freebsd-skq/sys/dev/sound/pci/atiixp.c

1427 lines
34 KiB
C
Raw Normal View History

/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2005 Ariff Abdullah <ariff@FreeBSD.org>
* 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, WHETHERIN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THEPOSSIBILITY OF
* SUCH DAMAGE.
*/
/*
* FreeBSD pcm driver for ATI IXP 150/200/250/300 AC97 controllers
*
* Features
* * 16bit playback / recording
* * 32bit native playback - yay!
* * 32bit native recording (seems broken on few hardwares)
*
* Issues / TODO:
* * SPDIF
* * Support for more than 2 channels.
* * VRA ? VRM ? DRA ?
* * 32bit native recording seems broken on few hardwares, most
* probably because of incomplete VRA/DRA cleanup.
*
*
* Thanks goes to:
*
* Shaharil @ SCAN Associates whom relentlessly providing me the
* mind blowing Acer Ferrari 4002 WLMi with this ATI IXP hardware.
*
* Reinoud Zandijk <reinoud@NetBSD.org> (auixp), which this driver is
* largely based upon although large part of it has been reworked. His
* driver is the primary reference and pretty much well documented.
*
* Takashi Iwai (ALSA snd-atiixp), for register definitions and some
* random ninja hackery.
*/
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
#ifdef HAVE_KERNEL_OPTION_HEADERS
#include "opt_snd.h"
#endif
#include <dev/sound/pcm/sound.h>
#include <dev/sound/pcm/ac97.h>
#include <dev/pci/pcireg.h>
#include <dev/pci/pcivar.h>
#include <sys/sysctl.h>
#include <sys/endian.h>
#include <dev/sound/pci/atiixp.h>
SND_DECLARE_FILE("$FreeBSD$");
#define ATI_IXP_DMA_RETRY_MAX 100
#define ATI_IXP_BUFSZ_MIN 4096
#define ATI_IXP_BUFSZ_MAX 65536
#define ATI_IXP_BUFSZ_DEFAULT 16384
#define ATI_IXP_BLK_MIN 32
#define ATI_IXP_BLK_ALIGN (~(ATI_IXP_BLK_MIN - 1))
#define ATI_IXP_CHN_RUNNING 0x00000001
#define ATI_IXP_CHN_SUSPEND 0x00000002
struct atiixp_dma_op {
volatile uint32_t addr;
volatile uint16_t status;
volatile uint16_t size;
volatile uint32_t next;
};
struct atiixp_info;
struct atiixp_chinfo {
struct snd_dbuf *buffer;
struct pcm_channel *channel;
struct atiixp_info *parent;
struct atiixp_dma_op *sgd_table;
bus_addr_t sgd_addr;
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
uint32_t enable_bit, flush_bit, linkptr_bit, dt_cur_bit;
uint32_t blksz, blkcnt;
uint32_t ptr, prevptr;
2005-12-25 00:43:03 +00:00
uint32_t fmt;
uint32_t flags;
int caps_32bit, dir;
};
struct atiixp_info {
device_t dev;
bus_space_tag_t st;
bus_space_handle_t sh;
bus_dma_tag_t parent_dmat;
bus_dma_tag_t sgd_dmat;
bus_dmamap_t sgd_dmamap;
bus_addr_t sgd_addr;
struct resource *reg, *irq;
int regtype, regid, irqid;
void *ih;
struct ac97_info *codec;
struct atiixp_chinfo pch;
struct atiixp_chinfo rch;
struct atiixp_dma_op *sgd_table;
struct intr_config_hook delayed_attach;
uint32_t bufsz;
uint32_t codec_not_ready_bits, codec_idx, codec_found;
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
uint32_t blkcnt;
int registered_channels;
struct mtx *lock;
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
struct callout poll_timer;
int poll_ticks, polling;
};
#define atiixp_rd(_sc, _reg) \
bus_space_read_4((_sc)->st, (_sc)->sh, _reg)
#define atiixp_wr(_sc, _reg, _val) \
bus_space_write_4((_sc)->st, (_sc)->sh, _reg, _val)
#define atiixp_lock(_sc) snd_mtxlock((_sc)->lock)
#define atiixp_unlock(_sc) snd_mtxunlock((_sc)->lock)
#define atiixp_assert(_sc) snd_mtxassert((_sc)->lock)
static uint32_t atiixp_fmt_32bit[] = {
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
SND_FORMAT(AFMT_S16_LE, 2, 0),
SND_FORMAT(AFMT_S32_LE, 2, 0),
0
};
static uint32_t atiixp_fmt[] = {
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
SND_FORMAT(AFMT_S16_LE, 2, 0),
0
};
static struct pcmchan_caps atiixp_caps_32bit = {
ATI_IXP_BASE_RATE,
ATI_IXP_BASE_RATE,
atiixp_fmt_32bit, 0
};
static struct pcmchan_caps atiixp_caps = {
ATI_IXP_BASE_RATE,
ATI_IXP_BASE_RATE,
atiixp_fmt, 0
};
static const struct {
uint16_t vendor;
uint16_t devid;
char *desc;
} atiixp_hw[] = {
{ ATI_VENDOR_ID, ATI_IXP_200_ID, "ATI IXP 200" },
{ ATI_VENDOR_ID, ATI_IXP_300_ID, "ATI IXP 300" },
{ ATI_VENDOR_ID, ATI_IXP_400_ID, "ATI IXP 400" },
{ ATI_VENDOR_ID, ATI_IXP_SB600_ID, "ATI IXP SB600" },
};
static void atiixp_enable_interrupts(struct atiixp_info *);
static void atiixp_disable_interrupts(struct atiixp_info *);
static void atiixp_reset_aclink(struct atiixp_info *);
static void atiixp_flush_dma(struct atiixp_chinfo *);
static void atiixp_enable_dma(struct atiixp_chinfo *);
static void atiixp_disable_dma(struct atiixp_chinfo *);
static int atiixp_waitready_codec(struct atiixp_info *);
static int atiixp_rdcd(kobj_t, void *, int);
static int atiixp_wrcd(kobj_t, void *, int, uint32_t);
static void *atiixp_chan_init(kobj_t, void *, struct snd_dbuf *,
struct pcm_channel *, int);
static int atiixp_chan_setformat(kobj_t, void *, uint32_t);
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
static uint32_t atiixp_chan_setspeed(kobj_t, void *, uint32_t);
static int atiixp_chan_setfragments(kobj_t, void *, uint32_t, uint32_t);
static uint32_t atiixp_chan_setblocksize(kobj_t, void *, uint32_t);
static void atiixp_buildsgdt(struct atiixp_chinfo *);
static int atiixp_chan_trigger(kobj_t, void *, int);
static __inline uint32_t atiixp_dmapos(struct atiixp_chinfo *);
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
static uint32_t atiixp_chan_getptr(kobj_t, void *);
static struct pcmchan_caps *atiixp_chan_getcaps(kobj_t, void *);
static void atiixp_intr(void *);
static void atiixp_dma_cb(void *, bus_dma_segment_t *, int, int);
static void atiixp_chip_pre_init(struct atiixp_info *);
static void atiixp_chip_post_init(void *);
static void atiixp_release_resource(struct atiixp_info *);
static int atiixp_pci_probe(device_t);
static int atiixp_pci_attach(device_t);
static int atiixp_pci_detach(device_t);
2005-12-25 00:43:03 +00:00
static int atiixp_pci_suspend(device_t);
static int atiixp_pci_resume(device_t);
/*
* ATI IXP helper functions
*/
static void
atiixp_enable_interrupts(struct atiixp_info *sc)
{
uint32_t value;
/* clear all pending */
atiixp_wr(sc, ATI_REG_ISR, 0xffffffff);
/* enable all relevant interrupt sources we can handle */
value = atiixp_rd(sc, ATI_REG_IER);
value |= ATI_REG_IER_IO_STATUS_EN;
/*
* Disable / ignore internal xrun/spdf interrupt flags
* since it doesn't interest us (for now).
