ss "Playing Audio Data"
136The
137.Xr write 1
138system call copies data from an application's buffer to the device output FIFO.
139Ordinarily,
140.Xr write 2
141blocks until the entire user
142buffer is transferred.
143The device can alternatively be set to a non-blocking
144mode, in which case
145.Xr write 2
146completes immediately, but might have
147transferred fewer bytes than requested.
148See
149.Xr write 2 .
150.Pp
151Although
152.Xr write 2
153returns when the data is successfully queued, the actual
154completion of audio output might take considerably longer.
155The
156.Dv SNDCTL_DSP_SYNC
157ioctl can be issued to allow an application to block
158until all of the queued output data has been played.
159.Pp
160The final
161.Xr close 2
162of the file descriptor waits until all of the audio output has drained.
163If a signal interrupts the
164.Xr close 2 ,
165or if the process
166exits without closing the device, any remaining data queued for audio output is
167flushed and the device is closed immediately.
168.Pp
169The output of playback data can be halted entirely, by calling the
170.Dv SNDCTL_DSP_HALT_OUTPUT
171ioctl.
172This also discards any data that is queued for playback in device FIFOs.
173.Pp
174Before changing playback parameters, the output should be drained using the
175.Dv SNDCTL_DSP_SYNC
176ioctl, and then stopped using the
177.Dv SNDCTL_DSP_HALT_OUTPUT
178ioctl, which also flushes the any underlying device output FIFOs.
179This is not necessary if the process never started playback, such as by calling
180.Xr write 2 .
181This is particularly important
182when new parameters result in a changed sample size.
183.Pp
184Output data is played from the playback buffers at a default rate of at least
1858000 bytes per second for \(*m-Law, A-Law or 8-bit PCM data (faster for 16-bit
186linear data or higher sampling rates).
187If the output FIFO becomes empty, the
188framework plays silence, resulting in audible stall or click in the output,
189until more data is supplied by the application.
190The condition is also noted as
191a play underrun, which can be determined using the
192.Dv SNDCTL_DSP_GETERROR
193ioctl.
194.Pp
195Playback volume for a stream can be adjusted by issuing the
196.Dv SNDCTL_DSP_SETPLAYVOL
197ioctl.
198The volume can also be retrieved using the
199.Dv SNDCTL_DSP_GETPLAYVOL .
200.Ss "Asynchronous I/O"
201The
202.Dv O_NONBLOCK
203flag can be set using the
204.Dv F_SETFL
205.Xr fcntl 2
206to enable non-blocking
207.Xr read 2
208and
209.Xr write 2
210requests.
211This is normally
212sufficient for applications to maintain an audio stream in the background.
213.Pp
214It is also possible to determine the amount of data that can be transferred for
215playback or recording without blocking using the
216.Dv SNDCTL_DSP_GETOSPACE
217or
218.Dv SNDCTL_DSP_GETISPACE
219ioctls, respectively.
220.Ss "Mixer Pseudo-Device"
221The
222.Pa /dev/mixer
223provides access to global hardware settings such as master volume settings, etc.
224It is also the interface used for determining the
225hardware configuration on the system.
226.Pp
227Applications should
228.Xr open 2
229.Pa /dev/mixer ,
230and use the
231.Dv SNDCTL_SYSINFO
232and
233.Dv SNDCTL_AUDIOINFO
234ioctls to determine the device node names of audio devices on the system.
235See
236.Xr mixer 4I
237for additional details.
238.Sh IOCTLS
239.Ss "Information IOCTLs"
240The following ioctls are supported on the audio device, as well as the mixer
241device.
242See
243.Xr mixer 4I
244for details.
245.Bd -literal -offset 2n
246.Dv OSS_GETVERSION
247.Dv SNDCTL_SYSINFO
248.Dv SNDCTL_AUDIOINFO
249.Dv SNDCTL_MIXERINFO
250.Dv SNDCTL_CARDINFO
251.Ed
252.Ss "Audio IOCTLs"
253The
254.Nm
255device supports the following ioctl commands:
256.Pp
257.\" A compact list is used so that items with multiple tags don't have
258.\" vertical whitespace.  However that means explicit .Pp statements must
259.\" be added.
