NetBSD/sys/arch/x68k/dev/bsd_audio.c

1424 lines
31 KiB
C

/* $NetBSD: bsd_audio.c,v 1.3 1996/10/13 03:34:40 christos Exp $ */
/*
* Copyright (c) 1991-1993 Regents of the University of California.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the Computer Systems
* Engineering Group at Lawrence Berkeley Laboratory.
* 4. Neither the name of the University nor of the Laboratory may be used
* to endorse or promote products derived from this software without
* specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
/*
* This is a (partially) SunOS-compatible /dev/audio driver for NetBSD.
*
* This code assumes SoundBlaster type hardware, supported by the
* code in isa/sb.c. A major problem with this hardware is that it
* is half-duplex. E.g., you cannot simultaneously record and play
* samples. Thus, it doesn't really make sense to allow O_RDWR access.
* However, opening and closing the device to "turn around the line"
* is relatively expensive and costs a card reset (which can take
* some time). Instead, we allow O_RDWR access, and provide an
* ioctl to set the "mode", e.g., playing or recording. If you
* write to the device in record mode, the data is tossed. If you
* read from the device in play mode, you get zero filled buffers
* at the rate at which samples are naturally generated.
*/
#include "audio.h"
#if NAUDIO > 0
#include <sys/param.h>
#include <sys/ioctl.h>
#include <sys/systm.h>
#include <sys/vnode.h>
#include <sys/select.h>
#include <sys/malloc.h>
#include <sys/file.h>
#include <sys/proc.h>
#include <sys/user.h>
#include <sys/device.h>
#include <machine/cpu.h>
#include <machine/pte.h>
extern u_int kvtop __P((register caddr_t addr));
int uiomove __P((caddr_t cp, int n, struct uio *uio));
#include <x68k/dev/bsd_audiovar.h>
#include <x68k/dev/bsd_audioreg.h>
#include <machine/bsd_audioio.h>
#include <x68k/x68k/iodevice.h>
#define AUDIODEBUG if (audiodebug) printf
int audiodebug = 0;
#define dma3 (IODEVbase->io_dma[3])
/*
* Initial/default block size is patchable.
*/
int audio_blocksize = DEFBLKSIZE;
int audio_backlog = 400; /* 50ms in samples */
/*
* Software state, per MSM6258V audio chip.
*/
struct audio_softc {
struct adpcm_softc sc_adpcm;
u_char sc_open; /* single use device */
u_char sc_mode; /* */
u_char sc_rbus; /* input dma in progress */
u_char sc_pbus; /* output dma in progress */
u_char sc_rencoding;
u_char sc_pencoding;
u_char sc_pad[2];
u_long sc_wseek; /* timestamp of last frame written */
u_long sc_rseek; /* timestamp of last frame read */
u_long sc_orate; /* input sampling rate */
u_long sc_irate; /* output sampling rate */
struct selinfo sc_wsel; /* write selector */
struct selinfo sc_rsel; /* read selector */
int sc_rlevel; /* record level */
int sc_plevel; /* play level */
/*
* Sleep channels for reading and writing.
*/
int sc_rchan;
int sc_wchan;
int sc_ochan;
/*
* Buffer management.
*/
u_char *sc_hp; /* head */
u_char *sc_tp; /* tail */
u_char *sc_bp; /* start of buffer */
u_char *sc_ep; /* end of buffer */
u_char *sc_zp; /* block of silence */
int sc_nblk;
int sc_maxblk;
int sc_lowat; /* xmit low water mark (for wakeup) */
int sc_hiwat; /* xmit high water mark (for wakeup) */
int sc_blksize; /* recv block (chunk) size */
int sc_backlog; /* # blks of xmit backlog to gen. */
int sc_finish;
/* sc_au is special in that the hardware interrupt handler uses it */
int sc_rblks; /* number of phantom record blocks */
} audio_softc[NAUDIO];
/* forward declarations */
inline static int audio_sleep __P((int *));
/* autoconfiguration driver */
int audioattach();
static int audio_default_level = 150;
static void ausetrgain __P((struct audio_softc *, int));
static void ausetpgain __P((struct audio_softc *, int));
static int audiosetinfo __P((struct audio_softc *, struct audio_info *));
static int audiogetinfo __P((struct audio_softc *, struct audio_info *));
struct sun_audio_info;
void audio_init_record __P((struct audio_softc *));
void audio_init_play __P((struct audio_softc *));
void audiostartr __P((struct audio_softc *));
void audiostartp __P((struct audio_softc *));
inline void audio_rint __P((struct audio_softc *));
inline void audio_pint __P((struct audio_softc *));
void audio_tomulaw __P((short *, register int));
void audio_frommulaw __P((u_char *, register int));
