Skip to contentSkip to conversation

Requested: CMI8330 testers?

DISCUSSION Takashi Iwai 2 participants 12 messages Dec 16, 2002 – Jan 21, 2003 3 messages not in the archive

lore
Share & tools
lore
#0 lore raw

Hi,

can anyone who owns the CMI8330 ISA test the attached patch? the patch will change the pcm interface of this soundchip as a more reasonable way. the playback and capture are combined into one pcm. it's for the latest cvs version. not sure whether it's applicable to rc6.

please check whether the full-duplex works. it's enough to run aplay and arecord simultaneously.

thanks,

Takashi

#1 lore raw
Quoted 3 lines · Takashi Iwai

Hi,

can anyone who owns the CMI8330 ISA test the attached patch?

I am your man !!! Happy to see that this card is no longer forgotten. 1 month ago, I have written the following to [email protected], but it never got there (I was not a subscriber at this time).

******************************************

Short story: (and main problem)
=========
I have tried to hack the cmi8330 driver version 0.9.0beta10, which does not work, by relying on the (semi) functional last 0.5 driver. It is difficult to appreciate the changes done between 0.5 and 0.9, as the CVS Changelog misses the 0.5 era.
Any hint to fill this hole would be appreciated.

More generally, I am wondering if ALSA is still considering maintenance on old sound chips drivers?
As an alsa-1.0 release is near, this topic seems important to the credibility of the project.

Long story: (plus a lot of other minor problems)
=========
I have access to 4 different Linux machines, with different sound cards. All the machine run Debian Woody, kernel 2.4.18, same config for 4 machines, as far as hardware permits:
AMD K6-II with a SB Vibra 16x PnP
AMD K6-II with a CMI8330
PIII with a intel 810
Athlon with a VIA8233

First, the two K6-II machines run fine the OSS sb driver. When I got the Athlon machines, as via8233 is not supported by OSS drivers, I decided to switch to ALSA 0.9 (woody packages 0.9.0beta12).

The Athlon with snd-via8233 runs fine, no problems with using the OSS compatibility level.
The PIII with snd-intel8x0 runs well, except that xquake (original bin from Id) runs without producing any sound. Moot point, because quakeforge version works well (alsa plugin).

So the problem here seems to be in the oss compatibility level ?

Now the AMD K6-II with a SB Vibra 16X PnP: runs with snd-sb16. In the mixer I must force the 16 bit DMA to be allocated to Capture
to have PCM sound for all kind of audio formats. If set to Auto, for instance .au format is not recognized. Also, xquake runs and produces sound only with option 8bits sound. quakeforge works well, but there is a time lag in producing sound, which does not exist with the OSS driver.
Tuning snd-mixahead does not help.

Finally the AMD K6-II with the CMI8330 and the snd-cmi8330 driver. Version 0.9 does not produce sound at all. But going back to alsa 0.5, I was able to produce PCM sound and play CDs. The mixer tuning is very buggy (this was documented in the alsa-user mail list 2 years ago!), but you can have it working. xquake does not produce sound though. quakeforge cannot be run (they use the 0.9 alsa API).

Another real pain with this driver is the use of
two devices (pcmC0D1p and pcmC0d0c) on the same card. This requires editing several configuration files to have OSS applications work. For instance, I must switch /dev/dsp with /dev/adsp and change in alsa.conf to the following line:
defaults.pcm.device 1

This was corrected in the 0.9 snd-cmi8330 driver, but as it now does not produce sound... I have analyzed the driver code differences between 0.5 and 0.9. The main difference in handling the hardware seems to be that in the 0.5 version, the SB mixer part is initialized, not in the 0.9 one. So I added such an initialization step. I added also the SB mixer controls to the cmi8330 ones, as it is no more possible to add a second mixer device in the 0.9 API, at least I do not see how.
(This addition is not understood
by alsamixer, but amixer with the complex controls works, with the same kind of curious behaviour as amixer-0.5)

And guess what, bingo, I got PCM sound ! But an horrible sound, very distorted, so something is still working badly. This
was just after removing the 0.5 driver and installing the modified 0.9 one. After rebooting the machine with the 0.9 running, no more sound!

