d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
79e6a82b48
freebsd-dev/sys/dev/sound/pcm/channel.c

1163 lines
27 KiB
C
Raw Normal View History

say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
/*
* Copyright (c) 1999 Cameron Grant <gandalf@vilnya.demon.co.uk>
* Portions Copyright by Luigi Rizzo - 1997-99
* 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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.
*/
repo-copied to make way for newmidi, this commit updates include paths
1999-11-20 16:50:33 +00:00
#include <dev/sound/pcm/sound.h>
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
#include "feeder_if.h"
many changes: * add new channels to the end of the list so channels used in order of addition * de-globalise definition of struct snddev_info and provide accessor functions where necessary. * move the $FreeBSD$ tag in each .c file into a macro and allow the /dev/sndstat handler to display these when set to maximum verbosity to aid debugging. * allow each device to register its own sndstat handler to reduce the amount of groping sndstat must do in foreign structs.
2001-08-23 11:30:52 +00:00
SND_DECLARE_FILE("$FreeBSD$");
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
#define MIN_CHUNK_SIZE 256 /* for uiomove etc. */
#define DMA_ALIGN_THRESHOLD 4
#define DMA_ALIGN_MASK (~(DMA_ALIGN_THRESHOLD - 1))
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
#define MIN(x, y) (((x) < (y))? (x) : (y))
#define CANCHANGE(c) (!(c->flags & CHN_F_TRIGGERED))
slight mods to isa dma and channel flushing code
2000-11-07 00:32:35 +00:00
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
/*
#define DEB(x) x
*/
slight mods to isa dma and channel flushing code
2000-11-07 00:32:35 +00:00
add a tunable/sysctl, hw.snd.targetirqrate, to modify the default target irq rate for apps that do not set a blocksize.
2001-09-18 14:45:09 +00:00
static int chn_targetirqrate = 32;
TUNABLE_INT("hw.snd.targetirqrate", &chn_targetirqrate);
static int
sysctl_hw_snd_targetirqrate(SYSCTL_HANDLER_ARGS)
{
int err, val;
val = chn_targetirqrate;
err = sysctl_handle_int(oidp, &val, sizeof(val), req);
if (val < 16 || val > 512)
err = EINVAL;
else
chn_targetirqrate = val;
return err;
}
SYSCTL_PROC(_hw_snd, OID_AUTO, targetirqrate, CTLTYPE_INT | CTLFLAG_RW,
0, sizeof(int), sysctl_hw_snd_targetirqrate, "I", "");
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
static int chn_buildfeeder(struct pcm_channel *c);
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
static void
chn_lockinit(struct pcm_channel *c)
{
c->lock = snd_mtxcreate(c->name);
}
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
static void
chn_lockdestroy(struct pcm_channel *c)
{
snd_mtxfree(c->lock);
}
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
static int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_polltrigger(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *bs = c->bufsoft;
unsigned amt, lim;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
if (c->flags & CHN_F_MAPPED) {
if (sndbuf_getprevblocks(bs) == 0)
return 1;
fix whitespace bogons
2001-03-25 21:43:24 +00:00
else
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
return (sndbuf_getblocks(bs) > sndbuf_getprevblocks(bs))? 1 : 0;
} else {
amt = (c->direction == PCMDIR_PLAY)? sndbuf_getfree(bs) : sndbuf_getready(bs);
lim = (c->flags & CHN_F_HAS_SIZE)? sndbuf_getblksz(bs) : 1;
lim = 1;
return (amt >= lim)? 1 : 0;
}
return 0;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
static int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_pollreset(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *bs = c->bufsoft;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
sndbuf_updateprevtotal(bs);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
return 1;
}
static void
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_wakeup(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *bs = c->bufsoft;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
if (sndbuf_getsel(bs)->si_pid && chn_polltrigger(c))
selwakeup(sndbuf_getsel(bs));
wakeup(bs);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
static int
chn_sleep(struct pcm_channel *c, char *str, int timeout)
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *bs = c->bufsoft;
int ret;
CHN_LOCKASSERT(c);
add defines and ifdefs so this code will compile on 4.x add spls so this code will work on 4.x
2001-06-23 17:36:51 +00:00
#ifdef USING_MUTEX
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = msleep(bs, c->lock, PRIBIO | PCATCH, str, timeout);
add defines and ifdefs so this code will compile on 4.x add spls so this code will work on 4.x
2001-06-23 17:36:51 +00:00
#else
ret = tsleep(bs, PRIBIO | PCATCH, str, timeout);
#endif
conditionalize "hwptr went backwards" messages modify default buffer size heuristic to not max out at 4k. mpg123 should now behave better under load.
2001-02-07 06:48:28 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
return ret;
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
}
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
/*
* chn_dmaupdate() tracks the status of a dma transfer,
* updating pointers. It must be called at spltty().
*/
fix dma underrun issues mutate some panics to kasserts add more spl protection PR: kern/14990 Partially Submitted by: Vladimir N.Silyaev <vns@delta.odessa.ua> Reviewed by: dfr
1999-12-05 19:09:13 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
static unsigned int
chn_dmaupdate(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b = c->bufhard;
unsigned int delta, old, hwptr, amt;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
KASSERT(sndbuf_getsize(b) > 0, ("bufsize == 0"));
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
fix mmapped playback remove dead debugging code
2001-12-21 22:34:01 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
old = sndbuf_gethwptr(b);
fix dma underrun issues mutate some panics to kasserts add more spl protection PR: kern/14990 Partially Submitted by: Vladimir N.Silyaev <vns@delta.odessa.ua> Reviewed by: dfr
1999-12-05 19:09:13 +00:00
hwptr = chn_getptr(c);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
delta = (sndbuf_getsize(b) + hwptr - old) % sndbuf_getsize(b);
sndbuf_sethwptr(b, hwptr);
conditionalize "hwptr went backwards" messages modify default buffer size heuristic to not max out at 4k. mpg123 should now behave better under load.