void audio_tolinear __P((struct audio_softc *, register u_char *, register int));
void audio_fromlinear __P((struct audio_softc *, register u_char *, register int));
void
audio_initbuf(sc)
struct audio_softc *sc;
{
register int nblk = NBPG / sc->sc_blksize;
sc->sc_ep = sc->sc_bp + nblk * sc->sc_blksize;
sc->sc_hp = sc->sc_tp = sc->sc_bp;
sc->sc_maxblk = nblk;
sc->sc_nblk = 0;
sc->sc_lowat = 3;
sc->sc_hiwat = nblk - 1;
}
static inline int
audio_sleep(int *chan)
{
int st;
*chan = 1;
st = (tsleep((caddr_t)chan, PWAIT | PCATCH, "audio", 0));
*chan = 0;
return (st);
}
static inline void
audio_wakeup(chan)
int *chan;
{
if (*chan) {
wakeup((caddr_t)chan);
*chan = 0;
}
}
/*XXX*/
int auzero[NBPG/4];
short transbuf[NBPG*2];
u_char transbuf2[NBPG];
int audiomatch __P((struct device *, void *, void *));
void audioattach __P((struct device *, struct device *, void *));
struct cfattach audio_ca = {
sizeof(struct audio_softc), audiomatch, audioattach
};
struct cfdriver audio_cd = {
NULL, "audio", DV_DULL
};
int
audiomatch(parent, match, aux)
struct device *parent;
void *match, *aux;
{
struct cfdata *cf = match;
if (strcmp(aux, "adpcm") || cf->cf_unit > 0)
return 0;
return 1;
}
void
audioattach(parent, self, aux)
struct device *parent;
struct device *self;
void *aux;
{
struct dmac *dmac = &IODEVbase->io_dma[3];
register struct audio_softc *sc = &audio_softc[0];
dmac->csr = 0xff;
dmac->dcr = 0x80;
dmac->ocr = 0x32;
dmac->scr = 0x04;
dmac->mfc = 0x05;
dmac->dfc = 0x05;
dmac->bfc = 0x05;
dmac->dar = (unsigned long)kvtop(&(adpcm.data));
dmac->niv = 0x6a;
dmac->eiv = 0x6b;
sc->sc_rbus = 0;
sc->sc_pbus = 0;
/*printf("audio0: MSM6258V ADPCM chip.\n");*/
}
int
audioopen(dev, flags, ifmt, p)
dev_t dev;
int flags;
int ifmt;
struct proc *p;
{
register struct audio_softc *sc = &audio_softc[0];
int s, i;
int unit = minor(dev);
int error;
AUDIODEBUG("audio: open\n");
if((unit & 0x0000003f) >= NAUDIO)
return (ENXIO);
if (sc->sc_open != 0)
return (EBUSY);
/*
* Allocate a single page so it won't cross a page boundary.
* This way the dma carried out in the sb module will be efficient
* (i.e., at_dma() won't have to make a copy)
*/
sc->sc_zp = malloc(NBPG, M_DEVBUF, M_WAITOK);
if (sc->sc_zp == 0)
goto nobufs;
sc->sc_bp = malloc(NBPG, M_DEVBUF, M_WAITOK);
if (sc->sc_bp == 0) {
free(sc->sc_zp, M_DEVBUF);
goto nobufs;
}
sc->sc_blksize = audio_blocksize;
sc->sc_backlog = audio_backlog;
audio_initbuf(sc);
/* nothing read or written yet */
sc->sc_rseek = 0;
sc->sc_wseek = 0;
sc->sc_rchan = 0;
sc->sc_wchan = 0;
sc->sc_adpcm.sc_amp = 0;
sc->sc_adpcm.sc_estim = 0;
/*
* Here's a hack: do ulaw conversion if 6-7 bit of
* minor device is set. That way, we can have /dev/audio
* (minor 0x80) do ulaw conversion, and /dev/sound or
* whatever, do adpcm.
*/
switch (minor(dev) >> 6) {
case 0x00:
/* /dev/adpcm */
AUDIODEBUG("audioopen: adpcm mode\n");
sc->sc_pencoding = sc->sc_rencoding = AUDIO_ENCODING_ADPCM;
sc->sc_orate = 15625;
sc->sc_irate = 15625;
for (i = NBPG / 4; --i >= 0; ) {
((u_long *)sc->sc_zp)[i] = 0x88008800;
auzero[i] = 0x88008800;
}
break;
case 0x01:
/* /dev/sound */
AUDIODEBUG("audioopen: linear mode\n");
sc->sc_pencoding = sc->sc_rencoding = AUDIO_ENCODING_LINEAR;
sc->sc_orate = 15625;
sc->sc_irate = 15625;
for (i = NBPG / 4; --i >= 0; ) {
((u_long *)sc->sc_zp)[i] = 0x88008800;
auzero[i] = 0x88008800;
}
break;
case 0x02:
/* /dev/audio */
AUDIODEBUG("audioopen: ulaw mode\n");
sc->sc_pencoding = sc->sc_rencoding = AUDIO_ENCODING_ULAW;
sc->sc_orate = 7813;
sc->sc_irate = 7813;
for (i = NBPG / 4; --i >= 0; ) {
((u_long *)sc->sc_zp)[i] = 0x7f7f7f7f;
auzero[i] = 0x80808080;
}
break;
case 0x03:
/* /dev/audio */
AUDIODEBUG("audioopen: ulaw mode\n");
sc->sc_pencoding = sc->sc_rencoding = AUDIO_ENCODING_ULAW;
sc->sc_orate = 7813;
sc->sc_irate = 7813;
for (i = NBPG / 4; --i >= 0; ) {
((u_long *)sc->sc_zp)[i] = 0x7f7f7f7f;
auzero[i] = 0x80808080;
}
break;
}
#if 0
ausetrgain(sc, audio_default_level);
ausetpgain(sc, audio_default_level);
#endif
/* XXX */
s = splaudio();
if ((sc->sc_pbus == 1) || (sc->sc_rbus == 1))
error = audio_sleep(&sc->sc_ochan);
sc->sc_open = 1;
if ((flags & FREAD) != 0) {
audio_init_record(sc);
} else {
audio_init_play(sc);
}
splx(s);
return (0);
nobufs:
sc->sc_open = 0;
return (ENOBUFS);
}
/*
* Must be called from task context.