So there is obviously something which has not been initialized well in the 0.9 driver alone. I have also tried to suppress completely any handling of the SB part (mixer + PCM), or only resetting it, to no avail so far.

As sb16 and cmi8330 are now old chips, I am wondering if there is today any action on the corresponding drivers more than "edit the driver file source to make it compile if there is an API change".

***************************************

Meanwhile, I have started to read your document "Writing an Alsa driver". This explains a lot of things, and fills a huge hole in the Alsa project, thanks !

Quoted 7 lines

the patch will change the pcm interface of this soundchip as a more reasonable way. the playback and capture are combined into one pcm. it's for the latest cvs version. not sure whether it's applicable to rc6.

please check whether the full-duplex works. it's enough to run aplay and arecord simultaneously.

I hope to be able to test your patch for next Monday.

Regards.

Signature · 7 lines
Bernard Urban


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
#2 lore raw

Hi Bernard,

thanks for your reply.

At 27 Dec 2002 10:19:10 +0100,

Quoted 22 lines · Bernard Urban
Quoted 3 lines · Takashi Iwai

Hi,

can anyone who owns the CMI8330 ISA test the attached patch?

I am your man !!! Happy to see that this card is no longer forgotten. 1 month ago, I have written the following to [email protected], but it never got there (I was not a subscriber at this time).

******************************************

Short story: (and main problem)
=========
I have tried to hack the cmi8330 driver version 0.9.0beta10, which does not work, by relying on the (semi) functional last 0.5 driver. It is difficult to appreciate the changes done between 0.5 and 0.9, as the CVS Changelog misses the 0.5 era.
Any hint to fill this hole would be appreciated.

Quoted 2 lines

More generally, I am wondering if ALSA is still considering maintenance on old sound chips drivers?

we can maintain the code but none of us have these hardwares for test...

Quoted 17 lines

As an alsa-1.0 release is near, this topic seems important to the credibility of the project.

Long story: (plus a lot of other minor problems)
=========
I have access to 4 different Linux machines, with different sound cards. All the machine run Debian Woody, kernel 2.4.18, same config for 4 machines, as far as hardware permits:
AMD K6-II with a SB Vibra 16x PnP
AMD K6-II with a CMI8330
PIII with a intel 810
Athlon with a VIA8233

First, the two K6-II machines run fine the OSS sb driver. When I got the Athlon machines, as via8233 is not supported by OSS drivers, I decided to switch to ALSA 0.9 (woody packages 0.9.0beta12).

Quoted 7 lines

The Athlon with snd-via8233 runs fine, no problems with using the OSS compatibility level.
The PIII with snd-intel8x0 runs well, except that xquake (original bin from Id) runs without producing any sound. Moot point, because quakeforge version works well (alsa plugin).

So the problem here seems to be in the oss compatibility level ?

most likely the problem is due to mmap on OSS emulation. the OSS emulation should be "direct" mode in some cases. you can change the behavior via writing to oss proc file like

echo "xquake 0 0 direct" > /proc/asound/card0/pcm0p/oss

please check alsa-kernel/Documentation/ALSA-Configuration.txt for details.

Quoted 5 lines

Now the AMD K6-II with a SB Vibra 16X PnP: runs with snd-sb16. In the mixer I must force the 16 bit DMA to be allocated to Capture
to have PCM sound for all kind of audio formats. If set to Auto, for instance .au format is not recognized. Also, xquake runs and produces sound only with option 8bits sound.

perhaps this is also related with the above.

Quoted 3 lines

quakeforge works well, but there is a time lag in producing sound, which does not exist with the OSS driver. Tuning snd-mixahead does not help.

this is not sure. does the delay happen only on quake or do you mean it generally? if it's general behavior, then it's easier to debug. anyway we need to trace at which point the delay occurs.

Quoted 14 lines

Finally the AMD K6-II with the CMI8330 and the snd-cmi8330 driver. Version 0.9 does not produce sound at all. But going back to alsa 0.5, I was able to produce PCM sound and play CDs. The mixer tuning is very buggy (this was documented in the alsa-user mail list 2 years ago!), but you can have it working. xquake does not produce sound though. quakeforge cannot be run (they use the 0.9 alsa API).

