Skip to contentSkip to conversation

Problems with snd_pcm_hw_params_can_pause (). etc.

DISCUSSION James Courtier-Dutton 2 participants 4 messages Oct 12, 2002

lore
Share & tools
lore
#0 lore raw

Currently, the state of play is that "snd_pcm_hw_params_can_pause ()" should not be called until one has first done the first "snd_pcm_hw_params()"

I start with
snd_pcm_hw_params_any(this->audio_fd, params);
then go about setting params, e.g.
snd_pcm_hw_params_set_access()
snd_pcm_hw_params_set_format()
etc.
Then
snd_pcm_hw_params()
then
snd_pcm_hw_params_can_pause (params)

If the "snd_pcm_hw_params_can_pause (params)" goes before the "snd_pcm_hw_params()" an invalid result is returned. I.E. the result is always 1 even for cards which don't support it.

I think the documentation should make it clear when one should use "snd_pcm_hw_params_can_pause ()".
or maybe "snd_pcm_hw_params_any()" should be modified to correctly fill the "can_pause" bit.
Maybe I have missunderstood the API, but I would prefer the approach of first retrieving the current hw params, then go about modifying it, then writting it back. If one has not set any hw params yet, it would fill the structure with the most general case, then allow the application to set params to further restrict the config.
I had assumed that "snd_pcm_hw_params_any()" did the retrieving of the current hw params, but reset to the least restrictive settings. I don't think "can_pause" is anything an application can set, the hardware on the card decides that, so the "snd_pcm_hw_params_any()" should cause "can_pause" to return a valid setting that matches the hardware, instead of having to wait for the params to be written to the card with "snd_pcm_hw_params()" before "can_pause" is set accurately.

Cheers
James

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

#1 lore raw
Quoted 19 lines · James Courtier-Dutton

Currently, the state of play is that "snd_pcm_hw_params_can_pause ()" should not be called until one has first done the first "snd_pcm_hw_params()"

I start with
snd_pcm_hw_params_any(this->audio_fd, params);
then go about setting params, e.g.
snd_pcm_hw_params_set_access()
snd_pcm_hw_params_set_format()
etc.
Then
snd_pcm_hw_params()
then
snd_pcm_hw_params_can_pause (params)

If the "snd_pcm_hw_params_can_pause (params)" goes before the "snd_pcm_hw_params()" an invalid result is returned. I.E. the result is always 1 even for cards which don't support it.

Quoted 7 lines

I think the documentation should make it clear when one should use "snd_pcm_hw_params_can_pause ()".
or maybe "snd_pcm_hw_params_any()" should be modified to correctly fill the "can_pause" bit.
Maybe I have missunderstood the API, but I would prefer the approach of first retrieving the current hw params, then go about modifying it, then writting it back. If one has not set any hw params yet, it would fill

But what's "current configuration"? What you want to prefer? rate? channels? period_size? No, application should give these hints to driver.

Quoted 9 lines

the structure with the most general case, then allow the application to set params to further restrict the config.
I had assumed that "snd_pcm_hw_params_any()" did the retrieving of the current hw params, but reset to the least restrictive settings. I don't think "can_pause" is anything an application can set, the hardware on the card decides that, so the "snd_pcm_hw_params_any()" should cause "can_pause" to return a valid setting that matches the hardware, instead of having to wait for the params to be written to the card with "snd_pcm_hw_params()" before "can_pause" is set accurately.

All these functions needs exactly one configuration at input. The snd_pcm_hw_params() function chooses one configuration and sets it to hardware. I've updated documentation.

Jaroslav

-----
Jaroslav Kysela <[email protected]>
Linux Kernel Sound Maintainer
ALSA Project http://www.alsa-project.org
SuSE Linux http://www.suse.com

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

#2 lore raw
Quoted 34 lines · Jaroslav Kysela
Quoted 13 lines · James Courtier-Dutton

Currently, the state of play is that "snd_pcm_hw_params_can_pause ()" should not be called until one has first done the first "snd_pcm_hw_params()"

I start with
snd_pcm_hw_params_any(this->audio_fd, params);
then go about setting params, e.g.
snd_pcm_hw_params_set_access()
snd_pcm_hw_params_set_format()
etc.
Then
snd_pcm_hw_params()
then
snd_pcm_hw_params_can_pause (params)

Quoted 7 lines

If the "snd_pcm_hw_params_can_pause (params)" goes before the "snd_pcm_hw_params()" an invalid result is returned. I.E. the result is always 1 even for cards which don't support it.

I think the documentation should make it clear when one should use "snd_pcm_hw_params_can_pause ()".
or maybe "snd_pcm_hw_params_any()" should be modified to correctly fill the "can_pause" bit.
Maybe I have missunderstood the API, but I would prefer the approach of first retrieving the current hw params, then go about modifying it, then writting it back. If one has not set any hw params yet, it would fill

But what's "current configuration"? What you want to prefer? rate? channels? period_size? No, application should give these hints to driver.

Quoted 4 lines

the structure with the most general case, then allow the application to set params to further restrict the config.
I had assumed that "snd_pcm_hw_params_any()" did the retrieving of the current hw params, but reset to the least restrictive settings. I don't think "can_pause" is anything an application can set, the hardware on the card decides that, so the "snd_pcm_hw_params_any()" should cause "can_pause" to return a valid setting that matches the hardware, instead of having to wait for the params to be written to the card with "snd_pcm_hw_params()" before "can_pause" is set accurately.