*/
void
audio_init_record(sc)
struct audio_softc *sc;
{
register int s;
s = splaudio();
dma3.csr = 0xff;
dma3.ocr = 0xb2;
sc->sc_mode = AUMODE_RECORD;
sc->sc_rblks = 0;
adpcm.stat = ADPCM_CMD_STOP;
adpcm_set_sr(sc->sc_irate);
splx(s);
}
/*
* Must be called from task context.
*/
void
audio_init_play(sc)
struct audio_softc *sc;
{
register int s;
s = splaudio();
dma3.csr = 0xff;
dma3.ocr = 0x32;
sc->sc_mode = AUMODE_PLAY;
sc->sc_rblks = 0;
adpcm.stat = ADPCM_CMD_STOP;
adpcm_set_sr(sc->sc_orate);
splx(s);
}
static int
audio_drain(sc)
register struct audio_softc *sc;
{
register int error;
while (sc->sc_nblk > 0) {
error = audio_sleep(&sc->sc_wchan);
if (error != 0) {
AUDIODEBUG("audio: Interrupted?\n");
return (error);
}
}
return (0);
}
/*
* Close an audio chip.
*/
/* ARGSUSED */
int
audioclose(dev, flags, ifmt, p)
dev_t dev;
int flags;
int ifmt;
struct proc *p;
{
register struct audio_softc *sc =&audio_softc[0];
register int s;
AUDIODEBUG("audio: close\n");
/*
* Block until output drains, but allow ^C interrupt.
*/
sc->sc_lowat = 0; /* avoid excessive wakeups */
s = splaudio();
/*
* If there is pending output, let it drain (unless
* the output is paused).
*/
if (sc->sc_pbus && sc->sc_nblk > 0)
(void)audio_drain(sc);
splx(s);
free(sc->sc_bp, M_DEVBUF);
free(sc->sc_zp, M_DEVBUF);
sc->sc_open = 0;
return (0);
}
int
audioread(dev, uio, ioflag)
dev_t dev;
struct uio *uio;
int ioflag;
{
register struct audio_softc *sc = &audio_softc[0];
register u_char *hp;
register int blocksize = sc->sc_blksize;
register int error, error2, s;
AUDIODEBUG("audio: read\n");
if (uio->uio_resid == 0)
return (0);
if (uio->uio_resid < blocksize)
return (EINVAL);
if (sc->sc_mode == AUMODE_PLAY) {
/*
* If we're in play mode, return silence blocks
* based on the number of blocks we have output.
*/
do {
s = splaudio();
while (sc->sc_rblks <= 0) {
if (ioflag & IO_NDELAY) {
splx(s);
return (EWOULDBLOCK);
}
error = audio_sleep(&sc->sc_rchan);
if (error != 0) {
splx(s);
return (error);
}
}
splx(s);
/*XXX handle ulaw 0 */
error = uiomove(sc->sc_zp, blocksize, uio);
if (error)
break;
--sc->sc_rblks;
} while (uio->uio_resid >= blocksize);
return (error);
}
error = error2 = 0;
do {
while ((sc->sc_nblk <= 0) && (error2 == 0)){
if (ioflag & IO_NDELAY) {
error = EWOULDBLOCK;
return (error);
}
s = splaudio();
if (!sc->sc_rbus)
audiostartr(sc);
error2 = error = audio_sleep(&sc->sc_rchan);
splx(s);
if (error != 0) {
AUDIODEBUG("audio: read Interrupted?\n");
if (sc->sc_rbus == 1){
s = splaudio();
dma3.csr = 0xff;
dma3.ccr = 0x08;
sc->sc_finish = 1;
splx(s);
audio_sleep(&sc->sc_rchan);
} else {
return (error);
}
}
}
hp = sc->sc_hp;
switch (sc->sc_rencoding) {
case AUDIO_ENCODING_ULAW:
audio_tolinear(sc, hp, blocksize);
audio_tomulaw(transbuf, blocksize*2);
error = uiomove((u_char *)transbuf2, blocksize*2, uio);
break;
case AUDIO_ENCODING_LINEAR:
audio_tolinear(sc, hp, blocksize);
error = uiomove((u_char *)transbuf, blocksize*4, uio);
break;
case AUDIO_ENCODING_ADPCM:
error = uiomove(hp, blocksize, uio);
break;
}
if (error) {
printf("audio: uiomove failed\n");
break;
}
hp += blocksize;
if (hp >= sc->sc_ep)
hp = sc->sc_bp;
sc->sc_hp = hp;
--sc->sc_nblk;
} while (uio->uio_resid >= blocksize);
return (error ? error : error2);
}
void
audio_clear(sc)
struct audio_softc *sc;
{
register int s = splaudio();
if (sc->sc_rbus || sc->sc_pbus) {
dma3.ccr = 0x10;
sc->sc_rbus = 0;
sc->sc_pbus = 0;
}
sc->sc_nblk = 0;
sc->sc_hp = sc->sc_tp = sc->sc_bp;
splx(s);
}
int
audiowrite(dev, uio, ioflag)
dev_t dev;
struct uio *uio;
int ioflag;
{
register struct audio_softc *sc;
register u_char *tp;
register int error, s, cc;
register int blocksize;
AUDIODEBUG("audio: write\n");
sc = &audio_softc[0];
blocksize = sc->sc_blksize;
/*
* If currently recording, throw away data.