Another real pain with this driver is the use of
two devices (pcmC0D1p and pcmC0d0c) on the same card. This requires editing several configuration files to have OSS applications work. For instance, I must switch /dev/dsp with /dev/adsp and change in alsa.conf to the following line:
defaults.pcm.device 1

Quoted 10 lines

This was corrected in the 0.9 snd-cmi8330 driver, but as it now does not produce sound... I have analyzed the driver code differences between 0.5 and 0.9. The main difference in handling the hardware seems to be that in the 0.5 version, the SB mixer part is initialized, not in the 0.9 one. So I added such an initialization step. I added also the SB mixer controls to the cmi8330 ones, as it is no more possible to add a second mixer device in the 0.9 API, at least I do not see how.
(This addition is not understood
by alsamixer, but amixer with the complex controls works, with the same kind of curious behaviour as amixer-0.5)

well, the mixers are all mixed up on alsa 0.9. it's possible to assign the ad1848 mixer elements with the index 1, though.

Quoted 4 lines

And guess what, bingo, I got PCM sound ! But an horrible sound, very distorted, so something is still working badly. This
was just after removing the 0.5 driver and installing the modified 0.9 one. After rebooting the machine with the 0.9 running, no more sound!

ok, it explains the hardware init problem.
i'll check the difference between 0.5 and 0.9.

btw, my last patch was already committed to cvs.

ciao,

Takashi

-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

#3 lore raw

Hi, it's me again.

At Wed, 08 Jan 2003 14:00:03 +0100,

Quoted 8 lines · I wrote:
Quoted 5 lines

And guess what, bingo, I got PCM sound ! But an horrible sound, very distorted, so something is still working badly. This
was just after removing the 0.5 driver and installing the modified 0.9 one. After rebooting the machine with the 0.9 running, no more sound!

ok, it explains the hardware init problem.
i'll check the difference between 0.5 and 0.9.

please try the attached patch.
it will add the SB mixers (all new ones starting "SB xxx" except for tone controls) and initializes the SB-mixer reg bit (as default).

the first step is to check whether and which SB mixer has influence to what output/input. there are also CMI8330-native controls, namely the controls without "SB" prefix. whether these controls affect the playback. the SB mixers can be disabled (removed) by commenting out ENABLE_SB_MIXER definition in the source.

the next step is to switch playback and capture by commenting out the definition PLAYBACK_ON_SB at the beginning of the code. then the playback will be done through AD1848 and capture will be on SB16. check again which mixer elements work for what.

ciao,

Takashi

#4 lore raw

Hi,

At Wed, 08 Jan 2003 17:17:17 +0100,

Quoted 4 lines · I wrote:

please try the attached patch.
it will add the SB mixers (all new ones starting "SB xxx" except for tone controls) and initializes the SB-mixer reg bit (as default).

i found the last patch including bugs.
now the new version was already applied to cvs, so just update the cvs tree.

Quoted 11 lines

the first step is to check whether and which SB mixer has influence to what output/input. there are also CMI8330-native controls, namely the controls without "SB" prefix. whether these controls affect the playback. the SB mixers can be disabled (removed) by commenting out ENABLE_SB_MIXER definition in the source.

the next step is to switch playback and capture by commenting out the definition PLAYBACK_ON_SB at the beginning of the code. then the playback will be done through AD1848 and capture will be on SB16. check again which mixer elements work for what.

and please check the above again...

ciao,

Takashi

-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

#8 lore raw

Re: CMI8330: success !

Hello!

I will try to summarize my tests of the current (13 January) CVS tree for this driver:

For playback, the following /etc/asound.state file works:

state.card0 {
	control.1 {
		comment.access 'read write'
		comment.type INTEGER
		comment.range '0 - 15'
		iface MIXER
		name 'Master Playback Volume'
		value.0 15
		value.1 15
	}
	control.2 {
		comment.access 'read write'
		comment.type BOOLEAN
		iface MIXER
		name 'Loud Playback Switch'
		value true
	}
	control.3 {
		comment.access 'read write'
		comment.type BOOLEAN
		iface MIXER
		name 'PCM Playback Switch'
		value.0 true
		value.1 true
	}
	control.4 {
		comment.access 'read write'
		comment.type INTEGER
		comment.range '0 - 63'
		iface MIXER
		name 'PCM Playback Volume'
		value.0 63
		value.1 63
	}
	control.21 {
		comment.access 'read write'
		comment.type BOOLEAN
		iface MIXER
		name '3D Control - Switch'
		value true
	}
	control.22 {
		comment.access 'read write'
		comment.type INTEGER
		comment.range '0 - 3'
		iface MIXER
		name 'PC Speaker Playback Volume'
		value 0
	}
	control.23 {
		comment.access 'read write'
		comment.type BOOLEAN
		iface MIXER
		name 'FM Playback Switch'
		value true
	}
	control.24 {
		comment.access 'read write'
		comment.type BOOLEAN
		iface MIXER
		name 'IEC958 Input Capture Switch'
		value true
	}
	control.25 {
		comment.access 'read write'
		comment.type BOOLEAN
		iface MIXER
		name 'IEC958 Input Playback Switch'
		value true
	}
}
state.CARD_0 {
	control.1 {
		comment.access 'read write'
		comment.type INTEGER
		comment.range '0 - 15'
		iface MIXER
		name 'Master Playback Volume'
		value.0 15
		value.1 15
	}
	control.2 {
		comment.access 'read write'
		comment.type BOOLEAN
		iface MIXER
		name 'Loud Playback Switch'
		value true
	}
	control.3 {
		comment.access 'read write'
		comment.type BOOLEAN
		iface MIXER
		name 'PCM Playback Switch'
		value.0 true
		value.1 true
	}
	control.4 {
		comment.access 'read write'
		comment.type INTEGER
		comment.range '0 - 63'
		iface MIXER
		name 'PCM Playback Volume'
		value.0 61
		value.1 61
	}
	control.9 {
		comment.access 'read write'
		comment.type BOOLEAN
		iface MIXER
		name 'CD Playback Switch'
		value.0 true
		value.1 true
	}
	control.11 {
		comment.access 'read write'
		comment.type INTEGER
		comment.range '0 - 15'
		iface MIXER
		name 'CD Playback Volume'
		value.0 12
		value.1 12
	}
	control.21 {
		comment.access 'read write'
		comment.type BOOLEAN
		iface MIXER
		name '3D Control - Switch'
		value true
	}
	control.23 {
		comment.access 'read write'
		comment.type BOOLEAN
		iface MIXER
		name 'FM Playback Switch'
		value true
	}
}

Notice that I have deleted the controls with false or zero values to save some place. For card0, the SB controls are absent. There are also some differences between controls for card0 and CARD_0, why ? I get the CARD_0 controls when using alsamixer.
The important thing to notice is that the SB
controls need not to be unmuted (as I first believe was necessary) ! So to have playback it is not needed
to define these SB controls in routine cmi8330_add_sb_mixers(). But the SB initialization step is probably mandatory (I will test it).

Actually, this is somehow curious, as PLAYBACK_ON_SB is defined, and so playback should go through SB ??? It is apparently not the case.

Only one application did not work correctly, namely the old libc5 based xquake (yes, I have done
echo "xquake.real 0 0 direct" >/proc/asound/card0/pcm0p/oss). There is sound, but extremely noisy.

Some words about the behaviour of the SB controls in alsamixer:
"SB" control follows the "SB Capture" control value. "SB Synth", "SB PCM" and "SB Speaker" do not respond to commands. "SB CD", "SB Line" and "SB Mic" cannot have values above 10%.

I have also tried both "SB Mic" and "Mic" controls in alsamixer, but capture did not work at all (only noise is registered and I have Larsen effect when the volume is pushed too high). Nevertheless, as "SB Mic" cannot have values above 10%, here lies perhaps the problem.

I will now try to use #undefine PLAYBACK_ON_SB.

Concerning the SB16 driver, I could not record either (arecord is blocking indefinitly). Changing the 16 bits DMA control did not help.

Regards.

Signature · 8 lines
Bernard Urban


-------------------------------------------------------
This SF.NET email is sponsored by: Thawte.com
Understand how to protect your customers personal information by implementing
SSL on your Apache Web Server. Click here to get our FREE Thawte Apache 
Guide: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0029en
#9 lore raw

Re: CMI8330: success !