*/
#if 1
value &= ~(ATI_REG_IER_IN_XRUN_EN | ATI_REG_IER_OUT_XRUN_EN |
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
ATI_REG_IER_SPDF_XRUN_EN | ATI_REG_IER_SPDF_STATUS_EN);
#else
value |= ATI_REG_IER_IN_XRUN_EN;
value |= ATI_REG_IER_OUT_XRUN_EN;
value |= ATI_REG_IER_SPDF_XRUN_EN;
value |= ATI_REG_IER_SPDF_STATUS_EN;
#endif
atiixp_wr(sc, ATI_REG_IER, value);
}
static void
atiixp_disable_interrupts(struct atiixp_info *sc)
{
/* disable all interrupt sources */
atiixp_wr(sc, ATI_REG_IER, 0);
/* clear all pending */
atiixp_wr(sc, ATI_REG_ISR, 0xffffffff);
}
static void
atiixp_reset_aclink(struct atiixp_info *sc)
{
uint32_t value, timeout;
/* if power is down, power it up */
value = atiixp_rd(sc, ATI_REG_CMD);
if (value & ATI_REG_CMD_POWERDOWN) {
/* explicitly enable power */
value &= ~ATI_REG_CMD_POWERDOWN;
atiixp_wr(sc, ATI_REG_CMD, value);
/* have to wait at least 10 usec for it to initialise */
DELAY(20);
}
/* perform a soft reset */
value = atiixp_rd(sc, ATI_REG_CMD);
value |= ATI_REG_CMD_AC_SOFT_RESET;
atiixp_wr(sc, ATI_REG_CMD, value);
/* need to read the CMD reg and wait aprox. 10 usec to init */
value = atiixp_rd(sc, ATI_REG_CMD);
DELAY(20);
/* clear soft reset flag again */
value = atiixp_rd(sc, ATI_REG_CMD);
value &= ~ATI_REG_CMD_AC_SOFT_RESET;
atiixp_wr(sc, ATI_REG_CMD, value);
/* check if the ac-link is working; reset device otherwise */
timeout = 10;
value = atiixp_rd(sc, ATI_REG_CMD);
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
while (!(value & ATI_REG_CMD_ACLINK_ACTIVE) && --timeout) {
2005-12-25 00:43:03 +00:00
#if 0
device_printf(sc->dev, "not up; resetting aclink hardware\n");
2005-12-25 00:43:03 +00:00
#endif
/* dip aclink reset but keep the acsync */
value &= ~ATI_REG_CMD_AC_RESET;
value |= ATI_REG_CMD_AC_SYNC;
atiixp_wr(sc, ATI_REG_CMD, value);
/* need to read CMD again and wait again (clocking in issue?) */
value = atiixp_rd(sc, ATI_REG_CMD);
DELAY(20);
/* assert aclink reset again */
value = atiixp_rd(sc, ATI_REG_CMD);
value |= ATI_REG_CMD_AC_RESET;
atiixp_wr(sc, ATI_REG_CMD, value);
/* check if its active now */
value = atiixp_rd(sc, ATI_REG_CMD);
}
if (timeout == 0)
device_printf(sc->dev, "giving up aclink reset\n");
2005-12-25 00:43:03 +00:00
#if 0
if (timeout != 10)
device_printf(sc->dev, "aclink hardware reset successful\n");
2005-12-25 00:43:03 +00:00
#endif
/* assert reset and sync for safety */
value = atiixp_rd(sc, ATI_REG_CMD);
value |= ATI_REG_CMD_AC_SYNC | ATI_REG_CMD_AC_RESET;
atiixp_wr(sc, ATI_REG_CMD, value);
}
static void
atiixp_flush_dma(struct atiixp_chinfo *ch)
{
atiixp_wr(ch->parent, ATI_REG_FIFO_FLUSH, ch->flush_bit);
}
static void
atiixp_enable_dma(struct atiixp_chinfo *ch)
{
uint32_t value;
value = atiixp_rd(ch->parent, ATI_REG_CMD);
if (!(value & ch->enable_bit)) {
value |= ch->enable_bit;
atiixp_wr(ch->parent, ATI_REG_CMD, value);
}
}
static void
atiixp_disable_dma(struct atiixp_chinfo *ch)
{
uint32_t value;
value = atiixp_rd(ch->parent, ATI_REG_CMD);
if (value & ch->enable_bit) {
value &= ~ch->enable_bit;
atiixp_wr(ch->parent, ATI_REG_CMD, value);
}
}
/*
* AC97 interface
*/
static int
atiixp_waitready_codec(struct atiixp_info *sc)
{
int timeout = 500;
do {
if ((atiixp_rd(sc, ATI_REG_PHYS_OUT_ADDR) &
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
ATI_REG_PHYS_OUT_ADDR_EN) == 0)
return (0);
DELAY(1);
} while (--timeout);
return (-1);
}
static int
atiixp_rdcd(kobj_t obj, void *devinfo, int reg)
{
struct atiixp_info *sc = devinfo;
uint32_t data;
int timeout;
if (atiixp_waitready_codec(sc))
return (-1);
data = (reg << ATI_REG_PHYS_OUT_ADDR_SHIFT) |
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
ATI_REG_PHYS_OUT_ADDR_EN | ATI_REG_PHYS_OUT_RW | sc->codec_idx;
atiixp_wr(sc, ATI_REG_PHYS_OUT_ADDR, data);
if (atiixp_waitready_codec(sc))
return (-1);
timeout = 500;
do {
data = atiixp_rd(sc, ATI_REG_PHYS_IN_ADDR);
if (data & ATI_REG_PHYS_IN_READ_FLAG)
return (data >> ATI_REG_PHYS_IN_DATA_SHIFT);
DELAY(1);
} while (--timeout);
if (reg < 0x7c)
device_printf(sc->dev, "codec read timeout! (reg 0x%x)\n", reg);
return (-1);
}
static int
atiixp_wrcd(kobj_t obj, void *devinfo, int reg, uint32_t data)
{
struct atiixp_info *sc = devinfo;
if (atiixp_waitready_codec(sc))
return (-1);
data = (data << ATI_REG_PHYS_OUT_DATA_SHIFT) |
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
(((uint32_t)reg) << ATI_REG_PHYS_OUT_ADDR_SHIFT) |
ATI_REG_PHYS_OUT_ADDR_EN | sc->codec_idx;
atiixp_wr(sc, ATI_REG_PHYS_OUT_ADDR, data);
return (0);
}
static kobj_method_t atiixp_ac97_methods[] = {
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
KOBJMETHOD(ac97_read, atiixp_rdcd),
KOBJMETHOD(ac97_write, atiixp_wrcd),
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
KOBJMETHOD_END
};
AC97_DECLARE(atiixp_ac97);
/*
* Playback / Record channel interface
*/
static void *
atiixp_chan_init(kobj_t obj, void *devinfo, struct snd_dbuf *b,
struct pcm_channel *c, int dir)
{
struct atiixp_info *sc = devinfo;
struct atiixp_chinfo *ch;
int num;
atiixp_lock(sc);
if (dir == PCMDIR_PLAY) {
ch = &sc->pch;
ch->linkptr_bit = ATI_REG_OUT_DMA_LINKPTR;
ch->enable_bit = ATI_REG_CMD_OUT_DMA_EN | ATI_REG_CMD_SEND_EN;
ch->flush_bit = ATI_REG_FIFO_OUT_FLUSH;
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
ch->dt_cur_bit = ATI_REG_OUT_DMA_DT_CUR;
/* Native 32bit playback working properly */
ch->caps_32bit = 1;
} else {
ch = &sc->rch;
ch->linkptr_bit = ATI_REG_IN_DMA_LINKPTR;
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
ch->enable_bit = ATI_REG_CMD_IN_DMA_EN | ATI_REG_CMD_RECEIVE_EN;
ch->flush_bit = ATI_REG_FIFO_IN_FLUSH;