260.Bl -tag -width "SNDCTL_DSP_CURRENT_IPTR" -compact
261.It Dv SNDCTL_DSP_SYNC
262The argument is ignored.
263This command suspends the calling process until the
264output FIFOs are empty and all queued samples have been played, or until a
265signal is delivered to the calling process.
266An implicit
267.Dv SNDCTL_DSP_SYNC
268is performed on the final
269.Xr close 2
270of the
271.Nm
272device.
273.Pp
274This ioctl should not be used unnecessarily, as if it is used in the middle of
275playback it causes a small click or pause, as the FIFOs are drained.
276The correct use of this ioctl is just before changing sample formats.
277.Pp
278.It Dv SNDCTL_DSP_HALT
279.It Dv SNDCTL_DSP_HALT_INPUT
280.It Dv SNDCTL_DSP_HALT_OUTPUT
281The argument is ignored.
282All input or output (or both) associated with the file
283is halted, and any pending data is discarded.
284.Pp
285.It Dv SNDCTL_DSP_SPEED
286The argument is a pointer to an integer, indicating the sample rate (in Hz) to
287be used.
288The rate applies to both input and output for the file descriptor.
289On return the actual rate, which can differ from that requested, is stored in
290the integer pointed to by the argument.
291To query the configured speed without
292changing it the value 0 can be used by the application.
293.Pp
294.It Dv SNDCTL_DSP_GETFMTS
295The argument is a pointer to an integer, which receives a bit mask of encodings
296supported by the device.
297Possible values are:
298.Pp
299.Bl -tag -width "AFMT_S24_PACKED" -compact -offset 2n
300.It Dv AFMT_MU_LAW
3018-bit unsigned \(*m-Law
302.It Dv AFMT_A_LAW
3038-bit unsigned a-Law
304.It Dv AFMT_U8
3058-bit unsigned linear PCM
306.It Dv AFMT_S16_LE
30716-bit signed little-endian linear PCM
308.It Dv AFMT_S16_BE
30916-bit signed big-endian linear PCM
310.It Dv AFMT_S16_NE
31116-bit signed native-endian linear PCM
312.It Dv AFMT_U16_LE
31316-bit unsigned little-endian linear PCM
314.It Dv AFMT_U16_BE
31516-bit unsigned big-endian linear PCM
316.It Dv AFMT_U16_NE
31716-bit unsigned native-endian linear PCM
318.It Dv AFMT_S24_LE
31924-bit signed little-endian linear PCM, 32-bit aligned
320.It Dv AFMT_S24_BE
32124-bit signed big-endian linear PCM, 32-bit aligned
322.It Dv AFMT_S24_NE
32324-bit signed native-endian linear PCM, 32-bit aligned
324.It Dv AFMT_S32_LE
32532-bit signed little-endian linear PCM
326.It Dv AFMT_S32_BE
32732-bit signed big-endian linear PCM
328.It Dv AFMT_S32_NE
32932-bit signed native-endian linear PCM
330.It Dv AFMT_S24_PACKED
33124-bit signed little-endian packed linear PCM
332.El
333.Pp
334Not all devices support all of these encodings.
335This implementation uses
336.Dv AFMT_S24_LE
337or
338.Dv AFMT_S24_BE ,
339whichever is native, internally.
340.Pp
341.It Dv SNDCTL_DSP_SETFMT
342The argument is a pointer to an integer, which indicates the encoding to be
343used.
344The same values as for
345.Dv SNDCTL_DSP_GETFMT
346can be used, but the caller can only specify a single option.
347The encoding is used for both input
348and output performed on the file descriptor.
349.Pp
350.It Dv SNDCTL_DSP_CHANNELS
351The argument is a pointer to an integer, indicating the number of channels to
352be used (1 for mono, 2 for stereo, etc.)
353The value applies to both input and output for the file descriptor.
354On return the actual channel configuration (which can differ from that
355requested) is stored in the integer pointed to by the argument.
356To query the configured channels without changing it the value 0
357can be used by the application.
358.Pp
359.It Dv SNDDCTL_DSP_GETCAPS
360The argument is a pointer to an integer bit mask, which indicates the
361capabilities of the device.
