2.6.3-rc1: snd_intel8x0 still too fast

2.6.3-rc1: snd_intel8x0 still too fast

[Meelis Roos]

I tried 2.6.3-rc1 to see if the problems with intel 8x0 audio are fixed.
Nope, the sound is still at least twice as fast as normal. KDE login
sound, mplayer sound etc. This is I815 integrated audio on Intel
D815EEA2 mainboard, Debian unstable up-to-date. 2.4.latest kernel with
OSS is OK. 2.6 with ALSA was OK at about 2.6.0. 2.6 with OSS is also OK.

Also, I can load both i810_audio and snd_intel8x0 simultaneously in 2.6
- probably a resource management problem.

-- 
Meelis Roos (mroos@linux.ee)

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Bill Davidsen]

Meelis Roos wrote:
> I tried 2.6.3-rc1 to see if the problems with intel 8x0 audio are fixed.
> Nope, the sound is still at least twice as fast as normal. KDE login
> sound, mplayer sound etc. This is I815 integrated audio on Intel
> D815EEA2 mainboard, Debian unstable up-to-date. 2.4.latest kernel with
> OSS is OK. 2.6 with ALSA was OK at about 2.6.0. 2.6 with OSS is also OK.
> 
> Also, I can load both i810_audio and snd_intel8x0 simultaneously in 2.6
> - probably a resource management problem.

Odd, I only have the problem with xmms, mplayer works fine.

-- 
bill davidsen <davidsen@tmr.com>
   CTO TMR Associates, Inc
   Doing interesting things with small computers since 1979

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Takashi Iwai]

At Sat, 7 Feb 2004 17:10:31 +0200 (EET),
Meelis Roos wrote:
> 
> I tried 2.6.3-rc1 to see if the problems with intel 8x0 audio are fixed.
> Nope, the sound is still at least twice as fast as normal. KDE login
> sound, mplayer sound etc. This is I815 integrated audio on Intel
> D815EEA2 mainboard, Debian unstable up-to-date. 2.4.latest kernel with
> OSS is OK. 2.6 with ALSA was OK at about 2.6.0. 2.6 with OSS is also OK.

which ac97 codec chip?  check /proc/asound/card0/codec97#0/* files.
also, you must have seen a kernel message like
	intel8x0: clocking to 48000
check which value was detected.

> Also, I can load both i810_audio and snd_intel8x0 simultaneously in 2.6
> - probably a resource management problem.

hmm, i810_audio.c doesn't check the return value of request_region()...

--
Takashi Iwai <tiwai@suse.de>		ALSA Developer - www.alsa-project.org

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Meelis Roos]

Today mplayer was OK when I tested, didn't retry KDE login. But the
kernel is the same, I have not rebooted inbetween.

> which ac97 codec chip?  check /proc/asound/card0/codec97#0/* files.

i810_audio that I have currently loaded tells it's
ac97_codec: AC97 Audio codec, id: ADS96 (Analog Devices AD1885)

ALSA loaded, from /proc/asound/card0/codec97#0/ac97#0-0

0-0/0: Analog Devices AD1885

Capabilities     : -headphone out-
DAC resolution   : 16-bit
ADC resolution   : 16-bit
3D enhancement   : Analog Devices Phat Stereo

Current setup
Mic gain         : +0dB [+0dB]
POP path         : pre 3D
Sim. stereo      : off
3D enhancement   : off
Loudness         : off
Mono output      : MIX
Mic select       : Mic1
ADC/DAC loopback : off
Extended ID      : codec=0 rev=0 DSA=0 VRA
Extended status  : VRA
PCM front DAC    : 48000Hz
PCM ADC          : 48000Hz



AD18XX configuration
Unchained        : 0x1000,0x0000,0x0000
Chained          : 0x0000,0x0000,0x0000


And +regs:

0:00 = 0410
0:02 = bf3f
0:04 = bf3f
0:06 = 801f
0:08 = 0000
0:0a = 801e
0:0c = 801f
0:0e = 801f
0:10 = 9f1f
0:12 = 9f1f
0:14 = 9f1f
0:16 = 9f1f
0:18 = 9f1f
0:1a = 0000
0:1c = 0000
0:1e = 0000
0:20 = 0000
0:22 = 0000
0:24 = 0000
0:26 = 000f
0:28 = 0001
0:2a = 0001
0:2c = bb80
0:2e = 0000
0:30 = 0000
0:32 = bb80
0:34 = 0000
0:36 = 0000
0:38 = 0000
0:3a = 0000
0:3c = 0000
0:3e = 0000
0:40 = 0000
0:42 = 0000
0:44 = 0000
0:46 = 0000
0:48 = 0000
0:4a = 0000
0:4c = 0000
0:4e = 0000
0:50 = 0000
0:52 = 0000
0:54 = 0000
0:56 = 0000
0:58 = 0000
0:5a = 0000
0:5c = 0000
0:5e = 0000
0:60 = 0000
0:62 = 0000
0:64 = 0000
0:66 = 0000
0:68 = 0000
0:6a = 0000
0:6c = 0000
0:6e = 0000
0:70 = 2000
0:72 = 0300
0:74 = 1000
0:76 = 0404
0:78 = bb80
0:7a = bb80
0:7c = 4144
0:7e = 5360

> also, you must have seen a kernel message like
> 	intel8x0: clocking to 48000

Did try twice:

intel8x0_measure_ac97_clock: measured 50040 usecs
intel8x0: clocking to 41146

intel8x0_measure_ac97_clock: measured 49395 usecs
intel8x0: clocking to 41145


-- 
Meelis Roos (mroos@linux.ee)

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Takashi Iwai]

At Mon, 9 Feb 2004 20:35:12 +0200 (EET),
Meelis Roos wrote:
> 
> Today mplayer was OK when I tested, didn't retry KDE login. But the
> kernel is the same, I have not rebooted inbetween.
 
weird...  cpufreq is running?

> > also, you must have seen a kernel message like
> > 	intel8x0: clocking to 48000
> 
> Did try twice:
> 
> intel8x0_measure_ac97_clock: measured 50040 usecs
> intel8x0: clocking to 41146
> 
> intel8x0_measure_ac97_clock: measured 49395 usecs
> intel8x0: clocking to 41145

and now you got the correct output?
if so, you can pass the value via ac97_clock option.

i guess the high pitch playback occured because of the wrong detection
of ac97 clock by some reason.


--
Takashi Iwai <tiwai@suse.de>		ALSA Developer - www.alsa-project.org

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Meelis Roos]

> > Today mplayer was OK when I tested, didn't retry KDE login. But the
> > kernel is the same, I have not rebooted inbetween.
>
> weird...  cpufreq is running?

No, it's a desktop Celeron 900.

> > Did try twice:
> >
> > intel8x0_measure_ac97_clock: measured 50040 usecs
> > intel8x0: clocking to 41146
> >
> > intel8x0_measure_ac97_clock: measured 49395 usecs
> > intel8x0: clocking to 41145
>
> and now you got the correct output?
> if so, you can pass the value via ac97_clock option.

I did retest things more toroughly today after I booted up 2.6.3-rc2. I
focused on testing KDE login since this is known to break. Startup
scripts (discover) loaded also i810_audio, I unloaded it.

Plain bootup gave
PCI: Setting latency timer of device 0000:00:1f.5 to 64
intel8x0_measure_ac97_clock: measured 49671 usecs
intel8x0: clocking to 41136

and it did not work. Unloading and reloading snd-intel8x0 gave
PCI: Setting latency timer of device 0000:00:1f.5 to 64
intel8x0_measure_ac97_clock: measured 49281 usecs
intel8x0: clocking to 41139

and it still did not work (way too fast, probably even more than twice).
Now I tried
modprobe snd-intel8x0 ac97_clock=41140
It loads silently but still does not work.

Now I started to suspect that loading i810_audio might screw things up.
Removed the module and rebooted, no i810_audio loaded, still the same.

Reproduced the problem with mplayer too. It depends on the input file
bitrate!

mplayer /usr/share/sounds/KDE_Startup.wav
gives the same very fast sound since it's a 22 KHz mono sample.

mplayer -srate 48000 /usr/share/sounds/KDE_Startup.wav
works fine. Same applies to all files.

So maybe it's a userspace resampling issue? I see that the hardware
talks 44 KHz.