2001-02-07 06:48:28 +00:00
DEB(
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
if (delta >= ((sndbuf_getsize(b) * 15) / 16)) {
warn if hwptr goes backwards
2000-08-09 00:42:00 +00:00
if (!(c->flags & (CHN_F_CLOSING | CHN_F_ABORTING)))
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
device_printf(c->parentsnddev->dev, "hwptr went backwards %d -> %d\n", old, hwptr);
warn if hwptr goes backwards
2000-08-09 00:42:00 +00:00
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
);
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
if (c->direction == PCMDIR_PLAY) {
amt = MIN(delta, sndbuf_getready(b));
if (amt > 0)
sndbuf_dispose(b, NULL, amt);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
} else {
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
amt = MIN(delta, sndbuf_getfree(b));
if (amt > 0)
sndbuf_acquire(b, NULL, amt);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
return delta;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
void
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_wrupdate(struct pcm_channel *c)
fix dma underrun issues mutate some panics to kasserts add more spl protection PR: kern/14990 Partially Submitted by: Vladimir N.Silyaev <vns@delta.odessa.ua> Reviewed by: dfr
1999-12-05 19:09:13 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
int ret;
fix dma underrun issues mutate some panics to kasserts add more spl protection PR: kern/14990 Partially Submitted by: Vladimir N.Silyaev <vns@delta.odessa.ua> Reviewed by: dfr
1999-12-05 19:09:13 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
KASSERT(c->direction == PCMDIR_PLAY, ("chn_wrupdate on bad channel"));
lock sound device when adding/removing channels implement setblocksize for vchans don't panic when doing certain ioctls or aborting on a vchan xmms now works with vchans
2001-06-07 20:06:22 +00:00
if ((c->flags & (CHN_F_MAPPED | CHN_F_VIRTUAL)) || !(c->flags & CHN_F_TRIGGERED))
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
return;
chn_dmaupdate(c);
ret = chn_wrfeed(c);
/* tell the driver we've updated the primary buffer */
chn_trigger(c, PCMTRIG_EMLDMAWR);
DEB(if (ret)
printf("chn_wrupdate: chn_wrfeed returned %d\n", ret);)
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
}
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
int
chn_wrfeed(struct pcm_channel *c)
{
struct snd_dbuf *b = c->bufhard;
struct snd_dbuf *bs = c->bufsoft;
unsigned int ret, amt;
CHN_LOCKASSERT(c);
DEB(
if (c->flags & CHN_F_CLOSING) {
sndbuf_dump(b, "b", 0x02);
sndbuf_dump(bs, "bs", 0x02);
})
fix mmapped playback remove dead debugging code
2001-12-21 22:34:01 +00:00
if (c->flags & CHN_F_MAPPED)
sndbuf_acquire(bs, NULL, sndbuf_getfree(bs));
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
amt = sndbuf_getfree(b);
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
if (sndbuf_getready(bs) < amt)
c->xruns++;
fix mmapped playback remove dead debugging code
2001-12-21 22:34:01 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = (amt > 0)? sndbuf_feed(bs, b, c, c->feeder, amt) : ENOSPC;
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
if (ret == 0 && sndbuf_getfree(b) < amt)
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_wakeup(c);
fix mmapped playback remove dead debugging code
2001-12-21 22:34:01 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
return ret;
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
}
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
static void
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_wrintr(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
int ret;
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
/* update pointers in primary buffer */
fix whitespace bogons
2001-03-25 21:43:24 +00:00
chn_dmaupdate(c);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
/* ...and feed from secondary to primary */
ret = chn_wrfeed(c);
/* tell the driver we've updated the primary buffer */
fix whitespace bogons
2001-03-25 21:43:24 +00:00
chn_trigger(c, PCMTRIG_EMLDMAWR);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
DEB(if (ret)
printf("chn_wrintr: chn_wrfeed returned %d\n", ret);)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
/*
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
* user write routine - uiomove data into secondary buffer, trigger if necessary
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
* if blocking, sleep, rinse and repeat.
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
*
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
* called externally, so must handle locking
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
*/
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_write(struct pcm_channel *c, struct uio *buf)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
int ret, timeout, newsize, count, sz;
struct snd_dbuf *bs = c->bufsoft;
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
/*
* XXX Certain applications attempt to write larger size
* of pcm data than c->blocksize2nd without blocking,
* resulting partial write. Expand the block size so that
* the write operation avoids blocking.
*/
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
if ((c->flags & CHN_F_NBIO) && buf->uio_resid > sndbuf_getblksz(bs)) {
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
DEB(device_printf(c->parentsnddev->dev, "broken app, nbio and tried to write %d bytes with fragsz %d\n",
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
buf->uio_resid, sndbuf_getblksz(bs)));
make mmap sort-of work. there seem to be interactions with certain hw drivers, so still work in progress. do various mmap-related ioctls right. improve blocksize control. bits of cleanup.
2000-04-15 05:04:12 +00:00
newsize = 16;
implement settrigger according to spec fixes for non-blocking mode
2000-04-23 18:09:18 +00:00
while (newsize < min(buf->uio_resid, CHN_2NDBUFMAXSIZE / 2))
make mmap sort-of work. there seem to be interactions with certain hw drivers, so still work in progress. do various mmap-related ioctls right. improve blocksize control. bits of cleanup.
2000-04-15 05:04:12 +00:00
newsize <<= 1;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_setblocksize(c, sndbuf_getblkcnt(bs), newsize);
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
DEB(device_printf(c->parentsnddev->dev, "frags reset to %d x %d\n", sndbuf_getblkcnt(bs), sndbuf_getblksz(bs)));
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = 0;
fix whitespace bogons
2001-03-25 21:43:24 +00:00
count = hz;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
while (!ret && (buf->uio_resid > 0) && (count > 0)) {
sz = sndbuf_getfree(bs);
if (sz == 0) {
if (c->flags & CHN_F_NBIO)
ret = EWOULDBLOCK;
else {
timeout = (hz * sndbuf_getblksz(bs)) / (sndbuf_getspd(bs) * sndbuf_getbps(bs));
if (timeout < 1)
timeout = 1;
tweaks to reduce latency/pauses in output
2001-08-29 09:17:43 +00:00
timeout = 1;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = chn_sleep(c, "pcmwr", timeout);
if (ret == EWOULDBLOCK) {
count -= timeout;
ret = 0;
} else if (ret == 0)
fix whitespace bogons
2001-03-25 21:43:24 +00:00
count = hz;
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
} else {
sz = MIN(sz, buf->uio_resid);
KASSERT(sz > 0, ("confusion in chn_write"));
/* printf("sz: %d\n", sz); */
ret = sndbuf_uiomove(bs, buf, sz);
if (ret == 0 && !(c->flags & CHN_F_TRIGGERED))
chn_start(c, 0);
}
}
/* printf("ret: %d left: %d\n", ret, buf->uio_resid); */
if (count <= 0) {
fix whitespace bogons
2001-03-25 21:43:24 +00:00
c->flags |= CHN_F_DEAD;
add a method for recording of specific channels for devices with more than one hardware record channel. new devices, /dev/dsprX.Y where X is unit number and Y is channel index.
2001-09-05 16:28:41 +00:00
printf("%s: play interrupt timeout, channel dead\n", c->name);
fix whitespace bogons
2001-03-25 21:43:24 +00:00
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
return ret;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
static int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_rddump(struct pcm_channel *c, unsigned int cnt)
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b = c->bufhard;
CHN_LOCKASSERT(c);
sndbuf_setxrun(b, sndbuf_getxrun(b) + cnt);
return sndbuf_dispose(b, NULL, cnt);
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
}
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
/*
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
* Feed new data from the read buffer. Can be called in the bottom half.
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
* Hence must be called at spltty.
*/
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_rdfeed(struct pcm_channel *c)
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b = c->bufhard;
struct snd_dbuf *bs = c->bufsoft;
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
unsigned int ret, amt;
conditionalize "hwptr went backwards" messages modify default buffer size heuristic to not max out at 4k. mpg123 should now behave better under load.
