Re: [Alsa-user] Re: Audigy and FPS 5300 rear speakers problem

Re: [Alsa-user] Re: Audigy and FPS 5300 rear speakers problem

[Takashi Iwai]

Hi,

sorry for the delay.

At 12 Jun 2002 23:49:26 +0200,
j_f@laposte.net wrote:
> 
> Hi, 
> See as attached files my freshly generated /etc/asound.state. I also
> attached /etc/init.d/alsasound, I think it comes from redhat and is
> responsible for loading modules and restore state of card (cause it
> loads the famous /etc/asound.state ;).

i tested this config, and it worked well on my system.
i heard very loud output from rear speakers...


> Notice that with this asound.state file I got sound in all speakers when
> reading mp3 files with xmms, but the problem with rear speakers is still
> here: sound is really weak. I also noticed that if I unmute the "Audigy
> Analog/Digital Output Jack" channel I lost sound coming from the center
> speaker (nothing change for rear ones). Also, playing audio cd with this
> config I've only sound from front speakers (nothing on rears or center).
> Playing a mpeg file with xine it seems (but I'm not sure) that there is
> absolutely nothing in the rear speakers.

the output from rear speaker is routed independently from ac97.
thus, for example, "Master" and "PCM" are irrelavant to the output
level of rear/center/lfe speakers.

however, the signals are usually enough high on the level of 60%.
it must not be so weak as you wrote.  i suspect something other
reasons, such like physical connection, etc.

how FPS5300 handles the cables?  are they all analog connections?

btw, cd cannot be played on the current driver from the rear
speakers, since the only pcm output is routed to surround (with the
amount of "Wave Surround").

the digital/analog swithc is one of mysterious things.
i'm not sure what and how it works.  it's not documented.
apparently this toggles the analog and digital outputs, but anyway,
audigy's digital output is still not supported on alsa, so we cannot
test it yet.

in the case of xine, it might use the 4channel pcm device explicitly,
so the routing of "wave surround" has no effect.


> I also noticed that if I launch alsamixer again after an audio
> application (xine or xmms) new item called EMU10K1 PCM, EMU10K1 PCM Send
> and EMU10K1 PCM Send Routing appear, and if I change anyone I lost sound
> from front-right speaker in xine but not in others apps. I've attached
> asound.state before changing an item and after, files are respctively
> called asound.state.xine and asound.state.xine.left

please don't tought these values.  they are actually used for efx
routing, and usually changed via alsa configuration file
(/usr/share/alsa/cards/EMU10k1.conf).

 
ciao,

Takashi


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
No, I will not fix your computer.
http://thinkgeek.com/sf

ALSA hangs under load

[Shaju Abraham]

Hi all,
         We usedALSA driver version( 0.5.12a)  on i810 motherboard, Red Hat
7.1 kernel 2.4.4.

We played a mp3 song using mpg123. when the song was being played, we opened
about 20 to 25 konsole windows and netscape navigators, ALSA stopped playing
the song.  After all windows were opened by linux, we checked the dmesg and
found that pcm_lib.c debug msg ----   DMA or IRQ problem?  ----   being
repeated.  cat /proc/interrupts repeatedly used shows that the number of ICH
interrupts does not increase at all.  Meaning ICH interrupts are no longer
being received at all.   Why does this happen? Is it a bug in ALSA?

--Shaju Abraham



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

Re: ALSA hangs under load

[Takashi Iwai]

At Thu, 18 Jul 2002 12:23:35 +0530,
Shaju Abraham wrote:
> 
> Hi all,
>          We usedALSA driver version( 0.5.12a)  on i810 motherboard, Red Hat
> 7.1 kernel 2.4.4.
> 
> We played a mp3 song using mpg123. when the song was being played, we opened
> about 20 to 25 konsole windows and netscape navigators, ALSA stopped playing
> the song.  After all windows were opened by linux, we checked the dmesg and
> found that pcm_lib.c debug msg ----   DMA or IRQ problem?  ----   being
> repeated.  cat /proc/interrupts repeatedly used shows that the number of ICH
> interrupts does not increase at all.  Meaning ICH interrupts are no longer
> being received at all.   Why does this happen? Is it a bug in ALSA?

it might be...

anyway version 0.5.x is obsolete now.  we won't developt it (although
we accept bug-fix patches, of course).

please try version 0.9.0, preferrably cvs version.


Takashi


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

snd_pcm_mmap_playback_avail & Co. still buggy.

[tomasz motylewski]

I have just downloaded alsa-lib 0.9.0rc2
alsa-lib-0.9.0rc2/src/pcm/pcm_local.h

The code below will return negative value when avail < - pcm->boundary
It may happen if appl.ptr==0 and hw.ptr+pcm->buffer_size goes from 0x7fffffff
to 0x80000000 .

static inline snd_pcm_uframes_t snd_pcm_mmap_playback_avail(snd_pcm_t *pcm)
{
        snd_pcm_sframes_t avail;
        avail = *pcm->hw.ptr + pcm->buffer_size - *pcm->appl.ptr;
        if (avail < 0)
                avail += pcm->boundary;
        else if ((snd_pcm_uframes_t) avail >= pcm->boundary)
                avail -= pcm->boundary;
        return avail;
}

May be the following code would be better?

static inline snd_pcm_uframes_t snd_pcm_mmap_playback_avail(snd_pcm_t *pcm)
{
        snd_pcm_sframes_t avail;
        avail = *pcm->hw.ptr + pcm->buffer_size - *pcm->appl.ptr;
        if (avail < 0)
                avail += pcm->boundary;
	if (avail < 0)
		avail += pcm->boundary;
        else if ((snd_pcm_uframes_t) avail >= pcm->boundary)
                avail -= pcm->boundary;
        return avail;
}

The solution is not as simple as I thought. There has to be a discontinuity
somewhere.

I have the following questions about the design:

1. Is hw.ptr allowed to be set by the alsa-driver >= than
boundary?

2. Is snd_pcm_mmap_playback_avail allowed to return negative number?

What is confusing is that there seems to be a protection against returning
negative values, but it does not work 100 %.

There are more functions nearby which have the same problem.

What I need is just an ((unsigned)hw.ptr)%boundary

Best regards,
--
Tomasz Motylewski



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