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
ch->dt_cur_bit = ATI_REG_IN_DMA_DT_CUR;
/* XXX Native 32bit recording appear to be broken */
ch->caps_32bit = 1;
}
ch->buffer = b;
ch->parent = sc;
ch->channel = c;
ch->dir = dir;
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
ch->blkcnt = sc->blkcnt;
ch->blksz = sc->bufsz / ch->blkcnt;
atiixp_unlock(sc);
if (sndbuf_alloc(ch->buffer, sc->parent_dmat, 0, sc->bufsz) == -1)
return (NULL);
atiixp_lock(sc);
num = sc->registered_channels++;
ch->sgd_table = &sc->sgd_table[num * ATI_IXP_DMA_CHSEGS_MAX];
ch->sgd_addr = sc->sgd_addr + (num * ATI_IXP_DMA_CHSEGS_MAX *
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
sizeof(struct atiixp_dma_op));
atiixp_disable_dma(ch);
atiixp_unlock(sc);
return (ch);
}
static int
atiixp_chan_setformat(kobj_t obj, void *data, uint32_t format)
{
struct atiixp_chinfo *ch = data;
struct atiixp_info *sc = ch->parent;
uint32_t value;
atiixp_lock(sc);
if (ch->dir == PCMDIR_REC) {
value = atiixp_rd(sc, ATI_REG_CMD);
value &= ~ATI_REG_CMD_INTERLEAVE_IN;
if ((format & AFMT_32BIT) == 0)
value |= ATI_REG_CMD_INTERLEAVE_IN;
atiixp_wr(sc, ATI_REG_CMD, value);
} else {
value = atiixp_rd(sc, ATI_REG_OUT_DMA_SLOT);
value &= ~ATI_REG_OUT_DMA_SLOT_MASK;
/* We do not have support for more than 2 channels, _yet_. */
value |= ATI_REG_OUT_DMA_SLOT_BIT(3) |
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
ATI_REG_OUT_DMA_SLOT_BIT(4);
value |= 0x04 << ATI_REG_OUT_DMA_THRESHOLD_SHIFT;
atiixp_wr(sc, ATI_REG_OUT_DMA_SLOT, value);
value = atiixp_rd(sc, ATI_REG_CMD);
value &= ~ATI_REG_CMD_INTERLEAVE_OUT;
if ((format & AFMT_32BIT) == 0)
value |= ATI_REG_CMD_INTERLEAVE_OUT;
atiixp_wr(sc, ATI_REG_CMD, value);
value = atiixp_rd(sc, ATI_REG_6CH_REORDER);
value &= ~ATI_REG_6CH_REORDER_EN;
atiixp_wr(sc, ATI_REG_6CH_REORDER, value);
}
2005-12-25 00:43:03 +00:00
ch->fmt = format;
atiixp_unlock(sc);
return (0);
}
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
static uint32_t
atiixp_chan_setspeed(kobj_t obj, void *data, uint32_t spd)
{
/* XXX We're supposed to do VRA/DRA processing right here */
return (ATI_IXP_BASE_RATE);
}
static int
atiixp_chan_setfragments(kobj_t obj, void *data,
uint32_t blksz, uint32_t blkcnt)
{
struct atiixp_chinfo *ch = data;
struct atiixp_info *sc = ch->parent;
blksz &= ATI_IXP_BLK_ALIGN;
if (blksz > (sndbuf_getmaxsize(ch->buffer) / ATI_IXP_DMA_CHSEGS_MIN))
blksz = sndbuf_getmaxsize(ch->buffer) / ATI_IXP_DMA_CHSEGS_MIN;
if (blksz < ATI_IXP_BLK_MIN)
blksz = ATI_IXP_BLK_MIN;
if (blkcnt > ATI_IXP_DMA_CHSEGS_MAX)
blkcnt = ATI_IXP_DMA_CHSEGS_MAX;
if (blkcnt < ATI_IXP_DMA_CHSEGS_MIN)
blkcnt = ATI_IXP_DMA_CHSEGS_MIN;
while ((blksz * blkcnt) > sndbuf_getmaxsize(ch->buffer)) {
if ((blkcnt >> 1) >= ATI_IXP_DMA_CHSEGS_MIN)
blkcnt >>= 1;
else if ((blksz >> 1) >= ATI_IXP_BLK_MIN)
blksz >>= 1;
else
break;
}
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
if ((sndbuf_getblksz(ch->buffer) != blksz ||
sndbuf_getblkcnt(ch->buffer) != blkcnt) &&
sndbuf_resize(ch->buffer, blkcnt, blksz) != 0)
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
device_printf(sc->dev, "%s: failed blksz=%u blkcnt=%u\n",
__func__, blksz, blkcnt);
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
ch->blksz = sndbuf_getblksz(ch->buffer);
ch->blkcnt = sndbuf_getblkcnt(ch->buffer);
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
return (0);
}
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
static uint32_t
atiixp_chan_setblocksize(kobj_t obj, void *data, uint32_t blksz)
{
struct atiixp_chinfo *ch = data;
struct atiixp_info *sc = ch->parent;
atiixp_chan_setfragments(obj, data, blksz, sc->blkcnt);
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
return (ch->blksz);
}
static void
atiixp_buildsgdt(struct atiixp_chinfo *ch)
{
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
struct atiixp_info *sc = ch->parent;
uint32_t addr, blksz, blkcnt;
int i;
addr = sndbuf_getbufaddr(ch->buffer);
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
if (sc->polling != 0) {
blksz = ch->blksz * ch->blkcnt;
blkcnt = 1;
} else {
blksz = ch->blksz;
blkcnt = ch->blkcnt;
}
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
for (i = 0; i < blkcnt; i++) {
ch->sgd_table[i].addr = htole32(addr + (i * blksz));
ch->sgd_table[i].status = htole16(0);
ch->sgd_table[i].size = htole16(blksz >> 2);
ch->sgd_table[i].next = htole32((uint32_t)ch->sgd_addr +
(((i + 1) % blkcnt) * sizeof(struct atiixp_dma_op)));
}
}
static __inline uint32_t
atiixp_dmapos(struct atiixp_chinfo *ch)
{
struct atiixp_info *sc = ch->parent;
uint32_t reg, addr, sz, retry;
volatile uint32_t ptr;
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
reg = ch->dt_cur_bit;
addr = sndbuf_getbufaddr(ch->buffer);
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
sz = ch->blkcnt * ch->blksz;
retry = ATI_IXP_DMA_RETRY_MAX;
do {
ptr = atiixp_rd(sc, reg);
if (ptr < addr)
continue;
ptr -= addr;
if (ptr < sz) {
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
#if 0
#ifdef ATI_IXP_DEBUG
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
if ((ptr & ~(ch->blksz - 1)) != ch->ptr) {
uint32_t delta;
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
delta = (sz + ptr - ch->prevptr) % sz;
#ifndef ATI_IXP_DEBUG_VERBOSE
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
if (delta < ch->blksz)
#endif
device_printf(sc->dev,
"PCMDIR_%s: incoherent DMA "
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
"prevptr=%u ptr=%u "
"ptr=%u blkcnt=%u "
"[delta=%u != blksz=%u] "
"(%s)\n",
(ch->dir == PCMDIR_PLAY) ?
