freebsd-dev/sys/dev/sound/pcm/sound.c

1528 lines
34 KiB
C
Raw Normal View History

/*-
2003-09-07 16:28:03 +00:00
* Copyright (c) 1999 Cameron Grant <cg@freebsd.org>
* (C) 1997 Luigi Rizzo
* 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.
*/
#include <dev/sound/pcm/sound.h>
Welcome to Once-a-year Sound Mega-Commit. Enjoy numerous updates and fixes in every sense. General ------- - Multichannel safe, endian safe, format safe * Large part of critical pcm filters such as vchan.c, feeder_rate.c, feeder_volume.c, feeder_fmt.c and feeder.c has been rewritten so that using them does not cause the pcm data to be converted to 16bit little endian. * Macrosses for accessing pcm data safely are defined within sound.h in the form of PCM_READ_* / PCM_WRITE_* * Currently, most of them are probably limited for mono/stereo handling, but the future addition of true multichannel will be much easier. - Low latency operation * Well, this require lot more works to do not just within sound driver, but we're heading towards right direction. Buffer/block sizing within channel.c is rewritten to calculate precise allocation for various combination of sample/data/rate size. As a result, applying correct SNDCTL_DSP_POLICY value will achive expected latency behaviour simmilar to what commercial 4front driver do. * Signal handling fix. ctrl+c of "cat /dev/zero > /dev/dsp" does not result long delay. * Eliminate sound truncation if the sound data is too small. DIY: 1) Download / extract http://people.freebsd.org/~ariff/lowlatency/shortfiles.tar.gz 2) Do a comparison between "cat state*.au > /dev/dsp" and "for x in state*.au ; do cat $x > /dev/dsp ; done" - there should be no "perceivable" differences. Double close for PR kern/31445. CAVEAT: Low latency come with (unbearable) price especially for poorly written applications. Applications that trying to act smarter by requesting (wrong) blocksize/blockcount will suffer the most. Fixup samples/patches can be found at: http://people.freebsd.org/~ariff/ports/ - Switch minimum/maximum sampling rate limit to "1" and "2016000" (48k * 42) due to closer compatibility with 4front driver. Discussed with: marcus@ (long time ago?) - All driver specific sysctls in the form of "hw.snd.pcm%d.*" have been moved to their own dev sysctl nodes, notably: hw.snd.pcm%d.vchans -> dev.pcm.%d.vchans Bump __FreeBSD_version. Driver specific --------------- - Ditto for sysctls. - snd_atiixp, snd_es137x, snd_via8233, snd_hda * Numerous cleanups and fixes. * _EXPERIMENTAL_ polling mode support using simple callout_* mechanisme. This was intended for pure debugging and latency measurement, but proven good enough in few unexpected and rare cases (such as problematic shared IRQ with GIANT devices - USB). Polling can be enabled/disabled through dev.pcm.0.polling. Disabled by default. - snd_ich * Fix possible overflow during speed calibration. Delay final initialization (pcm_setstatus) after calibration finished. PR: kern/100169 Tested by: Kevin Overman <oberman@es.net> * Inverted EAPD for few Nec VersaPro. PR: kern/104715 Submitted by: KAWATA Masahiko <kawata@mta.biglobe.ne.jp> Thanks to various people, notably Joel Dahl, Yuriy Tsibizov, Kevin Oberman, those at #freebsd-azalia @ freenode and others for testing. Joel Dahl will do the manpage update.
2006-11-26 12:24:06 +00:00
#include <dev/sound/pcm/ac97.h>
#include <dev/sound/pcm/vchan.h>
#include <dev/sound/pcm/dsp.h>
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
#include <dev/sound/version.h>
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
#include <sys/limits.h>
#include <sys/sysctl.h>
#include "feeder_if.h"
SND_DECLARE_FILE("$FreeBSD$");
devclass_t pcm_devclass;
2001-02-27 07:01:49 +00:00
int pcm_veto_load = 1;
2001-02-27 07:01:49 +00:00
#ifdef USING_DEVFS
int snd_unit = -1;
TUNABLE_INT("hw.snd.default_unit", &snd_unit);
2001-02-27 07:01:49 +00:00
#endif
static int snd_unit_auto = 0;
TUNABLE_INT("hw.snd.default_auto", &snd_unit_auto);
SYSCTL_INT(_hw_snd, OID_AUTO, default_auto, CTLFLAG_RW,
&snd_unit_auto, 0, "assign default unit to a newly attach device");
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
int snd_maxautovchans = 16;
/* XXX: a tunable implies that we may need more than one sound channel before
the system can change a sysctl (/etc/sysctl.conf), do we really need
this? */
TUNABLE_INT("hw.snd.maxautovchans", &snd_maxautovchans);
2001-02-27 07:01:49 +00:00
SYSCTL_NODE(_hw, OID_AUTO, snd, CTLFLAG_RD, 0, "Sound driver");
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
/*
* XXX I've had enough with people not telling proper version/arch
* while reporting problems, not after 387397913213th questions/requests.
*/
static const char snd_driver_version[] =
__XSTRING(SND_DRV_VERSION)"/"MACHINE_ARCH;
SYSCTL_STRING(_hw_snd, OID_AUTO, version, CTLFLAG_RD, &snd_driver_version,
0, "Driver version/arch");
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
/**
* @brief Unit number allocator for syncgroup IDs
*/
struct unrhdr *pcmsg_unrhdr = NULL;
static int sndstat_prepare_pcm(struct sbuf *s, device_t dev, int verbose);
void *
snd_mtxcreate(const char *desc, const char *type)
{
#ifdef USING_MUTEX
struct mtx *m;
m = malloc(sizeof(*m), M_DEVBUF, M_WAITOK | M_ZERO);
Change KASSERT() in feed_vchan16() into an explicit test and call to panic() so that the buffer overflow just beyond this point is always caught, even when the code is not compiled with INVARIANTS. Change chn_setblocksize() buffer reallocation code to attempt to avoid the feed_vchan16() buffer overflow by attempting to always keep the bufsoft buffer at least as large as the bufhard buffer. Print a diagnositic message Danger! %s bufsoft size increasing from %d to %d after CHANNEL_SETBLOCKSIZE() if our best attempts fail. If feed_vchan16() were to be called by the interrupt handler while locks are dropped in chn_setblocksize() to increase the size bufsoft to match the size of bufhard, the panic() code in feed_vchan16() will be triggered. If the diagnostic message is printed, it is a warning that a panic is possible if the system were to see events in an "unlucky" order. Change the locking code to avoid the need for MTX_RECURSIVE mutexes. Add the MTX_DUPOK option to the channel mutexes and change the locking sequence to always lock the parent channel before its children to avoid the possibility of deadlock. Actually implement locking assertions for the channel mutexes and fix the problems found by the resulting assertion violations. Clean up the locking code in dsp_ioctl(). Allocate the channel buffers using the malloc() M_WAITOK option instead of M_NOWAIT so that buffer allocation won't fail. Drop locks across the malloc() calls. Add/modify KASSERTS() in attempt to detect problems early. Abuse layering by adding a pointer to the snd_dbuf structure that points back to the pcm_channel that owns it. This allows sndbuf_resize() to do proper locking without having to change the its API, which is used by the hardware drivers. Don't dereference a NULL pointer when setting hw.snd.maxautovchans if a hardware driver is not loaded. Noticed by Ryan Sommers <ryans at gamersimpact.com>. Tested by: Stefan Ehmann <shoesoft AT gmx.net> Tested by: matk (Mathew Kanner) Tested by: Gordon Bergling <gbergling AT 0xfce3.net>
2004-01-28 08:02:15 +00:00
mtx_init(m, desc, type, MTX_DEF);
return m;
#else
return (void *)0xcafebabe;
#endif
}
void
snd_mtxfree(void *m)
{
#ifdef USING_MUTEX
struct mtx *mtx = m;
/* mtx_assert(mtx, MA_OWNED); */
mtx_destroy(mtx);
free(mtx, M_DEVBUF);
#endif
}
void
snd_mtxassert(void *m)
{
#ifdef USING_MUTEX
#ifdef INVARIANTS
struct mtx *mtx = m;
mtx_assert(mtx, MA_OWNED);
#endif
#endif
}
/*
void
snd_mtxlock(void *m)
{
#ifdef USING_MUTEX
struct mtx *mtx = m;
mtx_lock(mtx);
#endif
}
void
snd_mtxunlock(void *m)
{
#ifdef USING_MUTEX
struct mtx *mtx = m;
mtx_unlock(mtx);
#endif
}
*/
int
snd_setup_intr(device_t dev, struct resource *res, int flags, driver_intr_t hand, void *param, void **cookiep)
{
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
struct snddev_info *d;
#ifdef USING_MUTEX
flags &= INTR_MPSAFE;
flags |= INTR_TYPE_AV;
#else
flags = INTR_TYPE_AV;
#endif
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
d = device_get_softc(dev);
if (d != NULL && (flags & INTR_MPSAFE))
d->flags |= SD_F_MPSAFE;
return bus_setup_intr(dev, res, flags,
#if __FreeBSD_version >= 700031
NULL,
#endif
hand, param, cookiep);
}
#ifndef PCM_DEBUG_MTX
void
pcm_lock(struct snddev_info *d)
{
snd_mtxlock(d->lock);
}
void
pcm_unlock(struct snddev_info *d)
{
snd_mtxunlock(d->lock);
}
#endif
struct pcm_channel *
pcm_getfakechan(struct snddev_info *d)
{
return d->fakechan;
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
static void
pcm_clonereset(struct snddev_info *d)
{
int cmax;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
cmax = d->playcount + d->reccount - 1;
if (d->pvchancount > 0)
cmax += MAX(d->pvchancount, snd_maxautovchans) - 1;
if (d->rvchancount > 0)
cmax += MAX(d->rvchancount, snd_maxautovchans) - 1;
if (cmax > PCMMAXCLONE)
cmax = PCMMAXCLONE;
(void)snd_clone_gc(d->clones);