-- 
Meelis Roos (mroos@linux.ee)

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Takashi Iwai]

At Tue, 10 Feb 2004 13:22:25 +0200 (EET),
Meelis Roos wrote:
> 
> > > Today mplayer was OK when I tested, didn't retry KDE login. But the
> > > kernel is the same, I have not rebooted inbetween.
> >
> > weird...  cpufreq is running?
> 
> No, it's a desktop Celeron 900.
> 
> > > Did try twice:
> > >
> > > intel8x0_measure_ac97_clock: measured 50040 usecs
> > > intel8x0: clocking to 41146
> > >
> > > intel8x0_measure_ac97_clock: measured 49395 usecs
> > > intel8x0: clocking to 41145
> >
> > and now you got the correct output?
> > if so, you can pass the value via ac97_clock option.
> 
> I did retest things more toroughly today after I booted up 2.6.3-rc2. I
> focused on testing KDE login since this is known to break. Startup
> scripts (discover) loaded also i810_audio, I unloaded it.
 
oh never load two these modules at the same time!
it can crash the system.
it must be a bug of i810_audio that can coexist with ALSA driver.

> Plain bootup gave
> PCI: Setting latency timer of device 0000:00:1f.5 to 64
> intel8x0_measure_ac97_clock: measured 49671 usecs
> intel8x0: clocking to 41136
> 
> and it did not work. Unloading and reloading snd-intel8x0 gave
> PCI: Setting latency timer of device 0000:00:1f.5 to 64
> intel8x0_measure_ac97_clock: measured 49281 usecs
> intel8x0: clocking to 41139
> 
> and it still did not work (way too fast, probably even more than twice).
> Now I tried
> modprobe snd-intel8x0 ac97_clock=41140
> It loads silently but still does not work.
> 
> Now I started to suspect that loading i810_audio might screw things up.
> Removed the module and rebooted, no i810_audio loaded, still the same.
> 
> Reproduced the problem with mplayer too. It depends on the input file
> bitrate!
> 
> mplayer /usr/share/sounds/KDE_Startup.wav
> gives the same very fast sound since it's a 22 KHz mono sample.
> 
> mplayer -srate 48000 /usr/share/sounds/KDE_Startup.wav
> works fine. Same applies to all files.

then VRA of ac97 seems not working correctly.
check /proc/asound/card0/codec97#0/* files for each case whether the
DAC rate is set correctly.  (it must be tuned with the detected ac97
clock, i.e. rate * clock / 48000).


--
Takashi Iwai <tiwai@suse.de>		ALSA Developer - www.alsa-project.org

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Meelis Roos]

> > mplayer /usr/share/sounds/KDE_Startup.wav
> > gives the same very fast sound since it's a 22 KHz mono sample.
> >
> > mplayer -srate 48000 /usr/share/sounds/KDE_Startup.wav
> > works fine. Same applies to all files.
>
> then VRA of ac97 seems not working correctly.
> check /proc/asound/card0/codec97#0/* files for each case whether the
> DAC rate is set correctly.  (it must be tuned with the detected ac97
> clock, i.e. rate * clock / 48000).

PCM front DAC    : 41133Hz
PCM ADC          : 48000Hz

This is with "intel8x0: clocking to 41133" and playing with mplayer
-srate 48000 ...

PCM front DAC    : 18895Hz
PCM ADC          : 48000Hz

This is with mplayer with no options, it tells itself

Opening audio decoder: [pcm] Uncompressed PCM audio decoder
AUDIO: 22050 Hz, 1 ch, 16 bit (0x10), ratio: 44100->44100 (352,8 kbit)
Selected audio codec: [pcm] afm:pcm (Uncompressed PCM)
==========================================================================
Checking audio filter chain for 22050Hz/1ch/16bit -> 22050Hz/2ch/16bit...
AF_pre: af format: 2 bps, 1 ch, 22050 hz, little endian signed int
AF_pre: 22050Hz 1ch Signed 16-bit (Little-Endian)
SDL: Samplerate: 22050Hz Channels: Stereo Format Signed 16-bit (Little-Endian)
AO: [sdl] 22050Hz 2ch Signed 16-bit (Little-Endian) (2 bps)
Building audio filter chain for 22050Hz/1ch/16bit -> 22050Hz/2ch/16bit...

-- 
Meelis Roos (mroos@linux.ee)

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Peter Chubb]

>>>>> "Meelis" == Meelis Roos <mroos@linux.ee> writes:

>> > Today mplayer was OK when I tested, didn't retry KDE login. But
>> the > kernel is the same, I have not rebooted inbetween.
>> 
>> weird...  cpufreq is running?