"PLAY" : "REC",
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
ch->prevptr, ptr,
ch->ptr, ch->blkcnt,
delta, ch->blksz,
(delta < ch->blksz) ?
"OVERLAPPED!" : "Ok");
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
ch->ptr = ptr & ~(ch->blksz - 1);
}
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
ch->prevptr = ptr;
#endif
#endif
return (ptr);
}
} while (--retry);
device_printf(sc->dev, "PCMDIR_%s: invalid DMA pointer ptr=%u\n",
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
(ch->dir == PCMDIR_PLAY) ? "PLAY" : "REC", ptr);
return (0);
}
static __inline int
atiixp_poll_channel(struct atiixp_chinfo *ch)
{
uint32_t sz, delta;
volatile uint32_t ptr;
if (!(ch->flags & ATI_IXP_CHN_RUNNING))
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
return (0);
sz = ch->blksz * ch->blkcnt;
ptr = atiixp_dmapos(ch);
ch->ptr = ptr;
ptr %= sz;
ptr &= ~(ch->blksz - 1);
delta = (sz + ptr - ch->prevptr) % sz;
if (delta < ch->blksz)
return (0);
ch->prevptr = ptr;
return (1);
}
#define atiixp_chan_active(sc) (((sc)->pch.flags | (sc)->rch.flags) & \
ATI_IXP_CHN_RUNNING)
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
static void
atiixp_poll_callback(void *arg)
{
struct atiixp_info *sc = arg;
uint32_t trigger = 0;
if (sc == NULL)
return;
atiixp_lock(sc);
if (sc->polling == 0 || atiixp_chan_active(sc) == 0) {
atiixp_unlock(sc);
return;
}
trigger |= (atiixp_poll_channel(&sc->pch) != 0) ? 1 : 0;
trigger |= (atiixp_poll_channel(&sc->rch) != 0) ? 2 : 0;
/* XXX */
callout_reset(&sc->poll_timer, 1/*sc->poll_ticks*/,
atiixp_poll_callback, sc);
atiixp_unlock(sc);
if (trigger & 1)
chn_intr(sc->pch.channel);
if (trigger & 2)
chn_intr(sc->rch.channel);
}
static int
atiixp_chan_trigger(kobj_t obj, void *data, int go)
{
struct atiixp_chinfo *ch = data;
struct atiixp_info *sc = ch->parent;
uint32_t value;
int pollticks;
if (!PCMTRIG_COMMON(go))
return (0);
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
atiixp_lock(sc);
switch (go) {
case PCMTRIG_START:
atiixp_flush_dma(ch);
atiixp_buildsgdt(ch);
atiixp_wr(sc, ch->linkptr_bit, 0);
atiixp_enable_dma(ch);
atiixp_wr(sc, ch->linkptr_bit,
(uint32_t)ch->sgd_addr | ATI_REG_LINKPTR_EN);
if (sc->polling != 0) {
ch->ptr = 0;
ch->prevptr = 0;
pollticks = ((uint64_t)hz * ch->blksz) /
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
((uint64_t)sndbuf_getalign(ch->buffer) *
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
sndbuf_getspd(ch->buffer));
pollticks >>= 2;
if (pollticks > hz)
pollticks = hz;
if (pollticks < 1)
pollticks = 1;
if (atiixp_chan_active(sc) == 0 ||
pollticks < sc->poll_ticks) {
if (bootverbose) {
if (atiixp_chan_active(sc) == 0)
device_printf(sc->dev,
"%s: pollticks=%d\n",
__func__, pollticks);
else
device_printf(sc->dev,
"%s: pollticks %d -> %d\n",
__func__, sc->poll_ticks,
pollticks);
}
sc->poll_ticks = pollticks;
callout_reset(&sc->poll_timer, 1,
atiixp_poll_callback, sc);
}
}
ch->flags |= ATI_IXP_CHN_RUNNING;
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
break;
case PCMTRIG_STOP:
case PCMTRIG_ABORT:
atiixp_disable_dma(ch);
atiixp_flush_dma(ch);
ch->flags &= ~ATI_IXP_CHN_RUNNING;
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
if (sc->polling != 0) {
if (atiixp_chan_active(sc) == 0) {
callout_stop(&sc->poll_timer);
sc->poll_ticks = 1;
} else {
if (sc->pch.flags & ATI_IXP_CHN_RUNNING)
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
ch = &sc->pch;
else
ch = &sc->rch;
pollticks = ((uint64_t)hz * ch->blksz) /
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
((uint64_t)sndbuf_getalign(ch->buffer) *
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
sndbuf_getspd(ch->buffer));
pollticks >>= 2;
if (pollticks > hz)
pollticks = hz;
if (pollticks < 1)
pollticks = 1;
if (pollticks > sc->poll_ticks) {
if (bootverbose)
device_printf(sc->dev,
"%s: pollticks %d -> %d\n",
__func__, sc->poll_ticks,
pollticks);
sc->poll_ticks = pollticks;
callout_reset(&sc->poll_timer,
1, atiixp_poll_callback,
sc);
}
}
}
break;
default:
atiixp_unlock(sc);
return (0);
break;
}
/* Update bus busy status */
value = atiixp_rd(sc, ATI_REG_IER);
if (atiixp_rd(sc, ATI_REG_CMD) & (ATI_REG_CMD_SEND_EN |
ATI_REG_CMD_RECEIVE_EN | ATI_REG_CMD_SPDF_OUT_EN))
value |= ATI_REG_IER_SET_BUS_BUSY;
else
value &= ~ATI_REG_IER_SET_BUS_BUSY;
atiixp_wr(sc, ATI_REG_IER, value);
atiixp_unlock(sc);
return (0);
}
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
static uint32_t
atiixp_chan_getptr(kobj_t obj, void *data)
{
struct atiixp_chinfo *ch = data;
struct atiixp_info *sc = ch->parent;
uint32_t ptr;
atiixp_lock(sc);
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
if (sc->polling != 0)
ptr = ch->ptr;
else
ptr = atiixp_dmapos(ch);