(void)snd_clone_setmaxunit(d->clones, cmax);
}
static int
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
pcm_setvchans(struct snddev_info *d, int direction, int newcnt, int num)
{
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
struct pcm_channel *c, *ch, *nch;
int err, vcnt;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if ((direction == PCMDIR_PLAY && d->playcount < 1) ||
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
(direction == PCMDIR_REC && d->reccount < 1))
return (ENODEV);
Welcome to Once-a-year Sound Mega-Commit. Enjoy numerous updates and fixes in every sense. General ------- - Multichannel safe, endian safe, format safe * Large part of critical pcm filters such as vchan.c, feeder_rate.c, feeder_volume.c, feeder_fmt.c and feeder.c has been rewritten so that using them does not cause the pcm data to be converted to 16bit little endian. * Macrosses for accessing pcm data safely are defined within sound.h in the form of PCM_READ_* / PCM_WRITE_* * Currently, most of them are probably limited for mono/stereo handling, but the future addition of true multichannel will be much easier. - Low latency operation * Well, this require lot more works to do not just within sound driver, but we're heading towards right direction. Buffer/block sizing within channel.c is rewritten to calculate precise allocation for various combination of sample/data/rate size. As a result, applying correct SNDCTL_DSP_POLICY value will achive expected latency behaviour simmilar to what commercial 4front driver do. * Signal handling fix. ctrl+c of "cat /dev/zero > /dev/dsp" does not result long delay. * Eliminate sound truncation if the sound data is too small. DIY: 1) Download / extract http://people.freebsd.org/~ariff/lowlatency/shortfiles.tar.gz 2) Do a comparison between "cat state*.au > /dev/dsp" and "for x in state*.au ; do cat $x > /dev/dsp ; done" - there should be no "perceivable" differences. Double close for PR kern/31445. CAVEAT: Low latency come with (unbearable) price especially for poorly written applications. Applications that trying to act smarter by requesting (wrong) blocksize/blockcount will suffer the most. Fixup samples/patches can be found at: http://people.freebsd.org/~ariff/ports/ - Switch minimum/maximum sampling rate limit to "1" and "2016000" (48k * 42) due to closer compatibility with 4front driver. Discussed with: marcus@ (long time ago?) - All driver specific sysctls in the form of "hw.snd.pcm%d.*" have been moved to their own dev sysctl nodes, notably: hw.snd.pcm%d.vchans -> dev.pcm.%d.vchans Bump __FreeBSD_version. Driver specific --------------- - Ditto for sysctls. - snd_atiixp, snd_es137x, snd_via8233, snd_hda * Numerous cleanups and fixes. * _EXPERIMENTAL_ polling mode support using simple callout_* mechanisme. This was intended for pure debugging and latency measurement, but proven good enough in few unexpected and rare cases (such as problematic shared IRQ with GIANT devices - USB). Polling can be enabled/disabled through dev.pcm.0.polling. Disabled by default. - snd_ich * Fix possible overflow during speed calibration. Delay final initialization (pcm_setstatus) after calibration finished. PR: kern/100169 Tested by: Kevin Overman <oberman@es.net> * Inverted EAPD for few Nec VersaPro. PR: kern/104715 Submitted by: KAWATA Masahiko <kawata@mta.biglobe.ne.jp> Thanks to various people, notably Joel Dahl, Yuriy Tsibizov, Kevin Oberman, those at #freebsd-azalia @ freenode and others for testing. Joel Dahl will do the manpage update.
2006-11-26 12:24:06 +00:00
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (!(d->flags & SD_F_AUTOVCHAN))
return (EINVAL);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (newcnt < 0 || newcnt > SND_MAXVCHANS)
return (E2BIG);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (direction == PCMDIR_PLAY)
vcnt = d->pvchancount;
else if (direction == PCMDIR_REC)
vcnt = d->rvchancount;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
else
return (EINVAL);
if (newcnt > vcnt) {
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
KASSERT(num == -1 ||
(num >= 0 && num < SND_MAXVCHANS && (newcnt - 1) == vcnt),
("bogus vchan_create() request num=%d newcnt=%d vcnt=%d",
num, newcnt, vcnt));
/* add new vchans - find a parent channel first */
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
ch = NULL;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
CHN_FOREACH(c, d, channels.pcm) {
CHN_LOCK(c);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (c->direction == direction &&
((c->flags & CHN_F_HAS_VCHAN) || (vcnt == 0 &&
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
!(c->flags & (CHN_F_BUSY | CHN_F_VIRTUAL))))) {
ch = c;
break;
}
CHN_UNLOCK(c);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (ch == NULL)
return (EBUSY);
ch->flags |= CHN_F_BUSY;
err = 0;
while (err == 0 && newcnt > vcnt) {
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
err = vchan_create(ch, num);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (err == 0)
vcnt++;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
else if (err == E2BIG && newcnt > vcnt)
device_printf(d->dev,
"%s: err=%d Maximum channel reached.\n",
__func__, err);
}
if (vcnt == 0)
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
ch->flags &= ~CHN_F_BUSY;
CHN_UNLOCK(ch);
if (err != 0)
return (err);
else
pcm_clonereset(d);
} else if (newcnt < vcnt) {
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
KASSERT(num == -1,
("bogus vchan_destroy() request num=%d", num));
CHN_FOREACH(c, d, channels.pcm) {
CHN_LOCK(c);
if (c->direction != direction ||
CHN_EMPTY(c, children) ||
!(c->flags & CHN_F_HAS_VCHAN)) {
CHN_UNLOCK(c);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
continue;
}
CHN_FOREACH_SAFE(ch, c, nch, children) {
CHN_LOCK(ch);
if (!(ch->flags & CHN_F_BUSY)) {
CHN_UNLOCK(ch);
CHN_UNLOCK(c);
err = vchan_destroy(ch);
CHN_LOCK(c);
if (err == 0)
vcnt--;
} else
CHN_UNLOCK(ch);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (vcnt == newcnt)
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
break;
}
CHN_UNLOCK(c);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
break;
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
pcm_clonereset(d);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (0);
}
/* return error status and a locked channel */
int
pcm_chnalloc(struct snddev_info *d, struct pcm_channel **ch, int direction,
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
pid_t pid, int devunit)
{
struct pcm_channel *c;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
int err, vchancount;
KASSERT(d != NULL && ch != NULL && (devunit == -1 ||
!(devunit & ~(SND_U_MASK | SND_D_MASK | SND_C_MASK))) &&
(direction == PCMDIR_PLAY || direction == PCMDIR_REC),
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
("%s(): invalid d=%p ch=%p direction=%d pid=%d devunit=%d",
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
__func__, d, ch, direction, pid, devunit));
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
/* Double check again. */
if (devunit != -1) {
switch (snd_unit2d(devunit)) {
case SND_DEV_DSPHW_PLAY:
case SND_DEV_DSPHW_VPLAY:
if (direction != PCMDIR_PLAY)
return (EOPNOTSUPP);
break;
case SND_DEV_DSPHW_REC:
case SND_DEV_DSPHW_VREC:
if (direction != PCMDIR_REC)
return (EOPNOTSUPP);
break;
default:
if (!(direction == PCMDIR_PLAY ||
direction == PCMDIR_REC))
return (EOPNOTSUPP);
break;
}
}
retry_chnalloc:
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
err = EOPNOTSUPP;
/* scan for a free channel */
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
CHN_FOREACH(c, d, channels.pcm) {
CHN_LOCK(c);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (c->direction == direction && !(c->flags & CHN_F_BUSY) &&
(devunit == -1 || devunit == -2 || c->unit == devunit)) {
c->flags |= CHN_F_BUSY;
c->pid = pid;
*ch = c;
return (0);
} else if (c->unit == devunit) {
if (c->direction != direction)
err = EOPNOTSUPP;
else if (c->flags & CHN_F_BUSY)
err = EBUSY;
else
err = EINVAL;
CHN_UNLOCK(c);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
return (err);
} else if ((devunit == -1 || devunit == -2) &&
c->direction == direction && (c->flags & CHN_F_BUSY))
err = EBUSY;
CHN_UNLOCK(c);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (devunit == -2)
return (err);
/* no channel available */
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (devunit == -1 || snd_unit2d(devunit) == SND_DEV_DSPHW_VPLAY ||
snd_unit2d(devunit) == SND_DEV_DSPHW_VREC) {
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (direction == PCMDIR_PLAY)
vchancount = d->pvchancount;
else
vchancount = d->rvchancount;
if (!(vchancount > 0 && vchancount < snd_maxautovchans) &&
(devunit == -1 || snd_unit2c(devunit) < snd_maxautovchans))
return (err);
err = pcm_setvchans(d, direction, vchancount + 1,
(devunit == -1) ? -1 : snd_unit2c(devunit));
if (err == 0) {
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (devunit == -1)
devunit = -2;
goto retry_chnalloc;
}
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
return (err);
}
/* release a locked channel and unlock it */