362The bits returned can include:
363.Pp
364.Bl -tag -width "PCM_CAP_DUPLEX" -compact -offset 2n
365.It Dv PCM_CAP_OUTPUT
366Device supports playback
367.It Dv PCM_CAP_INPUT
368Device supports recording
369.It Dv PCM_CAP_DUPLEX
370Device supports simultaneous playback and recording
371.El
372.Pp
373.It Dv SNDCTL_DSP_GETPLAYVOL
374.It Dv SNDCTL_DSP_GETRECVOL
375The argument is a pointer to an integer to receive the volume level for either
376playback or record.
377The value is encoded as a stereo value with the values for
378two channels in the least significant two bytes.
379The value for each channel thus has a range of 0-100.
380In this implementation, only the low order byte is
381used, as the value is treated as a monophonic value, but a stereo value (with
382both channel levels being identical) is returned for compatibility.
383.Pp
384.It Dv SNDCTL_DSP_SETPLAYVOL
385.It Dv SNDCTL_DSP_SETRECVOL
386The argument is a pointer to an integer indicating volume level for either
387playback or record.
388The value is encoded as a stereo value with the values for
389two channels in the least significant two bytes.
390The value for each channel has a range of 0-100.
391Note that in this implementation, only the low order byte is
392used, as the value is treated as a monophonic value.
393Portable applications should assign the same value to both bytes.
394.Pp
395.It Dv SNDCTL_DSP_GETISPACE
396.It Dv SNDCTL_DSP_GETOSPACE
397The argument is a pointer to a
398.Vt struct audio_buf_info ,
399which has the following structure:
400.Bd -literal -offset 2n
401typedef struct audio_buf_info {
402   int fragments; /* # of available fragments */
403   int fragstotal;
404        /* Total # of fragments allocated */
405   int fragsize;
406        /* Size of a fragment in bytes */
407   int bytes;
408       /* Available space in bytes */
409   /*
410    * Note! 'bytes' could be more than
411    * fragments*fragsize
412    */
413} audio_buf_info;
414.Ed
415.Pp
416The fields
417.Fa fragments ,
418.Fa fragstotal ,
419and
420.Fa fragsize
421are intended for use with compatible applications (and in the future with
422.Xr mmap 2 )
423only, and need not be used by typical applications.
424On successful return the bytes member
425contains the number of bytes that can be transferred without blocking.
426.Pp
427.It Dv SNDCTL_DSP_CURRENT_IPTR
428.It Dv SNDCTL_DSP_CURRENT_OPTR
429The argument is a pointer to an
430.Vt oss_count_t ,
431which has the following definition:
432.Bd -literal -offset 2n
433typedef struct {
434    long long samples;
435       /* Total # of samples */
436    int fifo_samples;
437       /* Samples in device FIFO */
438    int filler[32]; /* For future use */
439} oss_count_t;
440.Ed
441.Pp
442The
443.Fa samples
444field contains the total number of samples transferred by the
445device so far.
446The
447.Fa fifo_samples
448is the depth of any hardware FIFO.
449This structure can be useful for accurate stream positioning and latency
450calculations.
451.Pp
452.It Dv SNDCTL_DSP_GETIPTR
453.It Dv SNDCTL_DSP_GETOPTR
454The argument is a pointer to a
455.Vt struct count_info ,
456which has the following definition:
457.Bd -literal -offset 2n
458typedef struct count_info {
459    unsigned int bytes;
460      /* Total # of bytes processed */
461    int blocks;
462      /*
463       * # of fragment transitions since
464       * last time
465       */
466    int ptr; /* Current DMA pointer value */
467} count_info;
468.Ed
469.Pp
470These ioctls are primarily supplied for compatibility, and should not be used
471by most applications.
472.Pp
473.It Dv SNDCTL_DSP_GETODELAY
474The argument is a pointer to an integer.
475On return, the integer contains the
476number of bytes still to be played before the next byte written are played.
477This can be used for accurate determination of device latency.
478The result can differ from actual value by up the depth of the internal device
479FIFO, which is typically 64 bytes.