Meelis> mplayer /usr/share/sounds/KDE_Startup.wav gives the same very
Meelis> fast sound since it's a 22 KHz mono sample.

It's a mono/stereo thing.  I've seen the same problem here: the quick
workaround is to convert mono to stereo before playing.

--
Dr Peter Chubb  http://www.gelato.unsw.edu.au  peterc AT gelato.unsw.edu.au
The technical we do immediately,  the political takes *forever*

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Takashi Iwai]

[-- Attachment #1: Type: text/plain, Size: 921 bytes --]

At Tue, 10 Feb 2004 14:03:40 +0200 (EET),
Meelis Roos wrote:
> 
> > > mplayer /usr/share/sounds/KDE_Startup.wav
> > > gives the same very fast sound since it's a 22 KHz mono sample.
> > >
> > > mplayer -srate 48000 /usr/share/sounds/KDE_Startup.wav
> > > works fine. Same applies to all files.
> >
> > then VRA of ac97 seems not working correctly.
> > check /proc/asound/card0/codec97#0/* files for each case whether the
> > DAC rate is set correctly.  (it must be tuned with the detected ac97
> > clock, i.e. rate * clock / 48000).
> 
> PCM front DAC    : 41133Hz
> PCM ADC          : 48000Hz
> 
> This is with "intel8x0: clocking to 41133" and playing with mplayer
> -srate 48000 ...
> 
> PCM front DAC    : 18895Hz
> PCM ADC          : 48000Hz

does the attached patch improve?

if it still doesn't help, try the following:

- stop ALSA once
- remove /etc/asound.state
- restart ALSA and tune up mixer again


Takashi

[-- Attachment #2: Type: text/plain, Size: 400 bytes --]

--- linux/sound/pci/ac97/ac97_patch.c	9 Feb 2004 11:38:03 -0000	1.35
+++ linux/sound/pci/ac97/ac97_patch.c	11 Feb 2004 11:56:19 -0000
@@ -682,6 +682,9 @@
 	jack = snd_ac97_read(ac97, AC97_AD_JACK_SPDIF);
 	snd_ac97_write_cache(ac97, AC97_AD_JACK_SPDIF, jack | 0x0300);
 
+	/* set default */
+	snd_ac97_write_cache(ac97, AC97_AD_MISC, 0);
+
 	ac97->build_ops = &patch_ad1885_build_ops;
 	return 0;
 }

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Meelis Roos]

> does the attached patch improve?

Unfortunately not.

> if it still doesn't help, try the following:
>
> - stop ALSA once
> - remove /etc/asound.state
> - restart ALSA and tune up mixer again

Did that (with the new patched driver), still the same.

-- 
Meelis Roos (mroos@ut.ee)      http://www.cs.ut.ee/~mroos/

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Takashi Iwai]

At Wed, 11 Feb 2004 18:40:36 +0200 (EET),
Meelis Roos wrote:
> 
> > does the attached patch improve?
> 
> Unfortunately not.
> 
> > if it still doesn't help, try the following:
> >
> > - stop ALSA once
> > - remove /etc/asound.state
> > - restart ALSA and tune up mixer again
> 
> Did that (with the new patched driver), still the same.

hmm, please show /proc/asound/card0/codec97#0/ac97#0-0 and
ac97#0-0+regs files again?


Takashi

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Meelis Roos]

> hmm, please show /proc/asound/card0/codec97#0/ac97#0-0 and
> ac97#0-0+regs files again?

0-0/0: Analog Devices AD1885

Capabilities     : -headphone out-
DAC resolution   : 16-bit
ADC resolution   : 16-bit
3D enhancement   : Analog Devices Phat Stereo

Current setup
Mic gain         : +0dB [+0dB]
POP path         : pre 3D
Sim. stereo      : off
3D enhancement   : off
Loudness         : off
Mono output      : MIX
Mic select       : Mic1
ADC/DAC loopback : off
Extended ID      : codec=0 rev=0 DSA=0 VRA
Extended status  : VRA
PCM front DAC    : 18901Hz
PCM ADC          : 48000Hz



AD18XX configuration
Unchained        : 0x1000,0x0000,0x0000
Chained          : 0x0000,0x0000,0x0000