int
pcm_chnrelease(struct pcm_channel *c)
{
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(c->parentsnddev);
CHN_LOCKASSERT(c);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
c->flags &= ~CHN_F_BUSY;
c->pid = -1;
CHN_UNLOCK(c);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (0);
}
int
pcm_chnref(struct pcm_channel *c, int ref)
{
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(c->parentsnddev);
CHN_LOCKASSERT(c);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
c->refcount += ref;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (c->refcount);
}
int
pcm_inprog(struct snddev_info *d, int delta)
{
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
snd_mtxassert(d->lock);
d->inprog += delta;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (d->inprog);
}
static void
pcm_setmaxautovchans(struct snddev_info *d, int num)
{
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (num < 0)
return;
if (num >= 0 && d->pvchancount > num)
(void)pcm_setvchans(d, PCMDIR_PLAY, num, -1);
else if (num > 0 && d->pvchancount == 0)
(void)pcm_setvchans(d, PCMDIR_PLAY, 1, -1);
if (num >= 0 && d->rvchancount > num)
(void)pcm_setvchans(d, PCMDIR_REC, num, -1);
else if (num > 0 && d->rvchancount == 0)
(void)pcm_setvchans(d, PCMDIR_REC, 1, -1);
pcm_clonereset(d);
}
#ifdef USING_DEVFS
static int
sysctl_hw_snd_default_unit(SYSCTL_HANDLER_ARGS)
{
struct snddev_info *d;
int error, unit;
unit = snd_unit;
error = sysctl_handle_int(oidp, &unit, 0, req);
if (error == 0 && req->newptr != NULL) {
d = devclass_get_softc(pcm_devclass, unit);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (!PCM_REGISTERED(d) || CHN_EMPTY(d, channels.pcm))
return EINVAL;
snd_unit = unit;
}
return (error);
}
/* XXX: do we need a way to let the user change the default unit? */
SYSCTL_PROC(_hw_snd, OID_AUTO, default_unit, CTLTYPE_INT | CTLFLAG_RW,
Welcome to Once-a-year Sound Mega-Commit. Enjoy numerous updates and fixes in every sense. General ------- - Multichannel safe, endian safe, format safe * Large part of critical pcm filters such as vchan.c, feeder_rate.c, feeder_volume.c, feeder_fmt.c and feeder.c has been rewritten so that using them does not cause the pcm data to be converted to 16bit little endian. * Macrosses for accessing pcm data safely are defined within sound.h in the form of PCM_READ_* / PCM_WRITE_* * Currently, most of them are probably limited for mono/stereo handling, but the future addition of true multichannel will be much easier. - Low latency operation * Well, this require lot more works to do not just within sound driver, but we're heading towards right direction. Buffer/block sizing within channel.c is rewritten to calculate precise allocation for various combination of sample/data/rate size. As a result, applying correct SNDCTL_DSP_POLICY value will achive expected latency behaviour simmilar to what commercial 4front driver do. * Signal handling fix. ctrl+c of "cat /dev/zero > /dev/dsp" does not result long delay. * Eliminate sound truncation if the sound data is too small. DIY: 1) Download / extract http://people.freebsd.org/~ariff/lowlatency/shortfiles.tar.gz 2) Do a comparison between "cat state*.au > /dev/dsp" and "for x in state*.au ; do cat $x > /dev/dsp ; done" - there should be no "perceivable" differences. Double close for PR kern/31445. CAVEAT: Low latency come with (unbearable) price especially for poorly written applications. Applications that trying to act smarter by requesting (wrong) blocksize/blockcount will suffer the most. Fixup samples/patches can be found at: http://people.freebsd.org/~ariff/ports/ - Switch minimum/maximum sampling rate limit to "1" and "2016000" (48k * 42) due to closer compatibility with 4front driver. Discussed with: marcus@ (long time ago?) - All driver specific sysctls in the form of "hw.snd.pcm%d.*" have been moved to their own dev sysctl nodes, notably: hw.snd.pcm%d.vchans -> dev.pcm.%d.vchans Bump __FreeBSD_version. Driver specific --------------- - Ditto for sysctls. - snd_atiixp, snd_es137x, snd_via8233, snd_hda * Numerous cleanups and fixes. * _EXPERIMENTAL_ polling mode support using simple callout_* mechanisme. This was intended for pure debugging and latency measurement, but proven good enough in few unexpected and rare cases (such as problematic shared IRQ with GIANT devices - USB). Polling can be enabled/disabled through dev.pcm.0.polling. Disabled by default. - snd_ich * Fix possible overflow during speed calibration. Delay final initialization (pcm_setstatus) after calibration finished. PR: kern/100169 Tested by: Kevin Overman <oberman@es.net> * Inverted EAPD for few Nec VersaPro. PR: kern/104715 Submitted by: KAWATA Masahiko <kawata@mta.biglobe.ne.jp> Thanks to various people, notably Joel Dahl, Yuriy Tsibizov, Kevin Oberman, those at #freebsd-azalia @ freenode and others for testing. Joel Dahl will do the manpage update.
2006-11-26 12:24:06 +00:00
0, sizeof(int), sysctl_hw_snd_default_unit, "I", "default sound device");
2001-02-27 07:01:49 +00:00
#endif
static int
sysctl_hw_snd_maxautovchans(SYSCTL_HANDLER_ARGS)
{
struct snddev_info *d;
int i, v, error;
v = snd_maxautovchans;
error = sysctl_handle_int(oidp, &v, 0, req);
if (error == 0 && req->newptr != NULL) {
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (v < 0)
v = 0;
if (v > SND_MAXVCHANS)
v = SND_MAXVCHANS;
snd_maxautovchans = v;
for (i = 0; pcm_devclass != NULL &&
i < devclass_get_maxunit(pcm_devclass); i++) {
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
d = devclass_get_softc(pcm_devclass, i);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (!PCM_REGISTERED(d))
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
continue;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_ACQUIRE_QUICK(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
pcm_setmaxautovchans(d, v);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_RELEASE_QUICK(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
}
}
return (error);
}
SYSCTL_PROC(_hw_snd, OID_AUTO, maxautovchans, CTLTYPE_INT | CTLFLAG_RW,
Welcome to Once-a-year Sound Mega-Commit. Enjoy numerous updates and fixes in every sense. General ------- - Multichannel safe, endian safe, format safe * Large part of critical pcm filters such as vchan.c, feeder_rate.c, feeder_volume.c, feeder_fmt.c and feeder.c has been rewritten so that using them does not cause the pcm data to be converted to 16bit little endian. * Macrosses for accessing pcm data safely are defined within sound.h in the form of PCM_READ_* / PCM_WRITE_* * Currently, most of them are probably limited for mono/stereo handling, but the future addition of true multichannel will be much easier. - Low latency operation * Well, this require lot more works to do not just within sound driver, but we're heading towards right direction. Buffer/block sizing within channel.c is rewritten to calculate precise allocation for various combination of sample/data/rate size. As a result, applying correct SNDCTL_DSP_POLICY value will achive expected latency behaviour simmilar to what commercial 4front driver do. * Signal handling fix. ctrl+c of "cat /dev/zero > /dev/dsp" does not result long delay. * Eliminate sound truncation if the sound data is too small. DIY: 1) Download / extract http://people.freebsd.org/~ariff/lowlatency/shortfiles.tar.gz 2) Do a comparison between "cat state*.au > /dev/dsp" and "for x in state*.au ; do cat $x > /dev/dsp ; done" - there should be no "perceivable" differences. Double close for PR kern/31445. CAVEAT: Low latency come with (unbearable) price especially for poorly written applications. Applications that trying to act smarter by requesting (wrong) blocksize/blockcount will suffer the most. Fixup samples/patches can be found at: http://people.freebsd.org/~ariff/ports/ - Switch minimum/maximum sampling rate limit to "1" and "2016000" (48k * 42) due to closer compatibility with 4front driver. Discussed with: marcus@ (long time ago?) - All driver specific sysctls in the form of "hw.snd.pcm%d.*" have been moved to their own dev sysctl nodes, notably: hw.snd.pcm%d.vchans -> dev.pcm.%d.vchans Bump __FreeBSD_version. Driver specific --------------- - Ditto for sysctls. - snd_atiixp, snd_es137x, snd_via8233, snd_hda * Numerous cleanups and fixes. * _EXPERIMENTAL_ polling mode support using simple callout_* mechanisme. This was intended for pure debugging and latency measurement, but proven good enough in few unexpected and rare cases (such as problematic shared IRQ with GIANT devices - USB). Polling can be enabled/disabled through dev.pcm.0.polling. Disabled by default. - snd_ich * Fix possible overflow during speed calibration. Delay final initialization (pcm_setstatus) after calibration finished. PR: kern/100169 Tested by: Kevin Overman <oberman@es.net> * Inverted EAPD for few Nec VersaPro. PR: kern/104715 Submitted by: KAWATA Masahiko <kawata@mta.biglobe.ne.jp> Thanks to various people, notably Joel Dahl, Yuriy Tsibizov, Kevin Oberman, those at #freebsd-azalia @ freenode and others for testing. Joel Dahl will do the manpage update.