Quoted 10 lines · Bernard Urban

Notice that I have deleted the controls with false or zero values to save some place. For card0, the SB controls are absent. There are also some differences between controls for card0 and CARD_0, why ? I get the CARD_0 controls when using alsamixer.
The important thing to notice is that the SB
controls need not to be unmuted (as I first believe was necessary) ! So to have playback it is not needed
to define these SB controls in routine cmi8330_add_sb_mixers(). But the SB initialization step is probably mandatory (I will test it).

Actually, the cmi8330 card works when undefining ENABLE_SB_MIXER. So the init step in snd_sbdsp_create() is sufficient. This is good, as you can get rid of these additional and dysfunctional controls.

Quoted 14 lines

Actually, this is somehow curious, as PLAYBACK_ON_SB is defined, and so playback should go through SB ??? It is apparently not the case.

Only one application did not work correctly, namely the old libc5 based xquake (yes, I have done
echo "xquake.real 0 0 direct" >/proc/asound/card0/pcm0p/oss). There is sound, but extremely noisy.

Some words about the behaviour of the SB controls in alsamixer:
"SB" control follows the "SB Capture" control value. "SB Synth", "SB PCM" and "SB Speaker" do not respond to commands. "SB CD", "SB Line" and "SB Mic" cannot have values above 10%.

Quoted 7 lines

I have also tried both "SB Mic" and "Mic" controls in alsamixer, but capture did not work at all (only noise is registered and I have Larsen effect when the volume is pushed too high). Nevertheless, as "SB Mic" cannot have values above 10%, here lies perhaps the problem.

I will now try to use #undefine PLAYBACK_ON_SB.

This works also, with ENABLE_SB_MIXER defined, but some .wav files played on events in the desktop are scratchy (sometimes, not always). So the original #define PLAYBACK_ON_SB
is better.

Quoted 3 lines

Concerning the SB16 driver, I could not record either (arecord is blocking indefinitly). Changing the 16 bits DMA control did not help.

No, this works. The blocking occurs when some application has an open file descriptor for playback (in my case, esd for gnome, even in the state "suspended"). But you must MANUALLY assign 16 bits DMA to "Playback" to obtain a successful capture (and conversely assign it to "Capture" for playback). The "Auto" value does not work. I recall my card is a SB 16 Vibra X PNP, if you want to put a warning in the README file about that.

Capture for cmi8330: no success at all, with or without PLAYBACK_ON_SB, and ENABLE_SB_MIXER defined.
But I noticed that I cannot capture (mic recording) with DOS or Windows either on this machine. The microphone is ok, it is the same used for the successful SB16 test. So it is not out of possibility there is a hardware problem on my mainboard (the chip is on it, not on an add-on card). I have not currently the possibility to test the "line in" entry. I will perhaps test the CD one if it makes sense ?

Regards.

Signature · 8 lines
Bernard Urban


-------------------------------------------------------
This SF.NET email is sponsored by: FREE  SSL Guide from Thawte
are you planning your Web Server Security? Click here to get a FREE
Thawte SSL guide and find the answers to all your  SSL security issues.
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0026en
#10 lore raw

Re: CMI8330: success !

At 20 Jan 2003 10:43:28 +0100,

Quoted 14 lines · Bernard Urban
Quoted 10 lines · Bernard Urban

Notice that I have deleted the controls with false or zero values to save some place. For card0, the SB controls are absent. There are also some differences between controls for card0 and CARD_0, why ? I get the CARD_0 controls when using alsamixer.
The important thing to notice is that the SB
controls need not to be unmuted (as I first believe was necessary) ! So to have playback it is not needed
to define these SB controls in routine cmi8330_add_sb_mixers(). But the SB initialization step is probably mandatory (I will test it).

Quoted 3 lines

Actually, the cmi8330 card works when undefining ENABLE_SB_MIXER. So the init step in snd_sbdsp_create() is sufficient. This is good, as you can get rid of these additional and dysfunctional controls.

thanks, nice to know that.
does it work w/o ENABLE_SB_MXIER even after rebooting (i mean, after resetting the hardware) ?

Quoted 20 lines
Quoted 20 lines

Actually, this is somehow curious, as PLAYBACK_ON_SB is defined, and so playback should go through SB ??? It is apparently not the case.