*/
if (sc->sc_mode != AUMODE_PLAY) {
uio->uio_offset += uio->uio_resid;
uio->uio_resid = 0;
return (0);
}
error = 0;
while (uio->uio_resid > 0) {
register int watermark = sc->sc_hiwat;
s = splaudio();
while (sc->sc_nblk > watermark) {
if (ioflag & IO_NDELAY) {
splx(s);
error = EWOULDBLOCK;
printf("audiowrite: ioflag=%x\n", ioflag);
return (error);
}
error = audio_sleep(&sc->sc_wchan);
if (error != 0) {
splx(s);
AUDIODEBUG("audiowrite: Interrupted?\n");
return (error);
}
watermark = sc->sc_lowat;
}
splx(s);
tp = sc->sc_tp;
cc = uio->uio_resid;
switch (sc->sc_pencoding) {
case AUDIO_ENCODING_ULAW:
if (cc < blocksize*2) {
error = uiomove((u_char *)transbuf2, cc, uio);
if (error){
printf("audio: uiomove failed\n");
break;
}
AUDIODEBUG("audiowrite: zero suppress(%x,%x,%d)\n",transbuf, cc, blocksize*4 - cc);
bcopy((char *)auzero, (u_char *)transbuf2 + cc, blocksize*2 - cc);
} else {
error = uiomove((u_char *)transbuf2, blocksize*2, uio);
if (error) {
printf("audio: uiomove failed\n");
break;
}
}
break;
case AUDIO_ENCODING_ADPCM:
if (cc < blocksize) {
error = uiomove(tp, cc, uio);
if (error){
printf("audio: uiomove failed\n");
break;
}
bcopy((char *)auzero, tp + cc, blocksize - cc);
} else {
error = uiomove(tp, blocksize, uio);
if (error) {
printf("audio: uiomove failed\n");
break;
}
}
break;
case AUDIO_ENCODING_LINEAR:
if (cc < blocksize*4) {
error = uiomove((u_char *)transbuf, cc, uio);
if (error){
printf("audio: uiomove failed\n");
break;
}
AUDIODEBUG("audiowrite: zero suppress(%x,%x,%d)\n",transbuf, cc, blocksize*4 - cc);
bzero((char *)transbuf + cc, blocksize*4 - cc);
} else {
error = uiomove((u_char *)transbuf, blocksize*4, uio);
if (error) {
printf("audio: uiomove failed\n");
break;
}
}
break;
}
switch (sc->sc_pencoding) {
case AUDIO_ENCODING_ULAW:
audio_frommulaw(transbuf2, blocksize*2);
audio_fromlinear(sc, tp, blocksize*2);
break;
case AUDIO_ENCODING_LINEAR:
audio_fromlinear(sc, tp, blocksize*2);
break;
case AUDIO_ENCODING_ADPCM:
break;
}
tp += blocksize;
if (tp >= sc->sc_ep)
tp = sc->sc_bp;
sc->sc_tp = tp;
++sc->sc_nblk;
sc->sc_finish = 0;
/*
* If output isn't active, start it up.
*/
s = splaudio();
if (sc->sc_pbus == 0)
audiostartp(sc);
splx(s);
}
AUDIODEBUG("audiowrite: exit\n");
return (error);
}
/* Sun audio compatibility */
struct sun_audio_prinfo {
u_int sample_rate;
u_int channels;
u_int precision;
u_int encoding;
u_int gain;
u_int port;
u_int reserved0[4];
u_int samples;
u_int eof;
u_char pause;
u_char error;
u_char waiting;
u_char balance;
u_short minordev;
u_char open;
u_char active;
};
struct sun_audio_info {
struct sun_audio_prinfo play;
struct sun_audio_prinfo record;
u_int monitor_gain;
u_int reserved[4];
};
int
audioioctl(dev, cmd, addr, flag, p)
dev_t dev;
u_long cmd;
caddr_t addr;
int flag;
struct proc *p;
{
register struct audio_softc *sc = &audio_softc[0];
int error = 0, s;
AUDIODEBUG("audio: ioctl(0x%x)\n", cmd);
switch (cmd) {
#if 0
case AUDIO_GETMAP:
bcopy((caddr_t)&sc->sc_map, addr, sizeof(sc->sc_map));
break;
case AUDIO_SETMAP:
bcopy(addr, (caddr_t)&sc->sc_map, sizeof(sc->sc_map));
sc->sc_map.mr_mmr2 &= 0x7f;
audio_setmap(sc->sc_au.au_msm, &sc->sc_map);
break;
case AUDIO_FLUSH:
sc->sc_wseek = 0;
sc->sc_rseek = 0;
break;
/*
* Number of read samples dropped. We don't know where or
* when they were dropped.