2006-11-26 12:24:06 +00:00
0, sizeof(int), sysctl_hw_snd_maxautovchans, "I", "maximum virtual channel");
struct pcm_channel *
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
pcm_chn_create(struct snddev_info *d, struct pcm_channel *parent, kobj_class_t cls, int dir, int num, void *devinfo)
{
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
struct pcm_channel *ch;
int direction, err, rpnum, *pnum, max;
int udc, device, chan;
char *dirs, *devname, buf[CHN_NAMELEN];
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(d);
snd_mtxassert(d->lock);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
KASSERT(num >= -1, ("invalid num=%d", num));
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
switch (dir) {
case PCMDIR_PLAY:
dirs = "play";
direction = PCMDIR_PLAY;
pnum = &d->playcount;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
device = SND_DEV_DSPHW_PLAY;
max = SND_MAXHWCHAN;
break;
case PCMDIR_PLAY_VIRTUAL:
dirs = "virtual";
direction = PCMDIR_PLAY;
pnum = &d->pvchancount;
device = SND_DEV_DSPHW_VPLAY;
max = SND_MAXVCHANS;
break;
case PCMDIR_REC:
dirs = "record";
direction = PCMDIR_REC;
pnum = &d->reccount;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
device = SND_DEV_DSPHW_REC;
max = SND_MAXHWCHAN;
break;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
case PCMDIR_REC_VIRTUAL:
dirs = "virtual";
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
direction = PCMDIR_REC;
pnum = &d->rvchancount;
device = SND_DEV_DSPHW_VREC;
max = SND_MAXVCHANS;
break;
default:
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (NULL);
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
chan = (num == -1) ? 0 : num;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (*pnum >= max || chan >= max)
return (NULL);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
rpnum = 0;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
CHN_FOREACH(ch, d, channels.pcm) {
if (CHN_DEV(ch) != device)
continue;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (chan == CHN_CHAN(ch)) {
if (num != -1) {
device_printf(d->dev,
"channel num=%d allocated!\n", chan);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (NULL);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
}
chan++;
if (chan >= max) {
device_printf(d->dev,
"chan=%d > %d\n", chan, max);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (NULL);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
}
}
rpnum++;
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (*pnum != rpnum) {
device_printf(d->dev,
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
"%s(): WARNING: pnum screwed : dirs=%s pnum=%d rpnum=%d\n",
__func__, dirs, *pnum, rpnum);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (NULL);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
}
udc = snd_mkunit(device_get_unit(d->dev), device, chan);
devname = dsp_unit2name(buf, sizeof(buf), udc);
if (devname == NULL) {
device_printf(d->dev,
"Failed to query device name udc=0x%08x\n", udc);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (NULL);
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
pcm_unlock(d);
ch = malloc(sizeof(*ch), M_DEVBUF, M_WAITOK | M_ZERO);
ch->methods = kobj_create(cls, M_DEVBUF, M_WAITOK | M_ZERO);
ch->unit = udc;
ch->pid = -1;
ch->parentsnddev = d;
ch->parentchannel = parent;
ch->dev = d->dev;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
ch->trigger = PCMTRIG_STOP;
snprintf(ch->name, sizeof(ch->name), "%s:%s:%s",
device_get_nameunit(ch->dev), dirs, devname);
err = chn_init(ch, devinfo, dir, direction);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
pcm_lock(d);
if (err) {
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
device_printf(d->dev, "chn_init(%s) failed: err = %d\n",
ch->name, err);
kobj_delete(ch->methods, M_DEVBUF);
free(ch, M_DEVBUF);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (NULL);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (ch);
}
int
pcm_chn_destroy(struct pcm_channel *ch)
{
struct snddev_info *d;
int err;
d = ch->parentsnddev;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(d);
err = chn_kill(ch);
if (err) {
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
device_printf(ch->dev, "chn_kill(%s) failed, err = %d\n",
ch->name, err);
return (err);
}
kobj_delete(ch->methods, M_DEVBUF);
free(ch, M_DEVBUF);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (0);
}
int
pcm_chn_add(struct snddev_info *d, struct pcm_channel *ch)
{
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
struct pcm_channel *tmp, *after;
int num;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(d);
snd_mtxassert(d->lock);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
KASSERT(ch != NULL && (ch->direction == PCMDIR_PLAY ||
ch->direction == PCMDIR_REC), ("Invalid pcm channel"));
after = NULL;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
tmp = NULL;
num = 0;
/*
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
* Look for possible device collision.
*/
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
CHN_FOREACH(tmp, d, channels.pcm) {
if (tmp->unit == ch->unit) {
device_printf(d->dev, "%s(): Device collision "
"old=%p new=%p devunit=0x%08x\n",
__func__, tmp, ch, ch->unit);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (ENODEV);
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (CHN_DEV(tmp) < CHN_DEV(ch)) {
if (num == 0)
after = tmp;
continue;
} else if (CHN_DEV(tmp) > CHN_DEV(ch))
break;
num++;
if (CHN_CHAN(tmp) < CHN_CHAN(ch))
after = tmp;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
else if (CHN_CHAN(tmp) > CHN_CHAN(ch))
break;
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (after != NULL) {
CHN_INSERT_AFTER(after, ch, channels.pcm);
} else {
CHN_INSERT_HEAD(d, ch, channels.pcm);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
switch (CHN_DEV(ch)) {
case SND_DEV_DSPHW_PLAY:
d->playcount++;
break;
case SND_DEV_DSPHW_VPLAY:
d->pvchancount++;
break;
case SND_DEV_DSPHW_REC:
d->reccount++;
break;
case SND_DEV_DSPHW_VREC:
d->rvchancount++;
break;
default:
break;
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
d->devcount++;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (0);
}
int
pcm_chn_remove(struct snddev_info *d, struct pcm_channel *ch)
{
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
struct pcm_channel *tmp;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(d);
snd_mtxassert(d->lock);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
tmp = NULL;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
CHN_FOREACH(tmp, d, channels.pcm) {
if (tmp == ch)
break;
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (tmp != ch)
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (EINVAL);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
CHN_REMOVE(d, ch, channels.pcm);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
switch (CHN_DEV(ch)) {
case SND_DEV_DSPHW_PLAY:
d->playcount--;
break;
case SND_DEV_DSPHW_VPLAY:
d->pvchancount--;
break;
case SND_DEV_DSPHW_REC:
d->reccount--;
break;
case SND_DEV_DSPHW_VREC:
d->rvchancount--;
break;
default:
break;
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
d->devcount--;
return (0);
}
int
pcm_addchan(device_t dev, int dir, kobj_class_t cls, void *devinfo)
{
struct snddev_info *d = device_get_softc(dev);
struct pcm_channel *ch;
int err;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(d);
pcm_lock(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
ch = pcm_chn_create(d, NULL, cls, dir, -1, devinfo);
if (!ch) {
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
device_printf(d->dev, "pcm_chn_create(%s, %d, %p) failed\n",
cls->name, dir, devinfo);
pcm_unlock(d);
return (ENODEV);
}
err = pcm_chn_add(d, ch);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
pcm_unlock(d);
if (err) {
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
device_printf(d->dev, "pcm_chn_add(%s) failed, err=%d\n",
ch->name, err);
pcm_chn_destroy(ch);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (err);
}
static int
pcm_killchan(device_t dev)
{
struct snddev_info *d = device_get_softc(dev);
struct pcm_channel *ch;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
int error;
PCM_BUSYASSERT(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
ch = CHN_FIRST(d, channels.pcm);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
pcm_lock(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
error = pcm_chn_remove(d, ch);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
pcm_unlock(d);
if (error)
return (error);
return (pcm_chn_destroy(ch));
}
int
pcm_setstatus(device_t dev, char *str)
{
struct snddev_info *d = device_get_softc(dev);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (d->playcount == 0 || d->reccount == 0)
d->flags |= SD_F_SIMPLEX;
if ((d->playcount > 0 || d->reccount > 0) &&
!(d->flags & SD_F_AUTOVCHAN)) {
d->flags |= SD_F_AUTOVCHAN;
vchan_initsys(dev);
}
pcm_setmaxautovchans(d, snd_maxautovchans);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
Welcome to Once-a-year Sound Mega-Commit. Enjoy numerous updates and fixes in every sense. General ------- - Multichannel safe, endian safe, format safe * Large part of critical pcm filters such as vchan.c, feeder_rate.c, feeder_volume.c, feeder_fmt.c and feeder.c has been rewritten so that using them does not cause the pcm data to be converted to 16bit little endian. * Macrosses for accessing pcm data safely are defined within sound.h in the form of PCM_READ_* / PCM_WRITE_* * Currently, most of them are probably limited for mono/stereo handling, but the future addition of true multichannel will be much easier. - Low latency operation * Well, this require lot more works to do not just within sound driver, but we're heading towards right direction. Buffer/block sizing within channel.c is rewritten to calculate precise allocation for various combination of sample/data/rate size. As a result, applying correct SNDCTL_DSP_POLICY value will achive expected latency behaviour simmilar to what commercial 4front driver do. * Signal handling fix. ctrl+c of "cat /dev/zero > /dev/dsp" does not result long delay. * Eliminate sound truncation if the sound data is too small. DIY: 1) Download / extract http://people.freebsd.org/~ariff/lowlatency/shortfiles.tar.gz 2) Do a comparison between "cat state*.au > /dev/dsp" and "for x in state*.au ; do cat $x > /dev/dsp ; done" - there should be no "perceivable" differences. Double close for PR kern/31445. CAVEAT: Low latency come with (unbearable) price especially for poorly written applications. Applications that trying to act smarter by requesting (wrong) blocksize/blockcount will suffer the most. Fixup samples/patches can be found at: http://people.freebsd.org/~ariff/ports/ - Switch minimum/maximum sampling rate limit to "1" and "2016000" (48k * 42) due to closer compatibility with 4front driver. Discussed with: marcus@ (long time ago?) - All driver specific sysctls in the form of "hw.snd.pcm%d.*" have been moved to their own dev sysctl nodes, notably: hw.snd.pcm%d.vchans -> dev.pcm.%d.vchans Bump __FreeBSD_version. Driver specific --------------- - Ditto for sysctls. - snd_atiixp, snd_es137x, snd_via8233, snd_hda * Numerous cleanups and fixes. * _EXPERIMENTAL_ polling mode support using simple callout_* mechanisme. This was intended for pure debugging and latency measurement, but proven good enough in few unexpected and rare cases (such as problematic shared IRQ with GIANT devices - USB). Polling can be enabled/disabled through dev.pcm.0.polling. Disabled by default. - snd_ich * Fix possible overflow during speed calibration. Delay final initialization (pcm_setstatus) after calibration finished. PR: kern/100169 Tested by: Kevin Overman <oberman@es.net> * Inverted EAPD for few Nec VersaPro. PR: kern/104715 Submitted by: KAWATA Masahiko <kawata@mta.biglobe.ne.jp> Thanks to various people, notably Joel Dahl, Yuriy Tsibizov, Kevin Oberman, those at #freebsd-azalia @ freenode and others for testing. Joel Dahl will do the manpage update.