Only one application did not work correctly, namely the old libc5 based xquake (yes, I have done
echo "xquake.real 0 0 direct" >/proc/asound/card0/pcm0p/oss). There is sound, but extremely noisy.

Some words about the behaviour of the SB controls in alsamixer:
"SB" control follows the "SB Capture" control value. "SB Synth", "SB PCM" and "SB Speaker" do not respond to commands. "SB CD", "SB Line" and "SB Mic" cannot have values above 10%.

I have also tried both "SB Mic" and "Mic" controls in alsamixer, but capture did not work at all (only noise is registered and I have Larsen effect when the volume is pushed too high). Nevertheless, as "SB Mic" cannot have values above 10%, here lies perhaps the problem.

Quoted 7 lines
Quoted 1 line

I will now try to use #undefine PLAYBACK_ON_SB.

This works also, with ENABLE_SB_MIXER defined, but some .wav files played on events in the desktop are scratchy (sometimes, not always). So the original #define PLAYBACK_ON_SB
is better.

ok.

Quoted 11 lines
Quoted 3 lines

Concerning the SB16 driver, I could not record either (arecord is blocking indefinitly). Changing the 16 bits DMA control did not help.

No, this works. The blocking occurs when some application has an open file descriptor for playback (in my case, esd for gnome, even in the state "suspended"). But you must MANUALLY assign 16 bits DMA to "Playback" to obtain a successful capture (and conversely assign it to "Capture" for playback). The "Auto" value does not work. I recall my card is a SB 16 Vibra X PNP, if you want to put a warning in the README file about that.

then we can set "playback" as default for Vibra X PnP.

i'm not sure whether it's a general problem for all SB16 Vibra
models, or Vibra X only. does anyone know that?

Quoted 8 lines

Capture for cmi8330: no success at all, with or without PLAYBACK_ON_SB, and ENABLE_SB_MIXER defined.
But I noticed that I cannot capture (mic recording) with DOS or Windows either on this machine. The microphone is ok, it is the same used for the successful SB16 test. So it is not out of possibility there is a hardware problem on my mainboard (the chip is on it, not on an add-on card). I have not currently the possibility to test the "line in" entry. I will perhaps test the CD one if it makes sense ?

yes, but make sure that you connected the CD-cable :)

ciao,

Takashi

-------------------------------------------------------
This SF.NET email is sponsored by: FREE SSL Guide from Thawte are you planning your Web Server Security? Click here to get a FREE Thawte SSL guide and find the answers to all your SSL security issues.
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0026en

#11 lore raw

Re: CMI8330: success !

Takashi Iwai <[email protected]> writes:

[...]

Quoted 4 lines

thanks, nice to know that.
does it work w/o ENABLE_SB_MXIER even after rebooting (i mean, after resetting the hardware) ?

Yes, even after power down/power up (and it is not an ATX power supply, so it is really down when you switch off).

[...]

Quoted 10 lines
Quoted 9 lines

Capture for cmi8330: no success at all, with or without PLAYBACK_ON_SB, and ENABLE_SB_MIXER defined.
But I noticed that I cannot capture (mic recording) with DOS or Windows either on this machine. The microphone is ok, it is the same used for the successful SB16 test. So it is not out of possibility there is a hardware problem on my mainboard (the chip is on it, not on an add-on card). I have not currently the possibility to test the "line in" entry. I will perhaps test the CD one if it makes sense ?

Quoted 2 lines

yes, but make sure that you connected the CD-cable :)

Good news, capture works with CD !

Only one gotcha (probably not related to the cmi8330 driver):
there are "Line Capture Volume", "Mic Capture Volume" and "CD Capture Volume" controls. The first 2 appear in alsamixer, but not the CD one ?
So you must raise the CD capture volume with amixer.

Sincerely.

Signature · 7 lines
Bernard Urban


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
#12 lore raw

Re: CMI8330: success !

Hi,

At 21 Jan 2003 11:11:55 +0100,

Quoted 13 lines · Bernard Urban
Quoted 10 lines
Quoted 9 lines

Capture for cmi8330: no success at all, with or without PLAYBACK_ON_SB, and ENABLE_SB_MIXER defined.
But I noticed that I cannot capture (mic recording) with DOS or Windows either on this machine. The microphone is ok, it is the same used for the successful SB16 test. So it is not out of possibility there is a hardware problem on my mainboard (the chip is on it, not on an add-on card). I have not currently the possibility to test the "line in" entry. I will perhaps test the CD one if it makes sense ?