*/
case AUDIO_RERROR:
*(int *)addr = sc->sc_au.au_rb.cb_drops != 0;
break;
/*
* How many samples will elapse until mike hears the first
* sample of what we last wrote?
*/
case AUDIO_WSEEK:
s = splaudio();
*(u_long *)addr = sc->sc_wseek - sc->sc_au.au_stamp
+ AUCB_LEN(&sc->sc_au.au_rb);
splx(s);
break;
#endif
case AUDIO_SETINFO:
error = audiosetinfo(sc, (struct audio_info *)addr);
break;
case AUDIO_GETINFO:
error = audiogetinfo(sc, (struct audio_info *)addr);
break;
case AUDIO_DRAIN:
error = audio_drain(sc);
break;
default:
error = EINVAL;
break;
}
return (error);
}
/* ARGSUSED */
int
audioselect(dev, rw, p)
dev_t dev;
int rw;
struct proc *p;
{
register struct audio_softc *sc = &audio_softc[0];
register int s = splaudio();
switch (rw) {
case FREAD:
if (sc->sc_mode == AUMODE_PLAY) {
if (sc->sc_rblks > 0) {
splx(s);
return (1);
}
} else if (sc->sc_nblk > 0) {
splx(s);
return (1);
}
selrecord(p, &sc->sc_rsel);
break;
case FWRITE:
/*
* Can write if we're recording because it gets preempted.
* Otherwise, can write when below low water.
* XXX this won't work right if we're in
* record mode -- we need to note that a write
* select has happed and flip the speaker.
*/
if (sc->sc_mode != AUMODE_PLAY ||
sc->sc_nblk < sc->sc_lowat) {
splx(s);
return (1);
}
selrecord(p, &sc->sc_wsel);
break;
}
splx(s);
return (0);
}
static inline void
audio_dmastart(read, addr, count)
int read;
u_char *addr;
int count;
{
dma3.csr = 0xff;
dma3.mtc = (u_short)count;
asm("nop");
asm("nop");
dma3.mar = (u_long)kvtop(addr);
#if defined(M68040)
/*
* Push back dirty cache lines
*/
if (mmutype == MMU_68040)
DCFP(kvtop(addr));
#endif
adpcm.stat = read ? ADPCM_CMD_PLAY : ADPCM_CMD_REC;
dma3.ccr = 0x88;
}
void
audiostartr(sc)
struct audio_softc *sc;
{
register u_char *tp = sc->sc_tp;
register int cc = sc->sc_blksize;
AUDIODEBUG("audio: startr\n");
audio_dmastart(0, tp, cc);
sc->sc_rbus = 1;
}
void
audiostartp(sc)
struct audio_softc *sc;
{
audio_dmastart(1, sc->sc_hp, sc->sc_blksize);
sc->sc_pbus = 1;
}
void
audiointr()
{
register struct audio_softc *sc = &audio_softc[0];
dma3.csr = 0xff;
PCIA(); /* XXX? by oki */
if (sc->sc_pbus == 1){
audio_pint(sc);
} else if (sc->sc_rbus == 1){
audio_rint(sc);
} else {
printf("audiointr: sc_pbus == sc_rbus == 0. Why interrupt?\n");
}
if (sc->sc_open == 0) {
audio_wakeup(&sc->sc_ochan);
}
}
void
audioerrintr()
{
register struct audio_softc *sc = &audio_softc[0];
printf("audioerrintr: software abort?\ncsr=%x, cer=%x\n pbus = %d, rbus = %d\n", dma3.csr, dma3.cer, sc->sc_pbus, sc->sc_rbus);
dma3.csr = 0xff;
if (sc->sc_pbus == 1){
audio_pint(sc);
} else if (sc->sc_rbus == 1){
audio_rint(sc);
} else {
printf("audioerrintr: sc_pbus == sc_rbus == 0. Why interrupt?\n");
}
if (sc->sc_open == 0) {
audio_wakeup(&sc->sc_ochan);
}
}
inline void
audio_pint(sc)
struct audio_softc *sc;
{
register u_char *hp = sc->sc_hp;
register int cc = sc->sc_blksize;
register int s;
AUDIODEBUG("audio: pint sc_nblk %d\n", sc->sc_nblk);
s = splaudio();
if (sc->sc_finish == 1) {
adpcm.stat = ADPCM_CMD_STOP;
sc->sc_pbus = 0;
} else {
--sc->sc_nblk;
hp = sc->sc_hp;
hp += cc;
if (hp >= sc->sc_ep)
hp = sc->sc_bp;
sc->sc_hp = hp;
if (sc->sc_nblk > 0) {
audio_dmastart(1, hp, cc);
sc->sc_finish = 0;
} else {
audio_dmastart(1, (char *)&auzero, cc);
sc->sc_finish = 1;
}
}
splx(s);
++sc->sc_rblks;
if (sc->sc_mode == AUMODE_PLAY) {
if (sc->sc_nblk <= sc->sc_lowat) {
audio_wakeup(&sc->sc_wchan);
selwakeup(&sc->sc_wsel);
}
}
}
/*
* Called from sb module on completion of dma input.