2006-11-26 12:24:06 +00:00
strlcpy(d->status, str, SND_STATUSLEN);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
pcm_lock(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
/* Last stage, enable cloning. */
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (d->clones != NULL)
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
(void)snd_clone_enable(d->clones);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
/* Done, we're ready.. */
d->flags |= SD_F_REGISTERED;
PCM_RELEASE(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
pcm_unlock(d);
if (snd_unit < 0 || snd_unit_auto != 0)
snd_unit = device_get_unit(dev);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (0);
}
uint32_t
pcm_getflags(device_t dev)
{
struct snddev_info *d = device_get_softc(dev);
return d->flags;
}
void
pcm_setflags(device_t dev, uint32_t val)
{
struct snddev_info *d = device_get_softc(dev);
d->flags = val;
}
void *
pcm_getdevinfo(device_t dev)
{
struct snddev_info *d = device_get_softc(dev);
return d->devinfo;
}
unsigned int
pcm_getbuffersize(device_t dev, unsigned int minbufsz, unsigned int deflt, unsigned int maxbufsz)
{
struct snddev_info *d = device_get_softc(dev);
int sz, x;
sz = 0;
if (resource_int_value(device_get_name(dev), device_get_unit(dev), "buffersize", &sz) == 0) {
x = sz;
RANGE(sz, minbufsz, maxbufsz);
if (x != sz)
device_printf(dev, "'buffersize=%d' hint is out of range (%d-%d), using %d\n", x, minbufsz, maxbufsz, sz);
x = minbufsz;
while (x < sz)
x <<= 1;
if (x > sz)
x >>= 1;
if (x != sz) {
device_printf(dev, "'buffersize=%d' hint is not a power of 2, using %d\n", sz, x);
sz = x;
}
} else {
sz = deflt;
}
d->bufsz = sz;
return sz;
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
#if defined(SND_DYNSYSCTL) && defined(SND_DEBUG)
static int
sysctl_dev_pcm_clone_flags(SYSCTL_HANDLER_ARGS)
{
struct snddev_info *d;
uint32_t flags;
int err;
d = oidp->oid_arg1;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (!PCM_REGISTERED(d) || d->clones == NULL)
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
return (ENODEV);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_ACQUIRE_QUICK(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
flags = snd_clone_getflags(d->clones);
err = sysctl_handle_int(oidp, &flags, 0, req);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (err == 0 && req->newptr != NULL) {
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (flags & ~SND_CLONE_MASK)
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
err = EINVAL;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
else
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
(void)snd_clone_setflags(d->clones, flags);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_RELEASE_QUICK(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
return (err);
}
static int
sysctl_dev_pcm_clone_deadline(SYSCTL_HANDLER_ARGS)
{
struct snddev_info *d;
int err, deadline;
d = oidp->oid_arg1;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (!PCM_REGISTERED(d) || d->clones == NULL)
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
return (ENODEV);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_ACQUIRE_QUICK(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
deadline = snd_clone_getdeadline(d->clones);
err = sysctl_handle_int(oidp, &deadline, 0, req);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (err == 0 && req->newptr != NULL) {
if (deadline < 0)
err = EINVAL;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
else
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
(void)snd_clone_setdeadline(d->clones, deadline);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_RELEASE_QUICK(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
return (err);
}
static int
sysctl_dev_pcm_clone_gc(SYSCTL_HANDLER_ARGS)
{
struct snddev_info *d;
int err, val;
d = oidp->oid_arg1;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (!PCM_REGISTERED(d) || d->clones == NULL)
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
return (ENODEV);
val = 0;
err = sysctl_handle_int(oidp, &val, 0, req);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (err == 0 && req->newptr != NULL && val != 0) {
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_ACQUIRE_QUICK(d);
val = snd_clone_gc(d->clones);
PCM_RELEASE_QUICK(d);
if (bootverbose != 0 || snd_verbose > 3)
device_printf(d->dev, "clone gc: pruned=%d\n", val);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
}
return (err);
}
static int
sysctl_hw_snd_clone_gc(SYSCTL_HANDLER_ARGS)
{
struct snddev_info *d;
int i, err, val;
val = 0;
err = sysctl_handle_int(oidp, &val, 0, req);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (err == 0 && req->newptr != NULL && val != 0) {
for (i = 0; pcm_devclass != NULL &&
i < devclass_get_maxunit(pcm_devclass); i++) {
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
d = devclass_get_softc(pcm_devclass, i);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (!PCM_REGISTERED(d) || d->clones == NULL)
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
continue;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_ACQUIRE_QUICK(d);
val = snd_clone_gc(d->clones);
PCM_RELEASE_QUICK(d);
if (bootverbose != 0 || snd_verbose > 3)
device_printf(d->dev, "clone gc: pruned=%d\n",
val);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
}
}
return (err);
}
SYSCTL_PROC(_hw_snd, OID_AUTO, clone_gc, CTLTYPE_INT | CTLFLAG_RW,
0, sizeof(int), sysctl_hw_snd_clone_gc, "I",
"global clone garbage collector");
#endif
int
pcm_register(device_t dev, void *devinfo, int numplay, int numrec)
{
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
struct snddev_info *d;
if (pcm_veto_load) {
device_printf(dev, "disabled due to an error while initialising: %d\n", pcm_veto_load);
return EINVAL;
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (device_get_unit(dev) > PCMMAXUNIT) {
device_printf(dev, "PCMMAXUNIT reached : unit=%d > %d\n",
device_get_unit(dev), PCMMAXUNIT);
device_printf(dev,
"Use 'hw.snd.maxunit' tunable to raise the limit.\n");
return ENODEV;
}
d = device_get_softc(dev);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
d->dev = dev;
d->lock = snd_mtxcreate(device_get_nameunit(dev), "sound cdev");
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
cv_init(&d->cv, device_get_nameunit(dev));
PCM_ACQUIRE_QUICK(d);
dsp_cdevinfo_init(d);
#if 0
/*
* d->flags should be cleared by the allocator of the softc.
* We cannot clear this field here because several devices set
* this flag before calling pcm_register().
*/
d->flags = 0;
#endif
d->devinfo = devinfo;
d->devcount = 0;
d->reccount = 0;
d->playcount = 0;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
d->pvchancount = 0;
d->rvchancount = 0;
d->pvchanrate = 0;
d->pvchanformat = 0;
d->rvchanrate = 0;
d->rvchanformat = 0;
d->inprog = 0;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
/*
* Create clone manager, disabled by default. Cloning will be
* enabled during final stage of driver iniialization through
* pcm_setstatus().