2001-02-07 06:48:28 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
DEB(
if (c->flags & CHN_F_CLOSING) {
sndbuf_dump(b, "b", 0x02);
sndbuf_dump(bs, "bs", 0x02);
})
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
amt = sndbuf_getready(b);
improve recording behaviour
2001-10-24 12:33:21 +00:00
if (sndbuf_getfree(bs) < amt) {
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
c->xruns++;
improve recording behaviour
2001-10-24 12:33:21 +00:00
amt = sndbuf_getfree(bs);
}
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
ret = (amt > 0)? sndbuf_feed(b, bs, c, c->feeder, amt) : 0;
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
don't try to dump nonexistent data when recording
2001-09-05 14:49:44 +00:00
amt = sndbuf_getready(b);
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
if (amt > 0)
chn_rddump(c, amt);
chn_wakeup(c);
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
return ret;
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
void
chn_rdupdate(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
int ret;
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
KASSERT(c->direction == PCMDIR_REC, ("chn_rdupdate on bad channel"));
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
if ((c->flags & CHN_F_MAPPED) || !(c->flags & CHN_F_TRIGGERED))
return;
chn_trigger(c, PCMTRIG_EMLDMARD);
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
chn_dmaupdate(c);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = chn_rdfeed(c);
if (ret)
printf("chn_rdfeed: %d\n", ret);
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
}
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
/* read interrupt routine. Must be called with interrupts blocked. */
static void
chn_rdintr(struct pcm_channel *c)
{
int ret;
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
/* tell the driver to update the primary buffer if non-dma */
fix whitespace bogons
2001-03-25 21:43:24 +00:00
chn_trigger(c, PCMTRIG_EMLDMARD);
tweaks to reduce latency/pauses in output
2001-08-29 09:17:43 +00:00
/* update pointers in primary buffer */
fix whitespace bogons
2001-03-25 21:43:24 +00:00
chn_dmaupdate(c);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
/* ...and feed from primary to secondary */
ret = chn_rdfeed(c);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
/*
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
* user read routine - trigger if necessary, uiomove data from secondary bufhard
* if blocking, sleep, rinse and repeat.
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
*
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
* called externally, so must handle locking
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
*/
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_read(struct pcm_channel *c, struct uio *buf)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
int ret, timeout, sz, count;
struct snd_dbuf *bs = c->bufsoft;
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
if (!(c->flags & CHN_F_TRIGGERED))
add reinit functions to mixers unstaticize chn_start() add reset/resetdone functions to channels
2000-10-26 20:46:58 +00:00
chn_start(c, 0);
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = 0;
fix whitespace bogons
2001-03-25 21:43:24 +00:00
count = hz;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
while (!ret && (buf->uio_resid > 0) && (count > 0)) {
improve recording behaviour
2001-10-24 12:33:21 +00:00
sz = MIN(buf->uio_resid, sndbuf_getready(bs));
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
improve recording behaviour
2001-10-24 12:33:21 +00:00
if (sz > 0) {
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = sndbuf_uiomove(bs, buf, sz);
} else {
improve recording behaviour
2001-10-24 12:33:21 +00:00
if (c->flags & CHN_F_NBIO) {
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = EWOULDBLOCK;
improve recording behaviour
2001-10-24 12:33:21 +00:00
} else {
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
timeout = (hz * sndbuf_getblksz(bs)) / (sndbuf_getspd(bs) * sndbuf_getbps(bs));
if (timeout < 1)
timeout = 1;
ret = chn_sleep(c, "pcmrd", timeout);
if (ret == EWOULDBLOCK) {
count -= timeout;
ret = 0;
improve recording behaviour
2001-10-24 12:33:21 +00:00
} else {
count = hz;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
}
improve recording behaviour
2001-10-24 12:33:21 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
}
add record channel irq timeouts too
2000-06-20 23:42:08 +00:00
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
}
if (count <= 0) {
fix whitespace bogons
2001-03-25 21:43:24 +00:00
c->flags |= CHN_F_DEAD;
add a method for recording of specific channels for devices with more than one hardware record channel. new devices, /dev/dsprX.Y where X is unit number and Y is channel index.
2001-09-05 16:28:41 +00:00
printf("%s: record interrupt timeout, channel dead\n", c->name);
fix whitespace bogons
2001-03-25 21:43:24 +00:00
}
improve recording
2000-06-06 22:30:22 +00:00
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
return ret;
}
void
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_intr(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCK(c);
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
c->interrupts++;
more mmap fixes
2000-04-17 17:06:47 +00:00
if (c->direction == PCMDIR_PLAY)
chn_wrintr(c);
else
chn_rdintr(c);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_UNLOCK(c);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
u_int32_t
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_start(struct pcm_channel *c, int force)
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
{
tweaks to reduce latency/pauses in output
2001-08-29 09:17:43 +00:00
u_int32_t i, j;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b = c->bufhard;
struct snd_dbuf *bs = c->bufsoft;
CHN_LOCKASSERT(c);
/* if we're running, or if we're prevented from triggering, bail */
MFS: make SNDCTL_DSP_SETTRIGGER work
2001-08-23 11:58:38 +00:00
if ((c->flags & CHN_F_TRIGGERED) || ((c->flags & CHN_F_NOTRIGGER) && !force))
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
return EINVAL;
i = (c->direction == PCMDIR_PLAY)? sndbuf_getready(bs) : sndbuf_getfree(bs);
tweaks to reduce latency/pauses in output
2001-08-29 09:17:43 +00:00
j = (c->direction == PCMDIR_PLAY)? sndbuf_getfree(b) : sndbuf_getready(b);
if (force || (i >= j)) {
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
c->flags |= CHN_F_TRIGGERED;
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
/*
* if we're starting because a vchan started, don't feed any data
* or it becomes impossible to start vchans synchronised with the
* first one. the hardbuf should be empty so we top it up with
* silence to give it something to chew. the real data will be
* fed at the first irq.
*/
if (c->direction == PCMDIR_PLAY) {
if (SLIST_EMPTY(&c->children))
chn_wrfeed(c);
else
sndbuf_fillsilence(b);
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
sndbuf_setrun(b, 1);
tweaks to reduce latency/pauses in output
2001-08-29 09:17:43 +00:00
c->xruns = 0;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_trigger(c, PCMTRIG_START);
return 0;
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
return 0;
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
}
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
void
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_resetbuf(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b = c->bufhard;
struct snd_dbuf *bs = c->bufsoft;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
make mmap sort-of work. there seem to be interactions with certain hw drivers, so still work in progress. do various mmap-related ioctls right. improve blocksize control. bits of cleanup.
2000-04-15 05:04:12 +00:00
c->blocks = 0;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
sndbuf_reset(b);
sndbuf_reset(bs);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
/*
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
* chn_sync waits until the space in the given channel goes above
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
* a threshold. The threshold is checked against fl or rl respectively.
* Assume that the condition can become true, do not check here...