atiixp_unlock(sc);
return (ptr);
}
static struct pcmchan_caps *
atiixp_chan_getcaps(kobj_t obj, void *data)
{
struct atiixp_chinfo *ch = data;
if (ch->caps_32bit)
return (&atiixp_caps_32bit);
return (&atiixp_caps);
}
static kobj_method_t atiixp_chan_methods[] = {
KOBJMETHOD(channel_init, atiixp_chan_init),
KOBJMETHOD(channel_setformat, atiixp_chan_setformat),
KOBJMETHOD(channel_setspeed, atiixp_chan_setspeed),
KOBJMETHOD(channel_setblocksize, atiixp_chan_setblocksize),
KOBJMETHOD(channel_setfragments, atiixp_chan_setfragments),
KOBJMETHOD(channel_trigger, atiixp_chan_trigger),
KOBJMETHOD(channel_getptr, atiixp_chan_getptr),
KOBJMETHOD(channel_getcaps, atiixp_chan_getcaps),
Sound Mega-commit. Expect further cleanup until code freeze. For a slightly thorough explaination, please refer to [1] http://people.freebsd.org/~ariff/SOUND_4.TXT.html . Summary of changes includes: 1 Volume Per-Channel (vpc). Provides private / standalone volume control unique per-stream pcm channel without touching master volume / pcm. Applications can directly use SNDCTL_DSP_[GET|SET][PLAY|REC]VOL, or for backwards compatibility, SOUND_MIXER_PCM through the opened dsp device instead of /dev/mixer. Special "bypass" mode is enabled through /dev/mixer which will automatically detect if the adjustment is made through /dev/mixer and forward its request to this private volume controller. Changes to this volume object will not interfere with other channels. Requirements: - SNDCTL_DSP_[GET|SET][PLAY|REC]_VOL are newer ioctls (OSSv4) which require specific application modifications (preferred). - No modifications required for using bypass mode, so applications like mplayer or xmms should work out of the box. Kernel hints: - hint.pcm.%d.vpc (0 = disable vpc). Kernel sysctls: - hw.snd.vpc_mixer_bypass (default: 1). Enable or disable /dev/mixer bypass mode. - hw.snd.vpc_autoreset (default: 1). By default, closing/opening /dev/dsp will reset the volume back to 0 db gain/attenuation. Setting this to 0 will preserve its settings across device closing/opening. - hw.snd.vpc_reset (default: 0). Panic/reset button to reset all volume settings back to 0 db. - hw.snd.vpc_0db (default: 45). 0 db relative to linear mixer value. 2 High quality fixed-point Bandlimited SINC sampling rate converter, based on Julius O'Smith's Digital Audio Resampling - http://ccrma.stanford.edu/~jos/resample/. It includes a filter design script written in awk (the clumsiest joke I've ever written) - 100% 32bit fixed-point, 64bit accumulator. - Possibly among the fastest (if not fastest) of its kind. - Resampling quality is tunable, either runtime or during kernel compilation (FEEDER_RATE_PRESETS). - Quality can be further customized during kernel compilation by defining FEEDER_RATE_PRESETS in /etc/make.conf. Kernel sysctls: - hw.snd.feeder_rate_quality. 0 - Zero-order Hold (ZOH). Fastest, bad quality. 1 - Linear Interpolation (LINEAR). Slightly slower than ZOH, better quality but still does not eliminate aliasing. 2 - (and above) - Sinc Interpolation(SINC). Best quality. SINC quality always start from 2 and above. Rough quality comparisons: - http://people.freebsd.org/~ariff/z_comparison/ 3 Bit-perfect mode. Bypasses all feeder/dsp effects. Pure sound will be directly fed into the hardware. 4 Parametric (compile time) Software Equalizer (Bass/Treble mixer). Can be customized by defining FEEDER_EQ_PRESETS in /etc/make.conf. 5 Transparent/Adaptive Virtual Channel. Now you don't have to disable vchans in order to make digital format pass through. It also makes vchans more dynamic by choosing a better format/rate among all the concurrent streams, which means that dev.pcm.X.play.vchanformat/rate becomes sort of optional. 6 Exclusive Stream, with special open() mode O_EXCL. This will "mute" other concurrent vchan streams and only allow a single channel with O_EXCL set to keep producing sound. Other Changes: * most feeder_* stuffs are compilable in userland. Let's not speculate whether we should go all out for it (save that for FreeBSD 16.0-RELEASE). * kobj signature fixups, thanks to Andriy Gapon <avg@freebsd.org> * pull out channel mixing logic out of vchan.c and create its own feeder_mixer for world justice. * various refactoring here and there, for good or bad. * activation of few more OSSv4 ioctls() (see [1] above). * opt_snd.h for possible compile time configuration: (mostly for debugging purposes, don't try these at home) SND_DEBUG SND_DIAGNOSTIC SND_FEEDER_MULTIFORMAT SND_FEEDER_FULL_MULTIFORMAT SND_FEEDER_RATE_HP SND_PCM_64 SND_OLDSTEREO Manual page updates are on the way. Tested by: joel, Olivier SMEDTS <olivier at gid0 d org>, too many unsung / unnamed heroes.