*/
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
d->clones = snd_clone_create(SND_U_MASK | SND_D_MASK, PCMMAXCLONE,
SND_CLONE_DEADLINE_DEFAULT, SND_CLONE_WAITOK |
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
SND_CLONE_GC_ENABLE | SND_CLONE_GC_UNREF |
SND_CLONE_GC_LASTREF | SND_CLONE_GC_EXPIRED);
if (bootverbose != 0 || snd_verbose > 3) {
device_printf(dev,
"clone manager: deadline=%dms flags=0x%08x\n",
snd_clone_getdeadline(d->clones),
snd_clone_getflags(d->clones));
}
CHN_INIT(d, channels.pcm);
CHN_INIT(d, channels.pcm.busy);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
/* XXX This is incorrect, but lets play along for now. */
if ((numplay == 0 || numrec == 0) && numplay != numrec)
d->flags |= SD_F_SIMPLEX;
d->fakechan = fkchan_setup(dev);
chn_init(d->fakechan, NULL, 0, 0);
2001-02-27 07:01:49 +00:00
#ifdef SND_DYNSYSCTL
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
sysctl_ctx_init(&d->play_sysctl_ctx);
d->play_sysctl_tree = SYSCTL_ADD_NODE(&d->play_sysctl_ctx,
SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO, "play",
CTLFLAG_RD, 0, "playback channels node");
sysctl_ctx_init(&d->rec_sysctl_ctx);
d->rec_sysctl_tree = SYSCTL_ADD_NODE(&d->rec_sysctl_ctx,
SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO, "rec",
CTLFLAG_RD, 0, "record channels node");
/* XXX: an user should be able to set this with a control tool, the
sysadmin then needs min+max sysctls for this */
Welcome to Once-a-year Sound Mega-Commit. Enjoy numerous updates and fixes in every sense. General ------- - Multichannel safe, endian safe, format safe * Large part of critical pcm filters such as vchan.c, feeder_rate.c, feeder_volume.c, feeder_fmt.c and feeder.c has been rewritten so that using them does not cause the pcm data to be converted to 16bit little endian. * Macrosses for accessing pcm data safely are defined within sound.h in the form of PCM_READ_* / PCM_WRITE_* * Currently, most of them are probably limited for mono/stereo handling, but the future addition of true multichannel will be much easier. - Low latency operation * Well, this require lot more works to do not just within sound driver, but we're heading towards right direction. Buffer/block sizing within channel.c is rewritten to calculate precise allocation for various combination of sample/data/rate size. As a result, applying correct SNDCTL_DSP_POLICY value will achive expected latency behaviour simmilar to what commercial 4front driver do. * Signal handling fix. ctrl+c of "cat /dev/zero > /dev/dsp" does not result long delay. * Eliminate sound truncation if the sound data is too small. DIY: 1) Download / extract http://people.freebsd.org/~ariff/lowlatency/shortfiles.tar.gz 2) Do a comparison between "cat state*.au > /dev/dsp" and "for x in state*.au ; do cat $x > /dev/dsp ; done" - there should be no "perceivable" differences. Double close for PR kern/31445. CAVEAT: Low latency come with (unbearable) price especially for poorly written applications. Applications that trying to act smarter by requesting (wrong) blocksize/blockcount will suffer the most. Fixup samples/patches can be found at: http://people.freebsd.org/~ariff/ports/ - Switch minimum/maximum sampling rate limit to "1" and "2016000" (48k * 42) due to closer compatibility with 4front driver. Discussed with: marcus@ (long time ago?) - All driver specific sysctls in the form of "hw.snd.pcm%d.*" have been moved to their own dev sysctl nodes, notably: hw.snd.pcm%d.vchans -> dev.pcm.%d.vchans Bump __FreeBSD_version. Driver specific --------------- - Ditto for sysctls. - snd_atiixp, snd_es137x, snd_via8233, snd_hda * Numerous cleanups and fixes. * _EXPERIMENTAL_ polling mode support using simple callout_* mechanisme. This was intended for pure debugging and latency measurement, but proven good enough in few unexpected and rare cases (such as problematic shared IRQ with GIANT devices - USB). Polling can be enabled/disabled through dev.pcm.0.polling. Disabled by default. - snd_ich * Fix possible overflow during speed calibration. Delay final initialization (pcm_setstatus) after calibration finished. PR: kern/100169 Tested by: Kevin Overman <oberman@es.net> * Inverted EAPD for few Nec VersaPro. PR: kern/104715 Submitted by: KAWATA Masahiko <kawata@mta.biglobe.ne.jp> Thanks to various people, notably Joel Dahl, Yuriy Tsibizov, Kevin Oberman, those at #freebsd-azalia @ freenode and others for testing. Joel Dahl will do the manpage update.
2006-11-26 12:24:06 +00:00
SYSCTL_ADD_INT(device_get_sysctl_ctx(dev),
SYSCTL_CHILDREN(device_get_sysctl_tree(dev)),
OID_AUTO, "buffersize", CTLFLAG_RD, &d->bufsz, 0, "allocated buffer size");
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
#ifdef SND_DEBUG
SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev),
SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO,
"clone_flags", CTLTYPE_UINT | CTLFLAG_RW, d, sizeof(d),
sysctl_dev_pcm_clone_flags, "IU",
"clone flags");
SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev),
SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO,
"clone_deadline", CTLTYPE_INT | CTLFLAG_RW, d, sizeof(d),
sysctl_dev_pcm_clone_deadline, "I",
"clone expiration deadline (ms)");
SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev),
SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO,
"clone_gc", CTLTYPE_INT | CTLFLAG_RW, d, sizeof(d),
sysctl_dev_pcm_clone_gc, "I",
"clone garbage collector");
#endif
2001-02-27 07:01:49 +00:00
#endif
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (numplay > 0 || numrec > 0) {
d->flags |= SD_F_AUTOVCHAN;
vchan_initsys(dev);
- channel.h * New definition CHN_F_HAS_VCHAN. - channel.c * Use CHN_F_HAS_VCHAN to mark channel with vchan capability instead of relying on SLIST_EMPTY(&channel->children) == true for better clarification and future possible usages of children (like 'slave' channel). * Various fixes, including blocksize / format bps allignment, better 24bit seeking (mplayer, others). * Improve format chain building, it's now possible to record something to a format non-native to the soundcard through various feeder format converters or to higher sampling rate. This also gains another feature, like doing vchan mixing on non s16le soundcard such as sb8. - sound.c * Increase robustness within various function that handle vchan creation / termination (these function need a total rewrite, but that would cause other major rewrite within various places too!). As far as its robustness can be guaranteed, leave it as is. * Optimize channel ordering, prefer *real* hardware playback channels over virtual channels. cat /dev/sndstat should look better. * Increase sndstat verbosity to include bufsoft/bufhard allocation. - vchan.c * Fix LOR 119. - http://sources.zabbadoz.net/freebsd/lor.html#119 * Reorder / increase robustness of vchan_create() / destroy(). Enforce destroy_dev() during destroy operation, fix possible panic / dangling character device. - http://lists.freebsd.org/pipermail/freebsd-current/2005-May/050308.html * Tolerate a little bit more during mixing process, this should help non s16le soundcards. Note: Recoring in a non-native rate/format may result in overruns. A friendly application is wavrec from audio/wavplay. The problem is under investigation. Submitted by: Ariff Abdullah <skywizard@MyBSD.org.my>
2005-09-10 18:10:31 +00:00
}
sndstat_register(dev, d->status, sndstat_prepare_pcm);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return 0;
}
int
pcm_unregister(device_t dev)
{
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
struct snddev_info *d;
struct pcm_channel *ch;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
struct thread *td;
int i;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
td = curthread;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
d = device_get_softc(dev);
if (!PCM_ALIVE(d)) {
device_printf(dev, "unregister: device not configured\n");
return (0);
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (sndstat_acquire(td) != 0) {
device_printf(dev, "unregister: sndstat busy\n");
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (EBUSY);
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
pcm_lock(d);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_WAIT(d);
if (d->inprog != 0) {
device_printf(dev, "unregister: operation in progress\n");
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
pcm_unlock(d);
sndstat_release(td);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (EBUSY);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_ACQUIRE(d);
pcm_unlock(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
CHN_FOREACH(ch, d, channels.pcm) {
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
CHN_LOCK(ch);
if (ch->refcount > 0) {
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
device_printf(dev,
"unregister: channel %s busy (pid %d)\n",
ch->name, ch->pid);
CHN_UNLOCK(ch);
PCM_RELEASE_QUICK(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
sndstat_release(td);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (EBUSY);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
CHN_UNLOCK(ch);
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (d->clones != NULL) {
if (snd_clone_busy(d->clones) != 0) {
device_printf(dev, "unregister: clone busy\n");
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_RELEASE_QUICK(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
sndstat_release(td);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (EBUSY);
} else {
pcm_lock(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
(void)snd_clone_disable(d->clones);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
pcm_unlock(d);
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
}
if (mixer_uninit(dev) == EBUSY) {
device_printf(dev, "unregister: mixer busy\n");
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
pcm_lock(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (d->clones != NULL)
(void)snd_clone_enable(d->clones);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_RELEASE(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
pcm_unlock(d);
sndstat_release(td);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (EBUSY);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
pcm_lock(d);
d->flags |= SD_F_DYING;
d->flags &= ~SD_F_REGISTERED;
pcm_unlock(d);
/*
* No lock being held, so this thing can be flushed without
* stucking into devdrn oblivion.