(this is after playing it 22 KHz)

0:00 = 0410
0:02 = bf3f
0:04 = 1010
0:06 = 801f
0:08 = 0000
0:0a = 801e
0:0c = 801f
0:0e = 801f
0:10 = 9f1f
0:12 = 9f1f
0:14 = 9f1f
0:16 = 9f1f
0:18 = 0909
0:1a = 0000
0:1c = 0000
0:1e = 0000
0:20 = 0000
0:22 = 0000
0:24 = 0000
0:26 = 000f
0:28 = 0001
0:2a = 0001
0:2c = 49d5
0:2e = 0000
0:30 = 0000
0:32 = bb80
0:34 = 0000
0:36 = 0000
0:38 = 0000
0:3a = 0000
0:3c = 0000
0:3e = 0000
0:40 = 0000
0:42 = 0000
0:44 = 0000
0:46 = 0000
0:48 = 0000
0:4a = 0000
0:4c = 0000
0:4e = 0000
0:50 = 0000
0:52 = 0000
0:54 = 0000
0:56 = 0000
0:58 = 0000
0:5a = 0000
0:5c = 0000
0:5e = 0000
0:60 = 0000
0:62 = 0000
0:64 = 0000
0:66 = 0000
0:68 = 0000
0:6a = 0000
0:6c = 0000
0:6e = 0000
0:70 = 2000
0:72 = 0300
0:74 = 1000
0:76 = 0000
0:78 = bb80
0:7a = 49d5
0:7c = 4144
0:7e = 5360


And after playing a 48 KHz sound, it changes to
PCM front DAC    : 41147Hz
PCM ADC          : 48000Hz

and regs changes to
0:2c = a0bb
0:7a = 49d5


-- 
Meelis Roos (mroos@linux.ee)

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Takashi Iwai]

[-- Attachment #1: Type: text/plain, Size: 531 bytes --]

At Wed, 11 Feb 2004 18:49:36 +0200 (EET),
Meelis Roos wrote:
> 
> > hmm, please show /proc/asound/card0/codec97#0/ac97#0-0 and
> > ac97#0-0+regs files again?

oops, the patch was wrong.  could you try this one (and remove
/etc/asound.state before starting ALSA?)

also, if it still doesn't work, try to change the ADC VRA.
for example, run "arecord -fS16_LE -r44100 foo.wav" once before
playing 44100 samples.

also, to make sure that it's a hardware problem, please avoid mono
samples but use 16bit stereo samples only.


Takashi

[-- Attachment #2: Type: text/plain, Size: 828 bytes --]

--- linux/sound/pci/ac97/ac97_patch.c	9 Feb 2004 11:38:03 -0000	1.35
+++ linux/sound/pci/ac97/ac97_patch.c	11 Feb 2004 12:08:20 -0000	1.36
@@ -652,7 +652,7 @@
 
 static const snd_kcontrol_new_t snd_ac97_controls_ad1885[] = {
 	AC97_SINGLE("Digital Mono Direct", AC97_AD_MISC, 11, 1, 0),
-	AC97_SINGLE("Digital Audio Mode", AC97_AD_MISC, 12, 1, 0),
+	/* AC97_SINGLE("Digital Audio Mode", AC97_AD_MISC, 12, 1, 0), */ /* seems problematic */
 	AC97_SINGLE("Low Power Mixer", AC97_AD_MISC, 14, 1, 0),
 	AC97_SINGLE("Zero Fill DAC", AC97_AD_MISC, 15, 1, 0),
 };
@@ -682,6 +682,9 @@
 	jack = snd_ac97_read(ac97, AC97_AD_JACK_SPDIF);
 	snd_ac97_write_cache(ac97, AC97_AD_JACK_SPDIF, jack | 0x0300);
 
+	/* set default */
+	snd_ac97_write_cache(ac97, AC97_AD_MISC, 0x0404);
+
 	ac97->build_ops = &patch_ad1885_build_ops;
 	return 0;
 }

Re: 2.6.3-rc1: snd_intel8x0 still too fast

[Meelis Roos]

> oops, the patch was wrong.  could you try this one (and remove
> /etc/asound.state before starting ALSA?)

This one works, thanks!

-- 
Meelis Roos (mroos@linux.ee)