2009-06-07 19:12:08 +00:00
KOBJMETHOD_END
};
CHANNEL_DECLARE(atiixp_chan);
/*
* PCI driver interface
*/
static void
atiixp_intr(void *p)
{
struct atiixp_info *sc = p;
uint32_t status, enable, detected_codecs;
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
uint32_t trigger = 0;
atiixp_lock(sc);
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
if (sc->polling != 0) {
atiixp_unlock(sc);
return;
}
status = atiixp_rd(sc, ATI_REG_ISR);
if (status == 0) {
atiixp_unlock(sc);
return;
}
if ((status & ATI_REG_ISR_OUT_STATUS) &&
(sc->pch.flags & ATI_IXP_CHN_RUNNING))
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
trigger |= 1;
if ((status & ATI_REG_ISR_IN_STATUS) &&
(sc->rch.flags & ATI_IXP_CHN_RUNNING))
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
trigger |= 2;
#if 0
if (status & ATI_REG_ISR_IN_XRUN) {
device_printf(sc->dev,
"Recieve IN XRUN interrupt\n");
}
if (status & ATI_REG_ISR_OUT_XRUN) {
device_printf(sc->dev,
"Recieve OUT XRUN interrupt\n");
}
#endif
if (status & CODEC_CHECK_BITS) {
/* mark missing codecs as not ready */
detected_codecs = status & CODEC_CHECK_BITS;
sc->codec_not_ready_bits |= detected_codecs;
2007-10-12 06:03:46 +00:00
/* disable detected interrupt sources */
enable = atiixp_rd(sc, ATI_REG_IER);
enable &= ~detected_codecs;
atiixp_wr(sc, ATI_REG_IER, enable);
wakeup(sc);
}
/* acknowledge */
atiixp_wr(sc, ATI_REG_ISR, status);
atiixp_unlock(sc);
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
if (trigger & 1)
chn_intr(sc->pch.channel);
if (trigger & 2)
chn_intr(sc->rch.channel);
}
static void
atiixp_dma_cb(void *p, bus_dma_segment_t *bds, int a, int b)
{
struct atiixp_info *sc = (struct atiixp_info *)p;
sc->sgd_addr = bds->ds_addr;
}
static void
atiixp_chip_pre_init(struct atiixp_info *sc)
{
uint32_t value;
atiixp_lock(sc);
/* disable interrupts */
atiixp_disable_interrupts(sc);
/* clear all DMA enables (preserving rest of settings) */
value = atiixp_rd(sc, ATI_REG_CMD);
value &= ~(ATI_REG_CMD_IN_DMA_EN | ATI_REG_CMD_OUT_DMA_EN |
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
ATI_REG_CMD_SPDF_OUT_EN );
atiixp_wr(sc, ATI_REG_CMD, value);
/* reset aclink */
atiixp_reset_aclink(sc);
sc->codec_not_ready_bits = 0;
/* enable all codecs to interrupt as well as the new frame interrupt */
atiixp_wr(sc, ATI_REG_IER, CODEC_CHECK_BITS);
atiixp_unlock(sc);
}
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
static int
sysctl_atiixp_polling(SYSCTL_HANDLER_ARGS)
{
struct atiixp_info *sc;
device_t dev;
int err, val;
dev = oidp->oid_arg1;
sc = pcm_getdevinfo(dev);
if (sc == NULL)
return (EINVAL);
atiixp_lock(sc);
val = sc->polling;
atiixp_unlock(sc);
err = sysctl_handle_int(oidp, &val, 0, req);
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
if (err || req->newptr == NULL)
return (err);
if (val < 0 || val > 1)
return (EINVAL);
atiixp_lock(sc);
if (val != sc->polling) {
if (atiixp_chan_active(sc) != 0)
err = EBUSY;
else if (val == 0) {
atiixp_enable_interrupts(sc);
sc->polling = 0;
DELAY(1000);
} else {
atiixp_disable_interrupts(sc);
sc->polling = 1;
DELAY(1000);
}
}
atiixp_unlock(sc);
return (err);
}
static void
atiixp_chip_post_init(void *arg)
{
struct atiixp_info *sc = (struct atiixp_info *)arg;
uint32_t subdev;
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
int i, timeout, found, polling;
char status[SND_STATUSLEN];
atiixp_lock(sc);
if (sc->delayed_attach.ich_func) {
config_intrhook_disestablish(&sc->delayed_attach);
sc->delayed_attach.ich_func = NULL;
}
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
polling = sc->polling;
sc->polling = 0;
timeout = 10;
if (sc->codec_not_ready_bits == 0) {
/* wait for the interrupts to happen */
do {
msleep(sc, sc->lock, PWAIT, "ixpslp", max(hz / 10, 1));
if (sc->codec_not_ready_bits != 0)
break;
} while (--timeout);
}
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
sc->polling = polling;
atiixp_disable_interrupts(sc);
if (sc->codec_not_ready_bits == 0 && timeout == 0) {
device_printf(sc->dev,
"WARNING: timeout during codec detection; "
"codecs might be present but haven't interrupted\n");
atiixp_unlock(sc);
goto postinitbad;
}
found = 0;
/*
* ATI IXP can have upto 3 codecs, but single codec should be
* suffice for now.
*/
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
if (!(sc->codec_not_ready_bits & ATI_REG_ISR_CODEC0_NOT_READY)) {
/* codec 0 present */
sc->codec_found++;
sc->codec_idx = 0;
found++;
}
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
if (!(sc->codec_not_ready_bits & ATI_REG_ISR_CODEC1_NOT_READY)) {
/* codec 1 present */
sc->codec_found++;
}
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
if (!(sc->codec_not_ready_bits & ATI_REG_ISR_CODEC2_NOT_READY)) {
/* codec 2 present */
sc->codec_found++;
}
atiixp_unlock(sc);
if (found == 0)
goto postinitbad;
/* create/init mixer */
sc->codec = AC97_CREATE(sc->dev, sc, atiixp_ac97);
if (sc->codec == NULL)
goto postinitbad;
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
subdev = (pci_get_subdevice(sc->dev) << 16) |
pci_get_subvendor(sc->dev);
switch (subdev) {
case 0x11831043: /* ASUS A6R */
case 0x2043161f: /* Maxselect x710s - http://maxselect.ru/ */
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
ac97_setflags(sc->codec, ac97_getflags(sc->codec) |
AC97_F_EAPD_INV);
break;
default:
break;
}
mixer_init(sc->dev, ac97_getmixerclass(), sc->codec);
if (pcm_register(sc->dev, sc, ATI_IXP_NPCHAN, ATI_IXP_NRCHAN))
goto postinitbad;
for (i = 0; i < ATI_IXP_NPCHAN; i++)
pcm_addchan(sc->dev, PCMDIR_PLAY, &atiixp_chan_class, sc);
for (i = 0; i < ATI_IXP_NRCHAN; i++)
pcm_addchan(sc->dev, PCMDIR_REC, &atiixp_chan_class, sc);
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_PROC(device_get_sysctl_ctx(sc->dev),
SYSCTL_CHILDREN(device_get_sysctl_tree(sc->dev)), OID_AUTO,
"polling", CTLTYPE_INT | CTLFLAG_RW, sc->dev, sizeof(sc->dev),
sysctl_atiixp_polling, "I", "Enable polling mode");