*/
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_sync(struct pcm_channel *c, int threshold)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
u_long rdy;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
int ret;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *bs = c->bufsoft;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
for (;;) {
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
rdy = (c->direction == PCMDIR_PLAY)? sndbuf_getfree(bs) : sndbuf_getready(bs);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
if (rdy <= threshold) {
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = chn_sleep(c, "pcmsyn", 1);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
if (ret == ERESTART || ret == EINTR) {
fix dma underrun issues mutate some panics to kasserts add more spl protection PR: kern/14990 Partially Submitted by: Vladimir N.Silyaev <vns@delta.odessa.ua> Reviewed by: dfr
1999-12-05 19:09:13 +00:00
DEB(printf("chn_sync: tsleep returns %d\n", ret));
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
return -1;
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
} else
break;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
return 0;
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
/* called externally, handle locking */
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
chn_poll(struct pcm_channel *c, int ev, struct thread *td)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *bs = c->bufsoft;
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
int ret;
various locking fixes, rework open logic and channel registration PR: kern/28084
2001-06-14 13:31:30 +00:00
CHN_LOCKASSERT(c);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
if (!(c->flags & CHN_F_MAPPED) && !(c->flags & CHN_F_TRIGGERED))
fix whitespace bogons
2001-03-25 21:43:24 +00:00
chn_start(c, 1);
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
ret = 0;
if (chn_polltrigger(c) && chn_pollreset(c))
ret = ev;
make mmap sort-of work. there seem to be interactions with certain hw drivers, so still work in progress. do various mmap-related ioctls right. improve blocksize control. bits of cleanup.
2000-04-15 05:04:12 +00:00
else
Use the passed in thread to selrecord() instead of curthread.
2001-09-21 22:46:54 +00:00
selrecord(td, sndbuf_getsel(bs));
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
return ret;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
/*
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
* chn_abort terminates a running dma transfer. it may sleep up to 200ms.
* it returns the number of bytes that have not been transferred.
*
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
* called from: dsp_close, dsp_ioctl, with channel locked
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
*/
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_abort(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
use a global devclass for all drivers - i'm not entirely sure why this worked before. mixer, dsp and sndstat are seperate devices - give them their own cdevsws instead of demuxing requests sent to a single cdevsw. use the si_drv1/si_drv2 fields in dev_t structures for holding information specific to an open instance of mixer/dsp. nuke /dev/{dsp,dspW,audio}[0-9]* links - this functionality is now provided using cloning. various locking fixes.
2001-06-16 21:25:10 +00:00
int missing = 0;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b = c->bufhard;
struct snd_dbuf *bs = c->bufsoft;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
if (!(c->flags & CHN_F_TRIGGERED))
handle closing differently - should fix the end-of-sample cutoff bug
2000-06-13 23:18:43 +00:00
return 0;
fix the buffer repetition on ^c problem
1999-12-18 22:24:50 +00:00
c->flags |= CHN_F_ABORTING;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
c->flags &= ~CHN_F_TRIGGERED;
/* kill the channel */
fix the buffer repetition on ^c problem
1999-12-18 22:24:50 +00:00
chn_trigger(c, PCMTRIG_ABORT);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
sndbuf_setrun(b, 0);
lock sound device when adding/removing channels implement setblocksize for vchans don't panic when doing certain ioctls or aborting on a vchan xmms now works with vchans
2001-06-07 20:06:22 +00:00
if (!(c->flags & CHN_F_VIRTUAL))
chn_dmaupdate(c);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
missing = sndbuf_getready(bs) + sndbuf_getready(b);
c->flags &= ~CHN_F_ABORTING;
fix whitespace bogons
2001-03-25 21:43:24 +00:00
return missing;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
/*
* this routine tries to flush the dma transfer. It is called
* on a close. We immediately abort any read DMA
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
* operation, and then wait for the play bufhard to drain.
*
* called from: dsp_close
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
*/
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_flush(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
int ret, count, resid, resid_p;
struct snd_dbuf *b = c->bufhard;
struct snd_dbuf *bs = c->bufsoft;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
KASSERT(c->direction == PCMDIR_PLAY, ("chn_wrupdate on bad channel"));
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
DEB(printf("chn_flush c->flags 0x%08x\n", c->flags));
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
if (!(c->flags & CHN_F_TRIGGERED))
slight mods to isa dma and channel flushing code
2000-11-07 00:32:35 +00:00
return 0;
fix whitespace bogons
2001-03-25 21:43:24 +00:00
c->flags |= CHN_F_CLOSING;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
resid = sndbuf_getready(bs) + sndbuf_getready(b);
fix whitespace bogons
2001-03-25 21:43:24 +00:00
resid_p = resid;
count = 10;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = 0;
while ((count > 0) && (resid > sndbuf_getsize(b)) && (ret == 0)) {
fix whitespace bogons
2001-03-25 21:43:24 +00:00
/* still pending output data. */
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = chn_sleep(c, "pcmflu", hz / 10);
if (ret == EWOULDBLOCK)
ret = 0;
if (ret == 0) {
resid = sndbuf_getready(bs) + sndbuf_getready(b);
handle closing differently - should fix the end-of-sample cutoff bug
2000-06-13 23:18:43 +00:00
if (resid >= resid_p)
count--;
resid_p = resid;
fix whitespace bogons
2001-03-25 21:43:24 +00:00
}
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
if (count == 0)
DEB(printf("chn_flush: timeout\n"));
c->flags &= ~CHN_F_TRIGGERED;
/* kill the channel */
chn_trigger(c, PCMTRIG_ABORT);
sndbuf_setrun(b, 0);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
c->flags &= ~CHN_F_CLOSING;
return 0;
}
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
int
fmtvalid(u_int32_t fmt, u_int32_t *fmtlist)
{
int i;
for (i = 0; fmtlist[i]; i++)
if (fmt == fmtlist[i])
return 1;
return 0;
}
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_reset(struct pcm_channel *c, u_int32_t fmt)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
do not blindly assume 8khz is supported on open(). try for 8khz but respect minspeed/maxspeed specified by the hw driver. Submitted by: Andrew Gordon <arg@arg1.demon.co.uk>
2000-11-18 03:43:04 +00:00
int hwspd, r = 0;
make mmap sort-of work. there seem to be interactions with certain hw drivers, so still work in progress. do various mmap-related ioctls right. improve blocksize control. bits of cleanup.
2000-04-15 05:04:12 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
c->flags &= CHN_F_RESET;
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
c->interrupts = 0;
c->xruns = 0;
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
CHANNEL_RESET(c->methods, c->devinfo);
more mmap fixes
2000-04-17 17:06:47 +00:00
if (fmt) {
do not blindly assume 8khz is supported on open(). try for 8khz but respect minspeed/maxspeed specified by the hw driver. Submitted by: Andrew Gordon <arg@arg1.demon.co.uk>
2000-11-18 03:43:04 +00:00
hwspd = DSP_DEFAULT_SPEED;
now we have the rate feeder, we don't need to constrain the default channel speed. however, continue to do so for record channels until the feederchain builder is fixed.