* Copy the input frame into the ring buffer at the
* current position. Do a wakeup if necessary.
*/
void
audio_rint(sc)
struct audio_softc *sc;
{
register u_char *tp = sc->sc_tp;
register int cc = sc->sc_blksize;
tp = sc->sc_tp;
tp += cc;
if (tp >= sc->sc_ep)
tp = sc->sc_bp;
if (++sc->sc_nblk < sc->sc_maxblk)
audio_dmastart(0, tp, cc);
else {
adpcm.stat = ADPCM_CMD_STOP;
sc->sc_rbus = 0;
}
sc->sc_tp = tp;
audio_wakeup(&sc->sc_rchan);
selwakeup(&sc->sc_rsel);
}
static void
ausetrgain(sc, level)
register struct audio_softc *sc;
register int level;
{
#ifdef x68k
/* XXX */
#endif
}
/*
* XXX Looks like we need a pro to do volume control...
*/
static void
ausetpgain(sc, level)
register struct audio_softc *sc;
register int level;
{
#ifdef x68k
/* XXX */
#endif
}
static int
audiosetinfo(sc, ai)
struct audio_softc *sc;
struct audio_info *ai;
{
struct audio_prinfo *r = &ai->record, *p = &ai->play;
register int cleared = 0;
register int s, bsize;
if (p->gain != ~0)
ausetpgain(sc, p->gain);
if (r->gain != ~0)
ausetrgain(sc, r->gain);
if (p->sample_rate != ~0) {
if (!cleared)
audio_clear(sc);
cleared = 1;
p->sample_rate = adpcm_round_sr(p->sample_rate);
printf("audiosetinfo: rate=%d\n", p->sample_rate);
sc->sc_orate = p->sample_rate;
if (sc->sc_mode == AUMODE_PLAY)
(void)adpcm_set_sr(sc->sc_orate);
}
if (r->sample_rate != ~0) {
if (!cleared)
audio_clear(sc);
cleared = 1;
r->sample_rate = adpcm_round_sr(r->sample_rate);
sc->sc_irate = r->sample_rate;
if (sc->sc_mode != AUMODE_PLAY)
(void)adpcm_set_sr(sc->sc_irate);
}
if (p->encoding != ~0) {
if (!cleared)
audio_clear(sc);
cleared = 1;
switch (p->encoding) {
case AUDIO_ENCODING_ULAW:
sc->sc_pencoding = AUDIO_ENCODING_ULAW;
break;
case AUDIO_ENCODING_LINEAR:
sc->sc_pencoding = AUDIO_ENCODING_LINEAR;
break;
default:
sc->sc_pencoding = AUDIO_ENCODING_ADPCM;
p->encoding = AUDIO_ENCODING_ADPCM;
}
}
if (r->encoding != ~0) {
switch (r->encoding) {
case AUDIO_ENCODING_ULAW:
sc->sc_rencoding = AUDIO_ENCODING_ULAW;
break;
case AUDIO_ENCODING_LINEAR:
sc->sc_rencoding = AUDIO_ENCODING_LINEAR;
break;
default:
sc->sc_rencoding = AUDIO_ENCODING_ADPCM;
r->encoding = AUDIO_ENCODING_ADPCM;
}
}
#ifdef notdef
if (p->pause != (u_char)~0)
sc->sc_au.au_wb.cb_pause = p->pause;
if (r->pause != (u_char)~0)
sc->sc_au.au_rb.cb_pause = r->pause;
#endif
if (ai->blocksize != ~0) {
if (!cleared)
audio_clear(sc);
cleared = 1;
if (ai->blocksize == 0)
bsize = audio_blocksize;
else if (ai->blocksize > NBPG/2)
bsize = NBPG/2;
else
bsize = ai->blocksize;
ai->blocksize = sc->sc_blksize = bsize;
audio_initbuf(sc);
}
if (ai->hiwat != ~0) {
if ((unsigned)ai->hiwat > sc->sc_maxblk)
ai->hiwat = sc->sc_maxblk;
sc->sc_hiwat = ai->hiwat;
}
if (ai->lowat != ~0) {
if ((unsigned)ai->lowat > sc->sc_maxblk)
ai->lowat = sc->sc_maxblk;
sc->sc_lowat = ai->lowat;
}
if (ai->backlog != ~0) {
if ((unsigned)ai->backlog > (sc->sc_maxblk/2))
ai->backlog = sc->sc_maxblk/2;
sc->sc_backlog = ai->backlog;
}
if (ai->mode != ~0) {
if (!cleared)
audio_clear(sc);
cleared = 1;
sc->sc_mode = ai->mode;
if (sc->sc_mode == AUMODE_PLAY)
audio_init_play(sc);
else
audio_init_record(sc);
}
#if 0
if (cleared) {
if (sc->sc_mode != AUMODE_PLAY)
audiostartr(sc);