*/
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
if (d->clones != NULL) {
snd_clone_destroy(d->clones);
d->clones = NULL;
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
#ifdef SND_DYNSYSCTL
if (d->play_sysctl_tree != NULL) {
sysctl_ctx_free(&d->play_sysctl_ctx);
d->play_sysctl_tree = NULL;
}
if (d->rec_sysctl_tree != NULL) {
sysctl_ctx_free(&d->rec_sysctl_ctx);
d->rec_sysctl_tree = NULL;
}
#endif
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
while (!CHN_EMPTY(d, channels.pcm))
pcm_killchan(dev);
chn_kill(d->fakechan);
fkchan_kill(d->fakechan);
2001-02-27 07:01:49 +00:00
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
dsp_cdevinfo_flush(d);
pcm_lock(d);
PCM_RELEASE(d);
cv_destroy(&d->cv);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
pcm_unlock(d);
snd_mtxfree(d->lock);
sndstat_unregister(dev);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
sndstat_release(td);
if (snd_unit == device_get_unit(dev)) {
/*
* Reassign default unit to the next available dev, but
* first, reset snd_unit to something ridiculous.
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
*/
snd_unit = -1;
for (i = 0; pcm_devclass != NULL &&
i < devclass_get_maxunit(pcm_devclass); i++) {
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (device_get_unit(dev) == i)
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
continue;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
d = devclass_get_softc(pcm_devclass, i);
if (PCM_REGISTERED(d)) {
snd_unit = i;
break;
}
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
}
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
return (0);
}
/************************************************************************/
static int
sndstat_prepare_pcm(struct sbuf *s, device_t dev, int verbose)
{
struct snddev_info *d;
struct pcm_channel *c;
struct pcm_feeder *f;
if (verbose < 1)
return 0;
d = device_get_softc(dev);
if (!d)
return ENXIO;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_BUSYASSERT(d);
if (CHN_EMPTY(d, channels.pcm)) {
sbuf_printf(s, " (mixer only)");
return 0;
}
sbuf_printf(s, " (%dp:%dv/%dr:%dv channels%s%s)",
d->playcount, d->pvchancount,
d->reccount, d->rvchancount,
(d->flags & SD_F_SIMPLEX)? "" : " duplex",
#ifdef USING_DEVFS
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
(device_get_unit(dev) == snd_unit)? " default" : ""
#else
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
""
#endif
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
);
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (verbose <= 1)
return 0;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
CHN_FOREACH(c, d, channels.pcm) {
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
KASSERT(c->bufhard != NULL && c->bufsoft != NULL,
("hosed pcm channel setup"));
sbuf_printf(s, "\n\t");
/* it would be better to indent child channels */
sbuf_printf(s, "%s[%s]: ", c->parentchannel? c->parentchannel->name : "", c->name);
sbuf_printf(s, "spd %d", c->speed);
if (c->speed != sndbuf_getspd(c->bufhard))
sbuf_printf(s, "/%d", sndbuf_getspd(c->bufhard));
sbuf_printf(s, ", fmt 0x%08x", c->format);
if (c->format != sndbuf_getfmt(c->bufhard))
sbuf_printf(s, "/0x%08x", sndbuf_getfmt(c->bufhard));
sbuf_printf(s, ", flags 0x%08x, 0x%08x", c->flags, c->feederflags);
if (c->pid != -1)
sbuf_printf(s, ", pid %d", c->pid);
sbuf_printf(s, "\n\t");
sbuf_printf(s, "interrupts %d, ", c->interrupts);
if (c->direction == PCMDIR_REC)
sbuf_printf(s, "overruns %d, feed %u, hfree %d, sfree %d [b:%d/%d/%d|bs:%d/%d/%d]",
c->xruns, c->feedcount, sndbuf_getfree(c->bufhard), sndbuf_getfree(c->bufsoft),
sndbuf_getsize(c->bufhard), sndbuf_getblksz(c->bufhard),
sndbuf_getblkcnt(c->bufhard),
sndbuf_getsize(c->bufsoft), sndbuf_getblksz(c->bufsoft),
sndbuf_getblkcnt(c->bufsoft));
else
sbuf_printf(s, "underruns %d, feed %u, ready %d [b:%d/%d/%d|bs:%d/%d/%d]",
c->xruns, c->feedcount, sndbuf_getready(c->bufsoft),
sndbuf_getsize(c->bufhard), sndbuf_getblksz(c->bufhard),
sndbuf_getblkcnt(c->bufhard),
sndbuf_getsize(c->bufsoft), sndbuf_getblksz(c->bufsoft),
sndbuf_getblkcnt(c->bufsoft));
sbuf_printf(s, "\n\t");
sbuf_printf(s, "{%s}", (c->direction == PCMDIR_REC)? "hardware" : "userland");
sbuf_printf(s, " -> ");
f = c->feeder;
while (f->source != NULL)
f = f->source;
while (f != NULL) {
sbuf_printf(s, "%s", f->class->name);
if (f->desc->type == FEEDER_FMT)
sbuf_printf(s, "(0x%08x -> 0x%08x)", f->desc->in, f->desc->out);
if (f->desc->type == FEEDER_RATE)
sbuf_printf(s, "(%d -> %d)", FEEDER_GET(f, FEEDRATE_SRC), FEEDER_GET(f, FEEDRATE_DST));
if (f->desc->type == FEEDER_ROOT || f->desc->type == FEEDER_MIXER ||
f->desc->type == FEEDER_VOLUME)
sbuf_printf(s, "(0x%08x)", f->desc->out);
sbuf_printf(s, " -> ");
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
f = f->parent;
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
sbuf_printf(s, "{%s}", (c->direction == PCMDIR_REC)? "userland" : "hardware");
}
return 0;
}
/************************************************************************/
#ifdef SND_DYNSYSCTL
int
sysctl_hw_snd_vchans(SYSCTL_HANDLER_ARGS)
{
struct snddev_info *d;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
int direction, vchancount;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
int err, cnt;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
d = devclass_get_softc(pcm_devclass, VCHAN_SYSCTL_UNIT(oidp->oid_arg1));
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (!PCM_REGISTERED(d) || !(d->flags & SD_F_AUTOVCHAN))
return (EINVAL);
pcm_lock(d);
PCM_WAIT(d);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
switch (VCHAN_SYSCTL_DIR(oidp->oid_arg1)) {
case VCHAN_PLAY:
direction = PCMDIR_PLAY;
vchancount = d->pvchancount;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
cnt = d->playcount;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
break;
case VCHAN_REC:
direction = PCMDIR_REC;
vchancount = d->rvchancount;
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
cnt = d->reccount;
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
break;
default:
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
pcm_unlock(d);
return (EINVAL);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
break;
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
if (cnt < 1) {
pcm_unlock(d);
return (ENODEV);
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_ACQUIRE(d);
pcm_unlock(d);
cnt = vchancount;
err = sysctl_handle_int(oidp, &cnt, 0, req);
if (err == 0 && req->newptr != NULL && vchancount != cnt) {
if (cnt < 0)
cnt = 0;
if (cnt > SND_MAXVCHANS)
cnt = SND_MAXVCHANS;
err = pcm_setvchans(d, direction, cnt, -1);
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
PCM_RELEASE_QUICK(d);
return err;
}
#endif
/************************************************************************/
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
/**
* @brief Handle OSSv4 SNDCTL_SYSINFO ioctl.
*
* @param si Pointer to oss_sysinfo struct where information about the
* sound subsystem will be written/copied.
*
* This routine returns information about the sound system, such as the
* current OSS version, number of audio, MIDI, and mixer drivers, etc.
* Also includes a bitmask showing which of the above types of devices
* are open (busy).
*
* @note
* Calling threads must not hold any snddev_info or pcm_channel locks.
*
* @author Ryan Beasley <ryanb@FreeBSD.org>
*/
void
sound_oss_sysinfo(oss_sysinfo *si)
{
static char si_product[] = "FreeBSD native OSS ABI";
static char si_version[] = __XSTRING(__FreeBSD_version);
static int intnbits = sizeof(int) * 8; /* Better suited as macro?
Must pester a C guru. */
struct snddev_info *d;
struct pcm_channel *c;
int i, j, ncards;
ncards = 0;
strlcpy(si->product, si_product, sizeof(si->product));
strlcpy(si->version, si_version, sizeof(si->version));
si->versionnum = SOUND_VERSION;
/*
* Iterate over PCM devices and their channels, gathering up data
* for the numaudios, ncards, and openedaudio fields.