2001-08-27 01:02:13 +00:00
/* only do this on a record channel until feederbuilder works */
if (c->direction == PCMDIR_REC)
RANGE(hwspd, chn_getcaps(c)->minspeed, chn_getcaps(c)->maxspeed);
do not blindly assume 8khz is supported on open(). try for 8khz but respect minspeed/maxspeed specified by the hw driver. Submitted by: Andrew Gordon <arg@arg1.demon.co.uk>
2000-11-18 03:43:04 +00:00
c->speed = hwspd;
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
r = chn_setformat(c, fmt);
if (r == 0)
do not blindly assume 8khz is supported on open(). try for 8khz but respect minspeed/maxspeed specified by the hw driver. Submitted by: Andrew Gordon <arg@arg1.demon.co.uk>
2000-11-18 03:43:04 +00:00
r = chn_setspeed(c, hwspd);
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
if (r == 0)
r = chn_setvolume(c, 100, 100);
more mmap fixes
2000-04-17 17:06:47 +00:00
}
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
r = chn_setblocksize(c, 0, 0);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
if (r == 0) {
fix whitespace bogons
2001-03-25 21:43:24 +00:00
chn_resetbuf(c);
CHANNEL_RESETDONE(c->methods, c->devinfo);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
return r;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_init(struct pcm_channel *c, void *devinfo, int dir)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
struct feeder_class *fc;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b, *bs;
more dma underrun fixes, using a per-channel 2nd buffer Submitted by: tanimura
1999-12-13 03:29:09 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_lockinit(c);
CHN_LOCK(c);
add some extra diagnostic info to sndstat output.
2001-08-29 02:49:54 +00:00
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
c->feeder = NULL;
fc = feeder_getclass(NULL);
if (fc == NULL)
return EINVAL;
if (chn_addfeeder(c, fc, NULL))
return EINVAL;
rework feeder sytem to allow feeders in klds modify driver capability reporting format to list every audio format seperately- required for above and because we could not previously indicate that mono was unsupported. there should be no functional impact.
2000-08-20 22:18:56 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
b = sndbuf_create(c->name, "primary");
if (b == NULL)
return ENOMEM;
bs = sndbuf_create(c->name, "secondary");
if (bs == NULL) {
sndbuf_destroy(b);
return ENOMEM;
}
sndbuf_setup(bs, NULL, 0);
c->bufhard = b;
c->bufsoft = bs;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
c->flags = 0;
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
c->feederflags = 0;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
c->devinfo = CHANNEL_INIT(c->methods, devinfo, b, c, dir);
if (c->devinfo == NULL) {
sndbuf_destroy(bs);
sndbuf_destroy(b);
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
return ENODEV;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
}
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
if ((sndbuf_getsize(b) == 0) && ((c->flags & CHN_F_VIRTUAL) == 0)) {
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
sndbuf_destroy(bs);
sndbuf_destroy(b);
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
return ENOMEM;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
}
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
chn_setdir(c, dir);
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
add a tunable/sysctl, hw.snd.targetirqrate, to modify the default target irq rate for apps that do not set a blocksize.
2001-09-18 14:45:09 +00:00
/* And the secondary buffer. */
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
sndbuf_setfmt(b, AFMT_U8);
sndbuf_setfmt(bs, AFMT_U8);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_UNLOCK(c);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
return 0;
}
change mixer api slightly change channel interface - kobj implementation coming soonish make pcm_makelinks not panic if modular add pcm_unregister() these changes support newpcm kld unloading, but this is only implemented by ds1.c
2000-09-01 20:09:24 +00:00
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_kill(struct pcm_channel *c)
change mixer api slightly change channel interface - kobj implementation coming soonish make pcm_makelinks not panic if modular add pcm_unregister() these changes support newpcm kld unloading, but this is only implemented by ds1.c
2000-09-01 20:09:24 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b = c->bufhard;
struct snd_dbuf *bs = c->bufsoft;
CHN_LOCK(c);
detach support remove un-needed setdir functions add bus_teardown_intr calls where necessary destroy our dma tags where necessary destroy ac97 before releasing resources
2000-09-09 19:21:04 +00:00
if (c->flags & CHN_F_TRIGGERED)
chn_trigger(c, PCMTRIG_ABORT);
change mixer api slightly change channel interface - kobj implementation coming soonish make pcm_makelinks not panic if modular add pcm_unregister() these changes support newpcm kld unloading, but this is only implemented by ds1.c
2000-09-01 20:09:24 +00:00
while (chn_removefeeder(c) == 0);
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
if (CHANNEL_FREE(c->methods, c->devinfo))
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
sndbuf_free(c->bufhard);
change mixer api slightly change channel interface - kobj implementation coming soonish make pcm_makelinks not panic if modular add pcm_unregister() these changes support newpcm kld unloading, but this is only implemented by ds1.c
2000-09-01 20:09:24 +00:00
c->flags |= CHN_F_DEAD;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
sndbuf_destroy(bs);
sndbuf_destroy(b);
chn_lockdestroy(c);
change mixer api slightly change channel interface - kobj implementation coming soonish make pcm_makelinks not panic if modular add pcm_unregister() these changes support newpcm kld unloading, but this is only implemented by ds1.c
2000-09-01 20:09:24 +00:00
return 0;
}
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_setdir(struct pcm_channel *c, int dir)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b = c->bufhard;
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
int r;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
c->direction = dir;
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
r = CHANNEL_SETDIR(c->methods, c->devinfo, c->direction);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
if (!r && ISA_DMA(b))
sndbuf_isadmasetdir(b, c->direction);
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
return r;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_setvolume(struct pcm_channel *c, int left, int right)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
/* could add a feeder for volume changing if channel returns -1 */
fix whitespace bogons
2001-03-25 21:43:24 +00:00
c->volume = (left << 8) | right;
return 0;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
make attempts to set unsupported speeds or formats non-sticky. this should fix problems with apps that probe for a variety of settings. Submitted by: Orion Hodson <O.Hodson@cs.ucl.ac.uk>
2001-02-13 21:57:34 +00:00
static int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_tryspeed(struct pcm_channel *c, int speed)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct pcm_feeder *f;
struct snd_dbuf *b = c->bufhard;
struct snd_dbuf *bs = c->bufsoft;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
int r, delta;
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
DEB(printf("setspeed, channel %s\n", c->name));
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
DEB(printf("want speed %d, ", speed));
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
if (speed <= 0)
return EINVAL;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
if (CANCHANGE(c)) {
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
r = 0;
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
c->speed = speed;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
sndbuf_setspd(bs, speed);
RANGE(speed, chn_getcaps(c)->minspeed, chn_getcaps(c)->maxspeed);
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
DEB(printf("try speed %d, ", speed));
sndbuf_setspd(b, CHANNEL_SETSPEED(c->methods, c->devinfo, speed));
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
DEB(printf("got speed %d\n", sndbuf_getspd(b)));
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