else
audiostartp(sc);
}
#endif
return (0);
}
static int
audiogetinfo(sc, ai)
struct audio_softc *sc;
struct audio_info *ai;
{
struct audio_prinfo *r = &ai->record, *p = &ai->play;
p->sample_rate = sc->sc_orate;
r->sample_rate = sc->sc_irate;
p->channels = r->channels = 1;
p->precision = r->precision = 8;
p->encoding = sc->sc_pencoding;
r->encoding = sc->sc_rencoding;
ai->monitor_gain = 0;
r->gain = sc->sc_rlevel;
p->gain = sc->sc_plevel;
r->port = 1; p->port = AUDIO_SPEAKER;
#ifdef notdef
p->pause = sc->sc_au.au_wb.cb_pause;
r->pause = sc->sc_au.au_rb.cb_pause;
p->error = sc->sc_au.au_wb.cb_drops != 0;
r->error = sc->sc_au.au_rb.cb_drops != 0;
#endif
p->open = sc->sc_open;
r->open = sc->sc_open;
#ifdef notdef
p->samples = sc->sc_au.au_stamp - sc->sc_au.au_wb.cb_pdrops;
r->samples = sc->sc_au.au_stamp - sc->sc_au.au_rb.cb_pdrops;
#endif
p->seek = sc->sc_wseek;
r->seek = sc->sc_rseek;
ai->blocksize = sc->sc_blksize;
ai->hiwat = sc->sc_hiwat;
ai->lowat = sc->sc_lowat;
ai->backlog = sc->sc_backlog;
ai->mode = sc->sc_mode;
return (0);
}
u_char mulawtolin[256] = {
128, 4, 8, 12, 16, 20, 24, 28,
32, 36, 40, 44, 48, 52, 56, 60,
64, 66, 68, 70, 72, 74, 76, 78,
80, 82, 84, 86, 88, 90, 92, 94,
96, 97, 98, 99, 100, 101, 102, 103,
104, 105, 106, 107, 108, 109, 110, 111,
112, 112, 113, 113, 114, 114, 115, 115,
116, 116, 117, 117, 118, 118, 119, 119,
120, 120, 120, 121, 121, 121, 121, 122,
122, 122, 122, 123, 123, 123, 123, 124,
124, 124, 124, 124, 125, 125, 125, 125,
125, 125, 125, 125, 126, 126, 126, 126,
126, 126, 126, 126, 126, 126, 126, 126,
127, 127, 127, 127, 127, 127, 127, 127,
127, 127, 127, 127, 127, 127, 127, 127,
127, 127, 127, 127, 127, 127, 127, 127,
255, 251, 247, 243, 239, 235, 231, 227,
223, 219, 215, 211, 207, 203, 199, 195,
191, 189, 187, 185, 183, 181, 179, 177,
175, 173, 171, 169, 167, 165, 163, 161,
159, 158, 157, 156, 155, 154, 153, 152,
151, 150, 149, 148, 147, 146, 145, 144,
143, 143, 142, 142, 141, 141, 140, 140,
139, 139, 138, 138, 137, 137, 136, 136,
135, 135, 135, 134, 134, 134, 134, 133,
133, 133, 133, 132, 132, 132, 132, 131,
131, 131, 131, 131, 130, 130, 130, 130,
130, 130, 130, 130, 129, 129, 129, 129,
129, 129, 129, 129, 129, 129, 129, 129,
128, 128, 128, 128, 128, 128, 128, 128,
128, 128, 128, 128, 128, 128, 128, 128,
128, 128, 128, 128, 128, 128, 128, 128,
};
u_char lintomulaw[256] = {
0, 0, 0, 0, 0, 1, 1, 1,
1, 2, 2, 2, 2, 3, 3, 3,
3, 4, 4, 4, 4, 5, 5, 5,
5, 6, 6, 6, 6, 7, 7, 7,
7, 8, 8, 8, 8, 9, 9, 9,
9, 10, 10, 10, 10, 11, 11, 11,
11, 12, 12, 12, 12, 13, 13, 13,
13, 14, 14, 14, 14, 15, 15, 15,
15, 16, 16, 17, 17, 18, 18, 19,
19, 20, 20, 21, 21, 22, 22, 23,
23, 24, 24, 25, 25, 26, 26, 27,
27, 28, 28, 29, 29, 30, 30, 31,
31, 32, 33, 34, 35, 36, 37, 38,
39, 40, 41, 42, 43, 44, 45, 46,
47, 48, 50, 52, 54, 56, 58, 60,
62, 65, 69, 73, 77, 83, 91, 103,
255, 231, 219, 211, 205, 201, 197, 193,
190, 188, 186, 184, 182, 180, 178, 176,
175, 174, 173, 172, 171, 170, 169, 168,
167, 166, 165, 164, 163, 162, 161, 160,
159, 159, 158, 158, 157, 157, 156, 156,
155, 155, 154, 154, 153, 153, 152, 152,
151, 151, 150, 150, 149, 149, 148, 148,
147, 147, 146, 146, 145, 145, 144, 144,
143, 143, 143, 143, 142, 142, 142, 142,