Quoted 2 lines

yes, but make sure that you connected the CD-cable :)

Quoted 7 lines

Good news, capture works with CD !

Only one gotcha (probably not related to the cmi8330 driver):
there are "Line Capture Volume", "Mic Capture Volume" and "CD Capture Volume" controls. The first 2 appear in alsamixer, but not the CD one ?
So you must raise the CD capture volume with amixer.

it's a typo in the driver.
please try the attached patch.

Takashi

#13 lore raw
In reply to Takashi Iwai

Takashi Iwai <[email protected]> writes:

[...]

Quoted 6 lines

echo "xquake 0 0 direct" > /proc/asound/card0/pcm0p/oss

please check alsa-kernel/Documentation/ALSA-Configuration.txt for details.

Quoted 7 lines
Quoted 5 lines

Now the AMD K6-II with a SB Vibra 16X PnP: runs with snd-sb16. In the mixer I must force the 16 bit DMA to be allocated to Capture
to have PCM sound for all kind of audio formats. If set to Auto, for instance .au format is not recognized. Also, xquake runs and produces sound only with option 8bits sound.

perhaps this is also related with the above.

Yes, this works ! No more flag for 8 bits sound required. I missed it the first time, as what you need actually is:
echo "xquake.real 0 0 direct" > /proc/asound/card0/pcm0p/oss

(xquake is a shell wrapper for the real binary xquake.real)

Quoted 7 lines
Quoted 3 lines

quakeforge works well, but there is a time lag in producing sound, which does not exist with the OSS driver. Tuning snd-mixahead does not help.

this is not sure. does the delay happen only on quake or do you mean it generally? if it's general behavior, then it's easier to debug. anyway we need to trace at which point the delay occurs.

It is not general behaviour, alas. More specifically:
quakeforge with alsa sound output and alsa driver has the time lag (nearly 0.5 second).
quakeforge with OSS sound output and OSS 'sb' driver works well. quakeforge with OSS sound output and alsa OSS emulation driver works well, using:

echo "nq-glx 0 0 direct" > /proc/asound/card0/pcm0p/oss

but when you quit, the program locks itself (no machine crash, only the program is blocked, you need to kill it).

Using quake2forge (a quake2 port, without alsa support) under alsa OSS emulation driver with:

echo "quake2 0 0 direct" > /proc/asound/card0/pcm0p/oss

it works also well, but same program lock when quitting as with quakeforge. I will try to investigate where it blocks (last message is about DMA buffer release).

Regards.

Signature · 7 lines
Bernard Urban


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
#14 lore raw

Re: SB 16 driver OSS emulation

This gives additional details about the blocking with the sb16 driver:

Bernard Urban <[email protected]> writes:

[...]

Quoted 11 lines

quakeforge with alsa sound output and alsa driver has the time lag (nearly 0.5 second).
quakeforge with OSS sound output and OSS 'sb' driver works well. quakeforge with OSS sound output and alsa OSS emulation driver works well, using:

echo "nq-glx 0 0 direct" > /proc/asound/card0/pcm0p/oss

but when you quit, the program locks itself (no machine crash, only the program is blocked, you need to kill it).

Quoted 8 lines

Using quake2forge (a quake2 port, without alsa support) under alsa OSS emulation driver with:

echo "quake2 0 0 direct" > /proc/asound/card0/pcm0p/oss

it works also well, but same program lock when quitting as with quakeforge. I will try to investigate where it blocks (last message is about DMA buffer release).

quake2forge obtains an 'audio_fd' file descriptor by opening "/dev/dsp". There is a DMA buffer mmaped associated.

At shutdown of the program, the DMA buffer is un-mmaped, and the following close(audio_fd) blocks. This seems really weird !!! A 'close' system call cannot block ???

Actually, doing or not
echo "quake2 0 0 direct" > /proc/asound/card0/pcm0p/oss makes no difference.

(By Monday, I hope I can send you the results of the CMI8330 patches.)

Regards.

Signature · 7 lines
Bernard Urban


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com