480.Pp
481.It Dv SNDCTL_DSP_GETERROR
482The argument is a pointer to a
483.Vt struct audio_errinfo ,
484defined as follows:
485.Bd -literal -offset 2n
486typedef struct audio_errinfo {
487     int play_underruns;
488     int rec_overruns;
489     unsigned int play_ptradjust;
490     unsigned int rec_ptradjust;
491     int play_errorcount;
492     int rec_errorcount;
493     int play_lasterror;
494     int rec_lasterror;
495     int play_errorparm;
496     int rec_errorparm;
497     int filler[16];
498} audio_errinfo;
499.Ed
500.Pp
501For this implementation, only the
502.Fa play_underruns
503and
504.Fa rec_overruns
505values are significant.
506No other fields are used in this implementation.
507.Pp
508These fields are reset to zero each time their value is retrieved using this
509ioctl.
510.El
511.Ss "Compatibility IOCTLS"
512These ioctls are supplied exclusively for compatibility with existing
513applications.
514Their use is not recommended, and they are not documented here.
515Many of these are implemented as simple no-ops.
516.Pp
517.Bl -tag -width "Dv" -offset 2n -compact
518.It Dv SNDCTL_DSP_POST
519.It Dv SNDCTL_DSP_STEREO
520.It Dv SNDCTL_DSP_SETDUPLEX
521.It Dv SNDCTL_DSP_LOW_WATER
522.It Dv SNDCTL_DSP_PROFILE
523.It Dv SNDCTL_DSP_GETBLKSIZE
524.It Dv SNDCTL_DSP_SUBDIVIDE
525.It Dv SNDCTL_DSP_SETFRAGMENT
526.It Dv SNDCTL_DSP_COOKEDMODE
527.It Dv SNDCTL_DSP_READCTL
528.It Dv SNDCTL_DSP_WRITECTL
529.It Dv SNDCTL_DSP_SILENCE
530.It Dv SNDCTL_DSP_SKIP
531.It Dv SNDCTL_DSP_POST
532.It Dv SNDCTL_DSP_GET_RECSRC
533.It Dv SNDCTL_DSP_SET_RECSRC
534.It Dv SNDCTL_DSP_SET_RECSRC_NAMES
535.It Dv SNDCTL_DSP_GET_PLAYTGT
536.It Dv SNDCTL_DSP_SET_PLAYTGT
537.It Dv SNDCTL_DSP_SET_PLAYTGT_NAMES
538.It Dv SNDCTL_DSP_GETTRIGGER
539.It Dv SNDCTL_DSP_SETTRIGGER
540.It Dv SNDCTL_AUDIOINFO_EX
541.It Dv SNDCTL_ENGINEINFO
542.El
543.Sh FILES
544The physical audio device names are system dependent and are rarely used by
545programmers.
546Programmers should use the generic device names listed below.
547.Bl -tag -width "/usr/share/audio/samples"
548.It Pa /dev/dsp
549Symbolic link to the system's primary audio device
550.It Pa /dev/mixer
551Symbolic link to the pseudo mixer device for the system
552.It Pa /dev/sndstat
553Symbolic link to the pseudo mixer device for the system
554.It Pa /usr/share/audio/samples
555Audio files
556.El
557.Sh ERRORS
558An
559.Xr open 2
560call fails if:
561.Bl -tag -width "EINVAL"
562.It Er EBUSY
563The requested play or record access is busy and either the
564.Dv O_NDELAY
565or
566.Dv O_NONBLOCK
567flag was set in the
568.Xr open 2
569request.
570.It Er EINTR
571The requested play or record access is busy and a signal interrupted the
572.Xr open 2
573request.
574.It Er EINVAL
575The device cannot support the requested play or record access.
576.El
577.Pp
578An
579.Xr ioctl 2
580call fails if:
581.Bl -tag -width "EINVAL"
582.It Er EINVAL
583The parameter changes requested in the ioctl are invalid or are not supported
584by the device.
585.El
586.Sh ARCHITECTURE
587SPARC
588X86
589.Sh INTERFACE STABILITY
590Uncommitted
591.Sh SEE ALSO
592.Xr close 2 ,
593.Xr fcntl 2 ,
594.Xr ioctl 2 ,
595.Xr mmap 2 ,
596.Xr open 2 ,
597.Xr poll 2 ,
598.Xr read 2 ,
599.Xr write 2 ,
600.Xr audio 4D ,
601.Xr mixer 4I ,
602.Xr attributes 7
603