Use uintmax_t (typedef'd to rman_res_t type) for rman ranges. On some architectures, u_long isn't large enough for resource definitions. Particularly, powerpc and arm allow 36-bit (or larger) physical addresses, but type `long' is only 32-bit. This extends rman's resources to uintmax_t. With this change, any resource can feasibly be placed anywhere in physical memory (within the constraints of the driver). Why uintmax_t and not something machine dependent, or uint64_t? Though it's possible for uintmax_t to grow, it's highly unlikely it will become 128-bit on 32-bit architectures. 64-bit architectures should have plenty of RAM to absorb the increase on resource sizes if and when this occurs, and the number of resources on memory-constrained systems should be sufficiently small as to not pose a drastic overhead. That being said, uintmax_t was chosen for source clarity. If it's specified as uint64_t, all printf()-like calls would either need casts to uintmax_t, or be littered with PRI*64 macros. Casts to uintmax_t aren't horrible, but it would also bake into the API for resource_list_print_type() either a hidden assumption that entries get cast to uintmax_t for printing, or these calls would need the PRI*64 macros. Since source code is meant to be read more often than written, I chose the clearest path of simply using uintmax_t. Tested on a PowerPC p5020-based board, which places all device resources in 0xfxxxxxxxx, and has 8GB RAM. Regression tested on qemu-system-i386 Regression tested on qemu-system-mips (malta profile) Tested PAE and devinfo on virtualbox (live CD) Special thanks to bz for his testing on ARM. Reviewed By: bz, jhb (previous) Relnotes: Yes Sponsored by: Alex Perez/Inertial Computing Differential Revision: https://reviews.freebsd.org/D4544
2016-03-18 01:28:41 +00:00
snprintf(status, SND_STATUSLEN, "at memory 0x%jx irq %jd %s",
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
rman_get_start(sc->reg), rman_get_start(sc->irq),
PCM_KLDSTRING(snd_atiixp));
pcm_setstatus(sc->dev, status);
atiixp_lock(sc);
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
if (sc->polling == 0)
atiixp_enable_interrupts(sc);
atiixp_unlock(sc);
return;
postinitbad:
atiixp_release_resource(sc);
}
static void
atiixp_release_resource(struct atiixp_info *sc)
{
if (sc == NULL)
return;
if (sc->registered_channels != 0) {
atiixp_lock(sc);
sc->polling = 0;
callout_stop(&sc->poll_timer);
atiixp_unlock(sc);
callout_drain(&sc->poll_timer);
}
if (sc->codec) {
ac97_destroy(sc->codec);
sc->codec = NULL;
}
if (sc->ih) {
bus_teardown_intr(sc->dev, sc->irq, sc->ih);
sc->ih = NULL;
}
if (sc->reg) {
bus_release_resource(sc->dev, sc->regtype, sc->regid, sc->reg);
sc->reg = NULL;
}
if (sc->irq) {
bus_release_resource(sc->dev, SYS_RES_IRQ, sc->irqid, sc->irq);
sc->irq = NULL;
}
if (sc->parent_dmat) {
bus_dma_tag_destroy(sc->parent_dmat);
sc->parent_dmat = NULL;
}
if (sc->sgd_addr) {
bus_dmamap_unload(sc->sgd_dmat, sc->sgd_dmamap);
sc->sgd_addr = 0;
}
if (sc->sgd_table) {
bus_dmamem_free(sc->sgd_dmat, sc->sgd_table, sc->sgd_dmamap);
sc->sgd_table = NULL;
}
if (sc->sgd_dmat) {
bus_dma_tag_destroy(sc->sgd_dmat);
sc->sgd_dmat = NULL;
}
if (sc->lock) {
snd_mtxfree(sc->lock);
sc->lock = NULL;
}
free(sc, M_DEVBUF);
}
static int
atiixp_pci_probe(device_t dev)
{
int i;
uint16_t devid, vendor;
vendor = pci_get_vendor(dev);
devid = pci_get_device(dev);
for (i = 0; i < sizeof(atiixp_hw) / sizeof(atiixp_hw[0]); i++) {
if (vendor == atiixp_hw[i].vendor &&
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
devid == atiixp_hw[i].devid) {
device_set_desc(dev, atiixp_hw[i].desc);
return (BUS_PROBE_DEFAULT);
}
}
return (ENXIO);
}
static int
atiixp_pci_attach(device_t dev)
{
struct atiixp_info *sc;
int i;
sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK | M_ZERO);
Fix severe out-of-bound mtx "type" pointer, causing WITNESS refcount confusions and panic provided that the following conditions are met: 1) WITNESS is enabled (watch/trace). 2) Using modules, instead of statically linked (Not a strict requirement, but easier to reproduce this way). 3) 2 or more modules share the same mtx type ("sound softc"). - They might share the same name (strcmp() == 0), but it always point to different address. 4) Repetitive kldunload/load on any module that shares the same mtx type (Not a strict requirement, but easier to reproduce this way). Consider module A and module B: - From enroll() - subr_witness.c: * Load module A. Everything seems fine right now. wA-w_refcount == 1 ; wA-w_name = "sound softc" * Load module B. * w->w_name == description will always fail. ("sound softc" from A and B point to different address). * wA->w_refcount > 0 && strcmp(description, wA->w_name) == 0 * enroll() will return wA instead of returning (possibly unique) wB. wA->w_refcount++ , == 2. * Unload module A, mtx_destroy(), wA->w_name become invalid, but wA->w_refcount-- become 1 instead of 0. wA will not be removed from witness list. * Some other places call mtx_init(), iterating witness list, found wA, failed on wA->w_name == description * wA->w_refcount > 0 && strcmp(description, wA->w_name) * Panic on strcmp() since wA->w_name no longer point to valid address. Note that this could happened in other places as well, not just sound (eg. consider lots of drivers that share simmilar MTX_NETWORK_LOCK). Solutions (for sound case): 1) Provide unique mtx type string for each mutex creation (chosen) or 2) Put "sound softc" global variable somewhere and use it.
2007-03-15 16:41:27 +00:00
sc->lock = snd_mtxcreate(device_get_nameunit(dev), "snd_atiixp softc");
sc->dev = dev;
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
callout_init(&sc->poll_timer, 1);
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
sc->poll_ticks = 1;
if (resource_int_value(device_get_name(sc->dev),
device_get_unit(sc->dev), "polling", &i) == 0 && i != 0)
sc->polling = 1;
else
sc->polling = 0;
pci_enable_busmaster(dev);
sc->regid = PCIR_BAR(0);
sc->regtype = SYS_RES_MEMORY;
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
sc->reg = bus_alloc_resource_any(dev, sc->regtype,
&sc->regid, RF_ACTIVE);
if (!sc->reg) {
device_printf(dev, "unable to allocate register space\n");
goto bad;
}
sc->st = rman_get_bustag(sc->reg);
sc->sh = rman_get_bushandle(sc->reg);
sc->bufsz = pcm_getbuffersize(dev, ATI_IXP_BUFSZ_MIN,
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
ATI_IXP_BUFSZ_DEFAULT, ATI_IXP_BUFSZ_MAX);
sc->irqid = 0;
sc->irq = bus_alloc_resource_any(dev, SYS_RES_IRQ, &sc->irqid,
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
RF_ACTIVE | RF_SHAREABLE);
if (!sc->irq || snd_setup_intr(dev, sc->irq, INTR_MPSAFE,
atiixp_intr, sc, &sc->ih)) {
device_printf(dev, "unable to map interrupt\n");
goto bad;
}
/*
* Let the user choose the best DMA segments.