delta = sndbuf_getspd(b) - sndbuf_getspd(bs);
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
if (delta < 0)
delta = -delta;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
c->feederflags &= ~(1 << FEEDER_RATE);
if (delta > 500)
c->feederflags |= 1 << FEEDER_RATE;
else
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
sndbuf_setspd(bs, sndbuf_getspd(b));
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
r = chn_buildfeeder(c);
DEB(printf("r = %d\n", r));
if (r)
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
goto out;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
r = chn_setblocksize(c, 0, 0);
if (r)
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
goto out;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
if (!(c->feederflags & (1 << FEEDER_RATE)))
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
goto out;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
r = EINVAL;
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
f = chn_findfeeder(c, FEEDER_RATE);
DEB(printf("feedrate = %p\n", f));
if (f == NULL)
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
goto out;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
r = FEEDER_SET(f, FEEDRATE_SRC, sndbuf_getspd(bs));
DEB(printf("feeder_set(FEEDRATE_SRC, %d) = %d\n", sndbuf_getspd(bs), r));
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
if (r)
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
goto out;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
r = FEEDER_SET(f, FEEDRATE_DST, sndbuf_getspd(b));
DEB(printf("feeder_set(FEEDRATE_DST, %d) = %d\n", sndbuf_getspd(b), r));
out:
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
DEB(printf("setspeed done, r = %d\n", r));
fix whitespace bogons
2001-03-25 21:43:24 +00:00
return r;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
} else
return EINVAL;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_setspeed(struct pcm_channel *c, int speed)
make attempts to set unsupported speeds or formats non-sticky. this should fix problems with apps that probe for a variety of settings. Submitted by: Orion Hodson <O.Hodson@cs.ucl.ac.uk>
2001-02-13 21:57:34 +00:00
{
int r, oldspeed = c->speed;
r = chn_tryspeed(c, speed);
if (r) {
DEB(printf("Failed to set speed %d falling back to %d\n", speed, oldspeed));
chn_tryspeed(c, oldspeed);
}
return r;
}
static int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_tryformat(struct pcm_channel *c, u_int32_t fmt)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b = c->bufhard;
struct snd_dbuf *bs = c->bufsoft;
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
int r;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
if (CANCHANGE(c)) {
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
DEB(printf("want format %d\n", fmt));
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
c->format = fmt;
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
r = chn_buildfeeder(c);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
if (r == 0) {
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
sndbuf_setfmt(b, c->feeder->desc->out);
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
sndbuf_setfmt(bs, fmt);
fix whitespace bogons
2001-03-25 21:43:24 +00:00
chn_resetbuf(c);
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
CHANNEL_SETFORMAT(c->methods, c->devinfo, sndbuf_getfmt(b));
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
r = chn_tryspeed(c, c->speed);
fix whitespace bogons
2001-03-25 21:43:24 +00:00
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
return r;
} else
return EINVAL;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
make attempts to set unsupported speeds or formats non-sticky. this should fix problems with apps that probe for a variety of settings. Submitted by: Orion Hodson <O.Hodson@cs.ucl.ac.uk>
2001-02-13 21:57:34 +00:00
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_setformat(struct pcm_channel *c, u_int32_t fmt)
make attempts to set unsupported speeds or formats non-sticky. this should fix problems with apps that probe for a variety of settings. Submitted by: Orion Hodson <O.Hodson@cs.ucl.ac.uk>
2001-02-13 21:57:34 +00:00
{
u_int32_t oldfmt = c->format;
int r;
r = chn_tryformat(c, fmt);
if (r) {
DEB(printf("Format change %d failed, reverting to %d\n", fmt, oldfmt));
chn_tryformat(c, oldfmt);
}
return r;
}
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_setblocksize(struct pcm_channel *c, int blkcnt, int blksz)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b = c->bufhard;
struct snd_dbuf *bs = c->bufsoft;
int bufsz, irqhz, tmp, ret;
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
if (!CANCHANGE(c) || (c->flags & CHN_F_MAPPED))
make mmap sort-of work. there seem to be interactions with certain hw drivers, so still work in progress. do various mmap-related ioctls right. improve blocksize control. bits of cleanup.
2000-04-15 05:04:12 +00:00
return EINVAL;
move various fields from pcm_channel to snd_dbuf improve chn_read code- don't stop/restart on overrun, just dump data more error checking on ioctls
2000-05-26 21:55:13 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = 0;
Update to C99, s/__FUNCTION__/__func__/, also don't use ANSI string concatenation.
2001-12-10 08:09:49 +00:00
DEB(printf("%s(%d, %d)\n", __func__, blkcnt, blksz));
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
if (blksz == 0 || blksz == -1) {
if (blksz == -1)
c->flags &= ~CHN_F_HAS_SIZE;
prevent excessively small buffer sizes resulting in hundreds of irqs per second
2001-01-24 01:15:50 +00:00
if (!(c->flags & CHN_F_HAS_SIZE)) {
add a tunable/sysctl, hw.snd.targetirqrate, to modify the default target irq rate for apps that do not set a blocksize.
2001-09-18 14:45:09 +00:00
blksz = (sndbuf_getbps(bs) * sndbuf_getspd(bs)) / chn_targetirqrate;
prevent excessively small buffer sizes resulting in hundreds of irqs per second
2001-01-24 01:15:50 +00:00
tmp = 32;
while (tmp <= blksz)
tmp <<= 1;
tmp >>= 1;
blksz = tmp;
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
blkcnt = CHN_2NDBUFMAXSIZE / blksz;
prevent excessively small buffer sizes resulting in hundreds of irqs per second
2001-01-24 01:15:50 +00:00
RANGE(blksz, 16, CHN_2NDBUFMAXSIZE / 2);
RANGE(blkcnt, 2, CHN_2NDBUFMAXSIZE / blksz);
Update to C99, s/__FUNCTION__/__func__/, also don't use ANSI string concatenation.
2001-12-10 08:09:49 +00:00
DEB(printf("%s: defaulting to (%d, %d)\n", __func__, blkcnt, blksz));
prevent excessively small buffer sizes resulting in hundreds of irqs per second
2001-01-24 01:15:50 +00:00
} else {
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
blkcnt = sndbuf_getblkcnt(bs);
blksz = sndbuf_getblksz(bs);
Update to C99, s/__FUNCTION__/__func__/, also don't use ANSI string concatenation.
2001-12-10 08:09:49 +00:00
DEB(printf("%s: updating (%d, %d)\n", __func__, blkcnt, blksz));
prevent excessively small buffer sizes resulting in hundreds of irqs per second
2001-01-24 01:15:50 +00:00
}
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
} else {
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = EINVAL;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
if ((blksz < 16) || (blkcnt < 2) || (blkcnt * blksz > CHN_2NDBUFMAXSIZE))
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
goto out;
ret = 0;
implement settrigger according to spec fixes for non-blocking mode
2000-04-23 18:09:18 +00:00
c->flags |= CHN_F_HAS_SIZE;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
}
make mmap sort-of work. there seem to be interactions with certain hw drivers, so still work in progress. do various mmap-related ioctls right. improve blocksize control. bits of cleanup.
2000-04-15 05:04:12 +00:00
bufsz = blkcnt * blksz;
- latest 2ndbuffer patch - make chn_setdir work for rec on isa cards - note: es1371 does not irq in smp Submitted by: tanimura
1999-12-29 03:46:54 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
ret = ENOMEM;
if (sndbuf_remalloc(bs, blkcnt, blksz))
goto out;
ret = 0;
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
/* adjust for different hw format/speed */
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
irqhz = (sndbuf_getbps(bs) * sndbuf_getspd(bs)) / sndbuf_getblksz(bs);
Update to C99, s/__FUNCTION__/__func__/, also don't use ANSI string concatenation.