141, 141, 141, 141, 140, 140, 140, 140,
139, 139, 139, 139, 138, 138, 138, 138,
137, 137, 137, 137, 136, 136, 136, 136,
135, 135, 135, 135, 134, 134, 134, 134,
133, 133, 133, 133, 132, 132, 132, 132,
131, 131, 131, 131, 130, 130, 130, 130,
129, 129, 129, 129, 128, 128, 128, 128,
};
#if 0
void
audio_tomulaw(p, cc)
register u_char *p;
register int cc;
{
register u_char *utab = lintomulaw;
while (--cc >= 0) {
*p = utab[*p];
++p;
}
}
#endif
static inline char
short2char(x)
short x;
{
if (x < 0){
x /= -256;
return (-1 * (char)x);
} else {
return ((char)x/256);
}
}
void
audio_tomulaw(p, cc)
short *p;
register int cc;
{
register u_char *utab = lintomulaw;
register int i;
for (i = 0; i < cc; i++) {
transbuf2[i] = utab[(u_char)(short2char(p[i]))];
}
}
void
audio_frommulaw(p, cc)
u_char *p;
register int cc;
{
register u_char *utab = mulawtolin;
register int i;
for (i = 0; i < cc; i++) {
transbuf[i] = (short)utab[p[i]];
}
}
double adpcm_estimindex[16] = {
1.0/8, 3.0/8, 5.0/8, 7.0/8, 9.0/8, 11.0/8, 13.0/8, 15.0/8,
-1.0/8, -3.0/8, -5.0/8, -7.0/8, -9.0/8, -11.0/8, -13.0/8, -15.0/8
};
double adpcm_estim[49] = {
16, 17, 19, 21, 23, 25, 28, 31, 34, 37,
41, 45, 50, 55, 60, 66, 73, 80, 88, 97,
107, 118, 130, 143, 157, 173, 190, 209, 230, 253,
279, 307, 337, 371, 408, 449, 494, 544, 598, 658,
724, 796, 875, 963, 1060, 1166, 1282, 1411, 1552 };
u_char adpcm_estimindex_correct[16] = {
-1, -1, -1, -1, 2, 4, 6, 8,
-1, -1, -1, -1, 2, 4, 6, 8
};
static inline void
adpcm2pcm_step(b, y, x)
u_char b;
short *y;
signed char *x;
{
*y += (short)(adpcm_estimindex[b] * adpcm_estim[*x]);
*x += adpcm_estimindex_correct[b];
if (*x < 0)
*x = 0;
else if (*x > 48)
*x = 48;
}
void
audio_tolinear(sc, p, cc)
struct audio_softc *sc;
register u_char *p;
register int cc;
{
signed char *x = &(sc->sc_adpcm.sc_estim);
short *y = &(sc->sc_adpcm.sc_amp);
u_char a, b;
int i;
AUDIODEBUG("audio_tolinear:\n");
for (i = 0; i < cc*2;) {
a = *p;
p++;
b = a & 0x0f;
adpcm2pcm_step(b, y, x);
transbuf[i++] = *y;
b = a >> 4;
adpcm2pcm_step(b, y, x);
transbuf[i++] = *y;
}
}
#if 0
void
audio_tolinear(sc, p, cc)
struct audio_softc *sc;
register u_char *p;
register int cc;
{
signed char x = 0;
short y = audio_softc[0].sc_amp;
u_char a, b;
int i;
for (i = 0; i < cc; i++) {
a = *p;
p++;
b = a & 0x0f;
y += (short)floor(adpcm_estimindex[b] * adpcm_estim[x]);
transbuf[i] = y;
x += adpcm_estimindex_correct[b];
if (x < 0)
x = 0;
else if (x > 48)
x = 48;
}
}
#endif
inline u_char
pcm2adpcm_step(a, y, x)
short a;
short *y;
signed char *x;
{
register u_char b;
double c, d;
a -= *y;
c = (double)a*4.0 / (d = adpcm_estim[*x]);
if (c < 0.0) {
b = (u_char)-c;
if (b >= 8)
b = 7;
b |= 0x08;
} else {
b = (u_char)c;
if (b >= 8)
b = 7;
}
*y += (short)(adpcm_estimindex[b] * d);
*x += adpcm_estimindex_correct[b];
if (*x < 0)
*x = 0;
else if (*x > 48)
*x = 48;
return b;
}
void
audio_fromlinear(sc, p, cc)
struct audio_softc *sc;
register u_char *p;
register int cc;
{
signed char *x = &(sc->sc_adpcm.sc_estim);
short *y = &(sc->sc_adpcm.sc_amp);
u_char f;
register int i;
for (i = 0; i < cc;) {
f = pcm2adpcm_step(transbuf[i++], y, x);
*p++ = f + (pcm2adpcm_step(transbuf[i++], y, x) << 4);
}
}
#endif