Quoted 7 lines

All these functions needs exactly one configuration at input.
The snd_pcm_hw_params() function chooses one configuration and sets it to hardware. I've updated documentation.

Jaroslav

Thanks for the documentation update. I currently use the inline docs contained in alsa-lib/pcm/pcm.c. Is that the ones you update ? I only mentioned it because I made the mistake of putting the can_pause() before the snd_pcm_hw_params() without knowing that it would not work correctly, so I just wanted the documentation to be updated so that no one else would make the same mistake. I think the point I was trying to make is that one should be able to call
snd_pcm_hw_can_pause() at any time. One should not have to wait until one has snd_pcm_hw_params() has been called. At this point I am assuming that the result of "snd_pcm_hw_can_pause()" is always constant, and does not depend of any hw_params that have been set. But now I come to think of it, this might not be true. I think the SB Live can do pause/resume in normal mode, but cannot do it in AC3 passthru mode using the DSP TRAM.

But as a general point, this does not really matter too much to me now, because I will be using snd_pcm_hw_params_current() that you have kindly added to the API. ;-) I will have to wait until the next release of alsa09 before I actually code the change, because I cannot expect all my users to install alsa09 from CVS. :-(

As a sub note, maybe snd_pcm_can_pause() should return some sort of error code, or throw an assert if the application programmer has put it before snd_pcm_hw_params() instead of returning a perfectly useable result that is baseless. I think an assert() would be acceptable, as this is a application programming error and not a user initiated error. It would then force the application programmer to "get it right first time". ;-)

Cheers
James

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

#3 lore raw
Quoted 82 lines · James Courtier-Dutton
Quoted 50 lines · Jaroslav Kysela
Quoted 19 lines · James Courtier-Dutton

Currently, the state of play is that "snd_pcm_hw_params_can_pause ()" should not be called until one has first done the first "snd_pcm_hw_params()"

I start with
snd_pcm_hw_params_any(this->audio_fd, params);
then go about setting params, e.g.
snd_pcm_hw_params_set_access()
snd_pcm_hw_params_set_format()
etc.
Then
snd_pcm_hw_params()
then
snd_pcm_hw_params_can_pause (params)

If the "snd_pcm_hw_params_can_pause (params)" goes before the "snd_pcm_hw_params()" an invalid result is returned. I.E. the result is always 1 even for cards which don't support it.

Quoted 9 lines

I think the documentation should make it clear when one should use "snd_pcm_hw_params_can_pause ()".
or maybe "snd_pcm_hw_params_any()" should be modified to correctly fill the "can_pause" bit.
Maybe I have missunderstood the API, but I would prefer the approach of first retrieving the current hw params, then go about modifying it, then writting it back. If one has not set any hw params yet, it would fill

But what's "current configuration"? What you want to prefer? rate? channels? period_size? No, application should give these hints to driver.

Quoted 11 lines

the structure with the most general case, then allow the application to set params to further restrict the config.
I had assumed that "snd_pcm_hw_params_any()" did the retrieving of the current hw params, but reset to the least restrictive settings. I don't think "can_pause" is anything an application can set, the hardware on the card decides that, so the "snd_pcm_hw_params_any()" should cause "can_pause" to return a valid setting that matches the hardware, instead of having to wait for the params to be written to the card with "snd_pcm_hw_params()" before "can_pause" is set accurately.

Quoted 8 lines

All these functions needs exactly one configuration at input. The snd_pcm_hw_params() function chooses one configuration and sets it to hardware. I've updated documentation.

Jaroslav

Thanks for the documentation update. I currently use the inline docs contained in alsa-lib/pcm/pcm.c. Is that the ones you update ? I only mentioned it because I made the mistake of putting the can_pause() before the snd_pcm_hw_params() without knowing that it would not work correctly, so I just wanted the documentation to be updated so that no one else would make the same mistake.
I think the point I was trying to make is that one should be able to call snd_pcm_hw_can_pause() at any time. One should not have to wait until one has snd_pcm_hw_params() has been called. At this point I am assuming that the result of "snd_pcm_hw_can_pause()" is always constant, and does not depend of any hw_params that have been set. But now I come to think of it, this might not be true. I think the SB Live can do pause/resume in normal mode, but cannot do it in AC3 passthru mode using the DSP TRAM.

But as a general point, this does not really matter too much to me now, because I will be using snd_pcm_hw_params_current() that you have kindly added to the API. ;-)
I will have to wait until the next release of alsa09 before I actually code the change, because I cannot expect all my users to install alsa09 from CVS. :-(

Quoted 7 lines

As a sub note, maybe snd_pcm_can_pause() should return some sort of error code, or throw an assert if the application programmer has put it before snd_pcm_hw_params() instead of returning a perfectly useable result that is baseless. I think an assert() would be acceptable, as this is a application programming error and not a user initiated error. It would then force the application programmer to "get it right first time". ;-)

Updated. I hope that it will work.

Jaroslav

-----
Jaroslav Kysela <[email protected]>
Linux Kernel Sound Maintainer
ALSA Project http://www.alsa-project.org
SuSE Linux http://www.suse.com

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