2001-12-10 08:09:49 +00:00
DEB(printf("%s: soft bps %d, spd %d, irqhz == %d\n", __func__, sndbuf_getbps(bs), sndbuf_getspd(bs), irqhz));
prevent excessively small buffer sizes resulting in hundreds of irqs per second
2001-01-24 01:15:50 +00:00
RANGE(irqhz, 16, 512);
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
sndbuf_setblksz(b, (sndbuf_getbps(b) * sndbuf_getspd(b)) / irqhz);
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
/* round down to 2^x */
blksz = 32;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
while (blksz <= sndbuf_getblksz(b))
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
blksz <<= 1;
blksz >>= 1;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
/* round down to fit hw bufhard size */
RANGE(blksz, 16, sndbuf_getmaxsize(b) / 2);
Update to C99, s/__FUNCTION__/__func__/, also don't use ANSI string concatenation.
2001-12-10 08:09:49 +00:00
DEB(printf("%s: hard blksz requested %d (maxsize %d), ", __func__, blksz, sndbuf_getmaxsize(b)));
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
sndbuf_setblksz(b, CHANNEL_SETBLOCKSIZE(c->methods, c->devinfo, blksz));
make mmap sort-of work. there seem to be interactions with certain hw drivers, so still work in progress. do various mmap-related ioctls right. improve blocksize control. bits of cleanup.
2000-04-15 05:04:12 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
irqhz = (sndbuf_getbps(b) * sndbuf_getspd(b)) / sndbuf_getblksz(b);
DEB(printf("got %d, irqhz == %d\n", sndbuf_getblksz(b), irqhz));
back out rev 1.49, looks like i committed it from the wrong tree. include the changes 1.49 was supposed to have. Noticed by: peter
2001-02-07 20:45:40 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_resetbuf(c);
out:
return ret;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_trigger(struct pcm_channel *c, int go)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct snd_dbuf *b = c->bufhard;
int ret;
CHN_LOCKASSERT(c);
if (ISA_DMA(b) && (go == PCMTRIG_EMLDMAWR || go == PCMTRIG_EMLDMARD))
sndbuf_isadmabounce(b);
ret = CHANNEL_TRIGGER(c->methods, c->devinfo, go);
return ret;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_getptr(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
fix dma underrun issues mutate some panics to kasserts add more spl protection PR: kern/14990 Partially Submitted by: Vladimir N.Silyaev <vns@delta.odessa.ua> Reviewed by: dfr
1999-12-05 19:09:13 +00:00
int hwptr;
int a = (1 << c->align) - 1;
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
hwptr = (c->flags & CHN_F_TRIGGERED)? CHANNEL_GETPTR(c->methods, c->devinfo) : 0;
fix dma underrun issues mutate some panics to kasserts add more spl protection PR: kern/14990 Partially Submitted by: Vladimir N.Silyaev <vns@delta.odessa.ua> Reviewed by: dfr
1999-12-05 19:09:13 +00:00
/* don't allow unaligned values in the hwa ptr */
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
#if 1
fix dma underrun issues mutate some panics to kasserts add more spl protection PR: kern/14990 Partially Submitted by: Vladimir N.Silyaev <vns@delta.odessa.ua> Reviewed by: dfr
1999-12-05 19:09:13 +00:00
hwptr &= ~a ; /* Apply channel align mask */
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
#endif
fix dma underrun issues mutate some panics to kasserts add more spl protection PR: kern/14990 Partially Submitted by: Vladimir N.Silyaev <vns@delta.odessa.ua> Reviewed by: dfr
1999-12-05 19:09:13 +00:00
hwptr &= DMA_ALIGN_MASK; /* Apply DMA align mask */
return hwptr;
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
struct pcmchan_caps *
chn_getcaps(struct pcm_channel *c)
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
{
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
return CHANNEL_GETCAPS(c->methods, c->devinfo);
say hello to newpcm. it is not yet enabled, requiring new pnp code from dfr to compile successfully. further details will be provided in the commit enabling newpcm.
1999-09-01 04:08:39 +00:00
}
rework feeder sytem to allow feeders in klds modify driver capability reporting format to list every audio format seperately- required for above and because we could not previously indicate that mono was unsupported. there should be no functional impact.
2000-08-20 22:18:56 +00:00
u_int32_t
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_getformats(struct pcm_channel *c)
rework feeder sytem to allow feeders in klds modify driver capability reporting format to list every audio format seperately- required for above and because we could not previously indicate that mono was unsupported. there should be no functional impact.
2000-08-20 22:18:56 +00:00
{
u_int32_t *fmtlist, fmts;
int i;
fmtlist = chn_getcaps(c)->fmtlist;
fmts = 0;
for (i = 0; fmtlist[i]; i++)
fmts |= fmtlist[i];
return fmts;
}
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
static int
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
chn_buildfeeder(struct pcm_channel *c)
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
{
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
struct feeder_class *fc;
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
struct pcm_feederdesc desc;
make the feederchain builder work for recording. this has not been tested extensively as none of my testboxes have speakers or an audio source at present, but the chains built look correct and reading /dev/audio (ulaw, translated from signed 16 bit little-endian) gives values within the expected range for silence.
2002-01-23 05:10:56 +00:00
u_int32_t tmp[2], type, flags, hwfmt;
rework feeder sytem to allow feeders in klds modify driver capability reporting format to list every audio format seperately- required for above and because we could not previously indicate that mono was unsupported. there should be no functional impact.
2000-08-20 22:18:56 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
CHN_LOCKASSERT(c);
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
while (chn_removefeeder(c) == 0);
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
KASSERT((c->feeder == NULL), ("feeder chain not empty"));
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
c->align = sndbuf_getalign(c->bufsoft);
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
if (SLIST_EMPTY(&c->children)) {
fc = feeder_getclass(NULL);
if (fc == NULL) {
DEB(printf("can't find root feeder\n"));
return EINVAL;
}
if (chn_addfeeder(c, fc, NULL)) {
DEB(printf("can't add root feeder\n"));
return EINVAL;
}
c->feeder->desc->out = c->format;
} else {
desc.type = FEEDER_MIXER;
desc.in = 0;
desc.out = c->format;
desc.flags = 0;
fc = feeder_getclass(&desc);
if (fc == NULL) {
DEB(printf("can't find vchan feeder\n"));
return EINVAL;
}
if (chn_addfeeder(c, fc, &desc)) {
DEB(printf("can't add vchan feeder\n"));
return EINVAL;
}
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
}
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
flags = c->feederflags;
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
if ((c->flags & CHN_F_MAPPED) && (flags != 0)) {
DEB(printf("can't build feeder chain on mapped channel\n"));
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
return EINVAL;
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
}
DEB(printf("not mapped, flags %x\n", flags));
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
for (type = FEEDER_RATE; type <= FEEDER_LAST; type++) {
if (flags & (1 << type)) {
desc.type = type;
desc.in = 0;
desc.out = 0;
desc.flags = 0;
DEB(printf("find feeder type %d, ", type));
kobjify. this gives us several benefits, including: * easier extensibility- new optional methods can be added to ac97/mixer/channel classes without having to fixup every driver. * forward compatibility for drivers, provided no new mandatory methods are added.