*/
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
if (resource_int_value(device_get_name(dev),
device_get_unit(dev), "blocksize", &i) == 0 && i > 0) {
i &= ATI_IXP_BLK_ALIGN;
if (i < ATI_IXP_BLK_MIN)
i = ATI_IXP_BLK_MIN;
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
sc->blkcnt = sc->bufsz / i;
i = 0;
while (sc->blkcnt >> i)
i++;
sc->blkcnt = 1 << (i - 1);
if (sc->blkcnt < ATI_IXP_DMA_CHSEGS_MIN)
sc->blkcnt = ATI_IXP_DMA_CHSEGS_MIN;
else if (sc->blkcnt > ATI_IXP_DMA_CHSEGS_MAX)
sc->blkcnt = ATI_IXP_DMA_CHSEGS_MAX;
} else
sc->blkcnt = ATI_IXP_DMA_CHSEGS;
/*
* DMA tag for scatter-gather buffers and link pointers
*/
if (bus_dma_tag_create(/*parent*/bus_get_dma_tag(dev), /*alignment*/2,
/*boundary*/0,
/*lowaddr*/BUS_SPACE_MAXADDR_32BIT,
/*highaddr*/BUS_SPACE_MAXADDR,
/*filter*/NULL, /*filterarg*/NULL,
/*maxsize*/sc->bufsz, /*nsegments*/1, /*maxsegz*/0x3ffff,
/*flags*/0, /*lockfunc*/NULL,
/*lockarg*/NULL, &sc->parent_dmat) != 0) {
device_printf(dev, "unable to create dma tag\n");
goto bad;
}
if (bus_dma_tag_create(/*parent*/bus_get_dma_tag(dev), /*alignment*/2,
/*boundary*/0,
/*lowaddr*/BUS_SPACE_MAXADDR_32BIT,
/*highaddr*/BUS_SPACE_MAXADDR,
/*filter*/NULL, /*filterarg*/NULL,
/*maxsize*/ATI_IXP_DMA_CHSEGS_MAX * ATI_IXP_NCHANS *
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
sizeof(struct atiixp_dma_op),
/*nsegments*/1, /*maxsegz*/0x3ffff,
/*flags*/0, /*lockfunc*/NULL,
/*lockarg*/NULL, &sc->sgd_dmat) != 0) {
device_printf(dev, "unable to create dma tag\n");
goto bad;
}
if (bus_dmamem_alloc(sc->sgd_dmat, (void **)&sc->sgd_table,
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
BUS_DMA_NOWAIT, &sc->sgd_dmamap) == -1)
goto bad;
if (bus_dmamap_load(sc->sgd_dmat, sc->sgd_dmamap, sc->sgd_table,
ATI_IXP_DMA_CHSEGS_MAX * ATI_IXP_NCHANS *
sizeof(struct atiixp_dma_op), atiixp_dma_cb, sc, 0))
goto bad;
atiixp_chip_pre_init(sc);
sc->delayed_attach.ich_func = atiixp_chip_post_init;
sc->delayed_attach.ich_arg = sc;
if (cold == 0 ||
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
config_intrhook_establish(&sc->delayed_attach) != 0) {
sc->delayed_attach.ich_func = NULL;
atiixp_chip_post_init(sc);
}
return (0);
bad:
atiixp_release_resource(sc);
return (ENXIO);
}
static int
atiixp_pci_detach(device_t dev)
{
int r;
struct atiixp_info *sc;
sc = pcm_getdevinfo(dev);
if (sc != NULL) {
if (sc->codec != NULL) {
r = pcm_unregister(dev);
if (r)
return (r);
}
sc->codec = NULL;
if (sc->st != 0 && sc->sh != 0)
atiixp_disable_interrupts(sc);
atiixp_release_resource(sc);
}
return (0);
}
2005-12-25 00:43:03 +00:00
static int
atiixp_pci_suspend(device_t dev)
{
struct atiixp_info *sc = pcm_getdevinfo(dev);
uint32_t value;
/* quickly disable interrupts and save channels active state */
atiixp_lock(sc);
atiixp_disable_interrupts(sc);
atiixp_unlock(sc);
/* stop everything */
if (sc->pch.flags & ATI_IXP_CHN_RUNNING) {
2005-12-25 00:43:03 +00:00
atiixp_chan_trigger(NULL, &sc->pch, PCMTRIG_STOP);
sc->pch.flags |= ATI_IXP_CHN_SUSPEND;
}
if (sc->rch.flags & ATI_IXP_CHN_RUNNING) {
2005-12-25 00:43:03 +00:00
atiixp_chan_trigger(NULL, &sc->rch, PCMTRIG_STOP);
sc->rch.flags |= ATI_IXP_CHN_SUSPEND;
}
2005-12-25 00:43:03 +00:00
/* power down aclink and pci bus */
atiixp_lock(sc);
value = atiixp_rd(sc, ATI_REG_CMD);
value |= ATI_REG_CMD_POWERDOWN | ATI_REG_CMD_AC_RESET;
atiixp_wr(sc, ATI_REG_CMD, ATI_REG_CMD_POWERDOWN);
atiixp_unlock(sc);
return (0);
2005-12-25 00:43:03 +00:00
}
static int
atiixp_pci_resume(device_t dev)
{
struct atiixp_info *sc = pcm_getdevinfo(dev);
atiixp_lock(sc);
/* reset / power up aclink */
atiixp_reset_aclink(sc);
atiixp_unlock(sc);
if (mixer_reinit(dev) == -1) {
device_printf(dev, "unable to reinitialize the mixer\n");
return (ENXIO);
2005-12-25 00:43:03 +00:00
}
/*
* Resume channel activities. Reset channel format regardless
* of its previous state.
*/
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
if (sc->pch.channel != NULL) {
if (sc->pch.fmt != 0)
2005-12-25 00:43:03 +00:00
atiixp_chan_setformat(NULL, &sc->pch, sc->pch.fmt);
if (sc->pch.flags & ATI_IXP_CHN_SUSPEND) {
sc->pch.flags &= ~ATI_IXP_CHN_SUSPEND;
2005-12-25 00:43:03 +00:00
atiixp_chan_trigger(NULL, &sc->pch, PCMTRIG_START);
}
2005-12-25 00:43:03 +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
if (sc->rch.channel != NULL) {
if (sc->rch.fmt != 0)
2005-12-25 00:43:03 +00:00
atiixp_chan_setformat(NULL, &sc->rch, sc->rch.fmt);
if (sc->rch.flags & ATI_IXP_CHN_SUSPEND) {
sc->rch.flags &= ~ATI_IXP_CHN_SUSPEND;
2005-12-25 00:43:03 +00:00
atiixp_chan_trigger(NULL, &sc->rch, PCMTRIG_START);
}
2005-12-25 00:43:03 +00:00
}
/* enable interrupts */
atiixp_lock(sc);
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
if (sc->polling == 0)
atiixp_enable_interrupts(sc);
2005-12-25 00:43:03 +00:00
atiixp_unlock(sc);
return (0);
2005-12-25 00:43:03 +00:00
}
static device_method_t atiixp_methods[] = {
DEVMETHOD(device_probe, atiixp_pci_probe),
DEVMETHOD(device_attach, atiixp_pci_attach),
DEVMETHOD(device_detach, atiixp_pci_detach),
2005-12-25 00:43:03 +00:00
DEVMETHOD(device_suspend, atiixp_pci_suspend),
DEVMETHOD(device_resume, atiixp_pci_resume),
{ 0, 0 }
};
static driver_t atiixp_driver = {
"pcm",
atiixp_methods,
PCM_SOFTC_SIZE,
};
DRIVER_MODULE(snd_atiixp, pci, atiixp_driver, pcm_devclass, 0, 0);
MODULE_DEPEND(snd_atiixp, sound, SOUND_MINVER, SOUND_PREFVER, SOUND_MAXVER);
MODULE_VERSION(snd_atiixp, 1);