*/
si->numaudios = 0;
bzero((void *)&si->openedaudio, sizeof(si->openedaudio));
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
j = 0;
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
for (i = 0; pcm_devclass != NULL &&
i < devclass_get_maxunit(pcm_devclass); i++) {
d = devclass_get_softc(pcm_devclass, i);
if (!PCM_REGISTERED(d))
continue;
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
/* XXX Need Giant magic entry ??? */
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
/* See note in function's docblock */
mtx_assert(d->lock, MA_NOTOWNED);
pcm_lock(d);
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
si->numaudios += d->devcount;
++ncards;
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
CHN_FOREACH(c, d, channels.pcm) {
mtx_assert(c->lock, MA_NOTOWNED);
CHN_LOCK(c);
if (c->flags & CHN_F_BUSY)
si->openedaudio[j / intnbits] |=
(1 << (j % intnbits));
CHN_UNLOCK(c);
j++;
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
}
Last (again ?!?) major commit for RELENG_7, featuring total Giant eradication in/from userland path, countless locking fixes, etc. - General sleep call through msleep(9) has been converted to condvar(9) with better consistencies. - Heavily guard every possible "slow path" entries (open(), close(), few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt started), they are free to fly on their own. - Rearrange locking sequences, resulting better concurrency and serialization. Large part doesn't even need locking at all, and will be removed in future. Less clutter, except in few places due to lock ordering. - Anonymous mixer object creation/deletion to simplify mixer handling beyond typical mixer ioctls. Submitted by: chibis (with modifications) - Add few mix_[get|set|..] functions to avoid calling mixer_ioctl() directly using cryptic arguments. - Locking fixes to avoid possible deadlock with (still under Giant) USB. - Better simplex/duplex device handling. - Recover mmap() functionality for recording, which has been lost since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still doesn't work (due to VM/page design), but people still can mmap both by opening each direction separately. mmaped playback is guarantee to work either way. - New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page mapping, due to recent changes in linux compatibility layer which require it. All linux applications that using sound + mmap() (mostly games) require this to be enabled. Disabled by default. - Other goodies.. too many, that will increase releng7 shareholder value and make users of releng6 (and below) cry ;) * This commit should be atomic. If anything goes wrong (not counting problem originated from elsewhere), I will not hesitate to revert everything back within 12 hours. This substantial changes itself not a rocket science and the process has begun for almost 2 years, and lots of incremental changes are already in place during that period of time. * Some issues does occur in snd_emu10kx (note the 'x') due to various internal locking issues and it is currently being worked on by chibis. Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira, many innocent souls...
2007-06-16 03:37:28 +00:00
pcm_unlock(d);
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
}
si->numsynths = 0; /* OSSv4 docs: this field is obsolete */
/**
* @todo Collect num{midis,timers}.
*
* Need access to sound/midi/midi.c::midistat_lock in order
* to safely touch midi_devices and get a head count of, well,
* MIDI devices. midistat_lock is a global static (i.e., local to
* midi.c), but midi_devices is a regular global; should the mutex
* be publicized, or is there another way to get this information?
*
* NB: MIDI/sequencer stuff is currently on hold.
*/
si->nummidis = 0;
si->numtimers = 0;
si->nummixers = mixer_count;
si->numcards = ncards;
/* OSSv4 docs: Intended only for test apps; API doesn't
really have much of a concept of cards. Shouldn't be
used by applications. */
/**
* @todo Fill in "busy devices" fields.
*
* si->openedmidi = " MIDI devices
*/
bzero((void *)&si->openedmidi, sizeof(si->openedmidi));
/*
* Si->filler is a reserved array, but according to docs each
* element should be set to -1.
*/
for (i = 0; i < sizeof(si->filler)/sizeof(si->filler[0]); i++)
si->filler[i] = -1;
}
/************************************************************************/
static int
sound_modevent(module_t mod, int type, void *data)
{
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
int ret;
#if 0
return (midi_modevent(mod, type, data));
#else
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
ret = 0;
switch(type) {
case MOD_LOAD:
pcmsg_unrhdr = new_unrhdr(1, INT_MAX, NULL);
break;
case MOD_UNLOAD:
case MOD_SHUTDOWN:
Last major commit and updates for RELENG_7: - Rework the entire pcm_channel structure: * Remove rarely used link placeholder, instead, make each pcm_channel as head/link of each own/each other. Unlock - Lock sequence due to sleep malloc has been reduced. * Implement "busy" queue which will contain list of busy/active channels. This greatly reduce locking contention for example while servicing interrupt for hardware with many channels or when virtual channels reach its 256 peak channels. - So I heard you like v chan ... O RLY? Welcome to Virtual **Record** Channels (vrec, rec vchans, vchans for recording, Rec-Chan, you decide), the ultimate solutions for your nagging O_RDWR full-duplex wannabe (note: flash plugins) monopolizing single record channel causing EBUSY. Vrec works exactly like Vchans (or, should I rename it to "Vplay" :) , except that it operates on the opposite direction (recording). Up to 256 vrecs (like vchans) are possible. Notes: * Relocate dev.pcm.%d.{vchans,vchanformat,vchanrate} to each of its respective node/direction: dev.pcm.%d.play.* for "play" (cdev = dsp%d.vp%d) dev.pcm.%d.rec.* for "record" (cdev = dsp%d.vr%d) * Don't expect that it will magically give you ability to split "recording source" (eg: 1 channel for cdrom, 1 channel for mic, etc). Just admit that you only have a *single* recording source / channel. Please bug your hardware vendor instead :) - Bump maxautovchans from 4 to 16. For a full-fledged multimedia desktop/workstation with too many soundservers installed (esound, artsd, jackd, pulse/polypaudio, ding-dong pling plong mudkip fuh fuh, etc), 4 seems inadequate. There will be no memory penalty here, since virtual channels are allocate only by demand. - Nuke/Rework the entire statically created cdev entries. Everything is clonable through snd own clone manager which designed to withstand many kind of abusive devfs droids such as: * while : ; do /bin/test -e /dev/dsp ; done * jot 16777216 0 | while read x ; do ls /dev/dsp0.$x ; done * hundreds (could be thousands) concurrent threads/process opening "/dev/dsp" (previously, this might result EBUSY even with just 3 contesting threads/procs). o Reusable clone objects (instead of creating new one like there's no tomorrow) after certain expiration deadline. The clone allocator will decide whether to reuse, share, or creating new clone. o Automatic garbage collector. - Dynamic unit magic allocator. Maximum attached soundcards can be tuned using tunable "hw.snd.maxunit" (Default to 512). Minimum is 16, and maximum is 2048. - ..other fixes, mostly related to concurrency issues. joel@ will do the manpage updates on sound(4). Have fun.
2007-05-31 18:43:33 +00:00
ret = sndstat_acquire(curthread);
if (ret != 0)
break;
MFp4 the sound Google Summer of Code project: The goal was to sync with the OSSv4 API 4Front Technologies uses in their proprietary OSS driver. This was successful as far as possible. The part of the API which is stable is implemented, for the rest there are some stubs already. New system ioctls: - SNDCTL_SYSINFO - obtain audio system info (version, # of audio/midi/ mixer devices, etc.) - SNDCTL_AUDIOINFO - fetch details about a specific audio device - SNDCTL_MIXERINFO - fetch details about a specific mixer device New audio ioctls: - Sync groups (SNDCTL_DSP_SYNCGROUP/SNDCTL_DSP_SYNCSTART) which allow triggered playback/recording on multiple devices (even across processes simultaneously). - Peak meters (SNDCTL_DSP_GETIPEAKS/SNDCTL_DSP_GETOPEAKS) - can query audio drivers for peak levels (needs driver support, disabled for now). - Per channel playback/recording levels - SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL. Note that these are still in name only, just wrapping around the AC97-style mixer at the moment. The next step is to push them down to the drivers. Audio ioctls still under development by 4Front (for which stubs may exist in this commit): - SNDCTL_GETNAME, SNDCTL_{GET,SET}{SONG,LABEL} - SNDCTL_DSP_{GET,SET}_CHNORDER - SNDCTL_MIX_ENUMINFO, SNDCTL_MIX_EXTINFO - (might be documented enough in the OSS releases to work on this. These ioctls cover the cool "twiddle any knob on your card" features.) Missing: - SNDCTL_DSP_COOKEDMODE -- this ioctl is used to give applications direct access to a card's buffers, bypassing the feeder architecture. It's a toughy -- "someone" needs to decide : (a) if this is desireable, and (b) if it's reasonably feasible. Updates for driver writers: So far, only two routines to the channel class (in channel_if.m) are added. One is for fetching a list of discrete supported playback/recording rates of a channel, and the other is for fetching peak level info (useful for drawing peak meters). Interested parties may want to help pushing down SNDCTL_DSP_{GET,SET}{PLAY,REC}VOL into the drivers. To use the new stuff you need to rebuild the sound drivers or your kernel (depending on if you use modules or not) and to install soundcard.h (a buildworld/installworld handles this). Sponsored by: Google SoC 2006 Submitted by: ryanb Many thanks to: 4Front Technologies for their cooperation, explanations and the nice license of their soundcard.h.
2006-09-23 20:45:47 +00:00
if (pcmsg_unrhdr != NULL) {
delete_unrhdr(pcmsg_unrhdr);
pcmsg_unrhdr = NULL;
}
break;
default:
ret = EOPNOTSUPP;
}
return ret;
#endif
}
DEV_MODULE(sound, sound_modevent, NULL);
MODULE_VERSION(sound, SOUND_MODVER);