2000-12-18 01:36:41 +00:00
fc = feeder_getclass(&desc);
update code dealing with snd_dbuf objects to do so using a functional interface modify chn_setblocksize() to pick a default soft-blocksize appropriate to the sample rate and format in use. it will aim for a power of two size small enough to generate block sizes of at most 20ms. it will also set the hard-blocksize taking into account rate/format conversions in use. update drivers to implement setblocksize correctly: updated, tested: sb16, emu10k1, maestro, solo updated, untested: ad1816, ess, mss, sb8, csa not updated: ds1, es137x, fm801, neomagic, t4dwave, via82c686 i lack hardware to test: ad1816, csa, fm801, neomagic others will be updated/tested in the next few days.
2000-12-23 03:16:13 +00:00
DEB(printf("got %p\n", fc));
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
if (fc == NULL) {
DEB(printf("can't find required feeder type %d\n", type));
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
return EINVAL;
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
}
if (c->feeder->desc->out != fc->desc->in) {
DEB(printf("build fmtchain from %x to %x: ", c->feeder->desc->out, fc->desc->in));
tmp[0] = fc->desc->in;
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
tmp[1] = 0;
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
if (chn_fmtchain(c, tmp) == 0) {
DEB(printf("failed\n"));
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
return EINVAL;
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
}
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
DEB(printf("ok\n"));
}
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
if (chn_addfeeder(c, fc, fc->desc)) {
DEB(printf("can't add feeder %p, output %x\n", fc, fc->desc->out));
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
return EINVAL;
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
}
DEB(printf("added feeder %p, output %x\n", fc, c->feeder->desc->out));
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
}
}
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
make the feederchain builder work for recording. this has not been tested extensively as none of my testboxes have speakers or an audio source at present, but the chains built look correct and reading /dev/audio (ulaw, translated from signed 16 bit little-endian) gives values within the expected range for silence.
2002-01-23 05:10:56 +00:00
if (fmtvalid(c->feeder->desc->out, chn_getcaps(c)->fmtlist)) {
hwfmt = c->feeder->desc->out;
} else {
if (c->direction == PCMDIR_REC) {
tmp[0] = c->format;
tmp[1] = NULL;
hwfmt = chn_fmtchain(c, tmp);
} else {
hwfmt = chn_fmtchain(c, chn_getcaps(c)->fmtlist);
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
}
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
}
minor tweaks in speed and format setting routines. don't stop exploring the feeders if a feeder fails to initialise.
2001-04-08 20:20:52 +00:00
make the feederchain builder work for recording. this has not been tested extensively as none of my testboxes have speakers or an audio source at present, but the chains built look correct and reading /dev/audio (ulaw, translated from signed 16 bit little-endian) gives values within the expected range for silence.
2002-01-23 05:10:56 +00:00
if (hwfmt == 0)
return EINVAL;
sndbuf_setfmt(c->bufhard, hwfmt);
prepare for adding a rate conversion feeder. move format conversion feeders to feeder_fmt.c - no pertinent history so no repo-copy.
2000-09-23 22:11:32 +00:00
return 0;
}
mega-commit. this introduces a new buffering mechanism which results in dramatic simplification of the channel manager. as several structures have changed, we take the opportunity to move their definitions into the source files where they are used, make them private and de-typedef them. the sound drivers are updated to use snd_setup_intr instead of bus_setup_intr, and to comply with the de-typedefed structures. the ac97, mixer and channel layers have been updated with finegrained locking, as have some drivers- not all though. the rest will follow soon.
2001-03-24 23:10:29 +00:00
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
int
chn_notify(struct pcm_channel *c, u_int32_t flags)
{
struct pcmchan_children *pce;
struct pcm_channel *child;
lock sound device when adding/removing channels implement setblocksize for vchans don't panic when doing certain ioctls or aborting on a vchan xmms now works with vchans
2001-06-07 20:06:22 +00:00
int run;
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
if (SLIST_EMPTY(&c->children))
return ENODEV;
lock sound device when adding/removing channels implement setblocksize for vchans don't panic when doing certain ioctls or aborting on a vchan xmms now works with vchans
2001-06-07 20:06:22 +00:00
run = (c->flags & CHN_F_TRIGGERED)? 1 : 0;
/*
* if the hwchan is running, we can't change its rate, format or
* blocksize
*/
if (run)
flags &= CHN_N_VOLUME | CHN_N_TRIGGER;
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
if (flags & CHN_N_RATE) {
/*
* we could do something here, like scan children and decide on
* the most appropriate rate to mix at, but we don't for now
*/
}
if (flags & CHN_N_FORMAT) {
/*
* we could do something here, like scan children and decide on
* the most appropriate mixer feeder to use, but we don't for now
*/
}
if (flags & CHN_N_VOLUME) {
/*
* we could do something here but we don't for now
*/
}
if (flags & CHN_N_BLOCKSIZE) {
int blksz;
/*
* scan the children, find the lowest blocksize and use that
* for the hard blocksize
*/
blksz = sndbuf_getmaxsize(c->bufhard) / 2;
SLIST_FOREACH(pce, &c->children, link) {
child = pce->channel;
if (sndbuf_getblksz(child->bufhard) < blksz)
blksz = sndbuf_getblksz(child->bufhard);
}
chn_setblocksize(c, 2, blksz);
}
if (flags & CHN_N_TRIGGER) {
lock sound device when adding/removing channels implement setblocksize for vchans don't panic when doing certain ioctls or aborting on a vchan xmms now works with vchans
2001-06-07 20:06:22 +00:00
int nrun;
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
/*
* scan the children, and figure out if any are running
* if so, we need to be running, otherwise we need to be stopped
* if we aren't in our target sstate, move to it
*/
lock sound device when adding/removing channels implement setblocksize for vchans don't panic when doing certain ioctls or aborting on a vchan xmms now works with vchans
2001-06-07 20:06:22 +00:00
nrun = 0;
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
SLIST_FOREACH(pce, &c->children, link) {
child = pce->channel;
if (child->flags & CHN_F_TRIGGERED)
lock sound device when adding/removing channels implement setblocksize for vchans don't panic when doing certain ioctls or aborting on a vchan xmms now works with vchans
2001-06-07 20:06:22 +00:00
nrun = 1;
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
}
lock sound device when adding/removing channels implement setblocksize for vchans don't panic when doing certain ioctls or aborting on a vchan xmms now works with vchans
2001-06-07 20:06:22 +00:00
if (nrun && !run)
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
chn_start(c, 1);
lock sound device when adding/removing channels implement setblocksize for vchans don't panic when doing certain ioctls or aborting on a vchan xmms now works with vchans
2001-06-07 20:06:22 +00:00
if (!nrun && run)
beginnings of virtual playback channel support instead of using two malloced arrays for storing channel lists, use an slist. convert the sndstat device to use sbufs and optionally provide more detail about channel state. vchans are software mixed playback channels. they are not enabled by this commit. they use the feeder infrastructure to emulate normal playback channels in a manner transparent to applications, whilst providing as many channels are desired, especially suitable for devices with only one hardware playback channel. in the future they will provide additional features. those wishing to test this functionality will need to add vchan.c to sys/conf/files and use 'sysctl -w hw.snd.pcm0.vchans' to enable it. blocksize and auto-rate selection are not yet supported.
2001-05-27 17:22:00 +00:00
chn_abort(c);
}
return 0;
}
Reference in New Issue Copy Permalink