Bug in pcm_lib.c ?? (using Xmms)

Bug in pcm_lib.c ?? (using Xmms)

[Benny Sjostrand]

Hi!

Just updated CVS tree and recompiled. Everything works OK execpt that I 
got this message in log
when using Xmms with ALSA-plugin and OSS-plugin. Using aplay the message 
dont apear.

Dec  8 13:40:16 cactus kernel: ALSA ../alsa-kernel/core/pcm_lib.c:176: 
BUG: stream = 0, pos = 0x4000, buffer size = 0x4000, period size = 0x200
Dec  8 13:40:16 cactus kernel: ALSA ../alsa-kernel/core/pcm_lib.c:176: 
BUG: stream = 0, pos = 0x4000, buffer size = 0x4000, period size = 0x200
Dec  8 13:40:16 cactus kernel: ALSA ../alsa-kernel/core/pcm_lib.c:110: 
BUG: stream = 0, pos = 0x4000, buffer size = 0x4000, period size = 0x200
Dec  8 13:40:16 cactus kernel: ALSA ../alsa-kernel/core/pcm_lib.c:176: 
BUG: stream = 0, pos = 0x4000, buffer size = 0x4000, period size = 0x200
Dec  8 13:40:16 cactus kernel: ALSA ../alsa-kernel/core/pcm_lib.c:176: 
BUG: stream = 0, pos = 0x4000, buffer size = 0x4000, period size = 0x200
Dec  8 13:40:16 cactus kernel: ALSA 
../../alsa-kernel/pci/cs46xx/dsp_spos_scb_lib.c:226: 
cs46xx_dsp_proc_free_scb_desc: freeing PCMReader_SCB0
Dec  8 13:40:16 cactus kernel: ALSA 
../../alsa-kernel/pci/cs46xx/dsp_spos_scb_lib.c:226: 
cs46xx_dsp_proc_free_scb_desc: freeing SrcTask_SCB0

With the ALSA-plugin I got the message only one time, with the 
OSS-plugin it apears about a 1000 of times.
(have not done any changes to the cs46xx driver yet,  except what's in 
current CVS ;) )

/Benny



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

Re: Bug in pcm_lib.c ?? (using Xmms)

[Jaroslav Kysela]

On Sun, 8 Dec 2002, Benny Sjostrand wrote:

> Hi!
> 
> Just updated CVS tree and recompiled. Everything works OK execpt that I 
> got this message in log
> when using Xmms with ALSA-plugin and OSS-plugin. Using aplay the message 
> dont apear.
> 
> Dec  8 13:40:16 cactus kernel: ALSA ../alsa-kernel/core/pcm_lib.c:176: 
> BUG: stream = 0, pos = 0x4000, buffer size = 0x4000, period size = 0x200
> Dec  8 13:40:16 cactus kernel: ALSA ../alsa-kernel/core/pcm_lib.c:176: 
> BUG: stream = 0, pos = 0x4000, buffer size = 0x4000, period size = 0x200
> Dec  8 13:40:16 cactus kernel: ALSA ../alsa-kernel/core/pcm_lib.c:110: 
> BUG: stream = 0, pos = 0x4000, buffer size = 0x4000, period size = 0x200
> Dec  8 13:40:16 cactus kernel: ALSA ../alsa-kernel/core/pcm_lib.c:176: 
> BUG: stream = 0, pos = 0x4000, buffer size = 0x4000, period size = 0x200
> Dec  8 13:40:16 cactus kernel: ALSA ../alsa-kernel/core/pcm_lib.c:176: 
> BUG: stream = 0, pos = 0x4000, buffer size = 0x4000, period size = 0x200
> Dec  8 13:40:16 cactus kernel: ALSA 
> ../../alsa-kernel/pci/cs46xx/dsp_spos_scb_lib.c:226: 
> cs46xx_dsp_proc_free_scb_desc: freeing PCMReader_SCB0
> Dec  8 13:40:16 cactus kernel: ALSA 
> ../../alsa-kernel/pci/cs46xx/dsp_spos_scb_lib.c:226: 
> cs46xx_dsp_proc_free_scb_desc: freeing SrcTask_SCB0
> 
> With the ALSA-plugin I got the message only one time, with the 
> OSS-plugin it apears about a 1000 of times.
> (have not done any changes to the cs46xx driver yet,  except what's in 
> current CVS ;) )

It's definitely a bug in CS46xx driver. I've changed the test for the ring 
buffer pointer to be in range 0 .. buffer_size-1 . pointer equal to 
buffer_size doesn't make any sense and it should be really zero.

						Jaroslav

-----
Jaroslav Kysela <perex@suse.cz>
Linux Kernel Sound Maintainer
ALSA Project, SuSE Labs



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

Re: Bug in pcm_lib.c ?? (using Xmms)

[Benny Sjostrand]

>
>
>It's definitely a bug in CS46xx driver. I've changed the test for the ring 
>buffer pointer to be in range 0 .. buffer_size-1 . pointer equal to 
>buffer_size doesn't make any sense and it should be really zero.
>  
>
I believe that this is not a new bug, just undiscovered with previous 
releases.
For the moment I got no idea what possibly could be wrong, depending on
which appl. used it only a few times when playbacking a stream.

/Benny



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

Re: Bug in pcm_lib.c ?? (using Xmms)

[Benny Sjostrand]

>
>
>It's definitely a bug in CS46xx driver. I've changed the test for the ring 
>buffer pointer to be in range 0 .. buffer_size-1 . pointer equal to 
>buffer_size doesn't make any sense and it should be really zero.
>
>  
>
I guess I found a fix for this problem (at least the message disapears 
from the log)
in cs46xx_lib.c in snd_cs46xx_playback_indirect_pointer(...) function:


    if (bytes < 0)
        bytes += buffer_size;
    cpcm->hw_io = ptr;
    cpcm->hw_ready -= bytes;
    cpcm->sw_io += bytes;
-    if (cpcm->sw_io > cpcm->sw_bufsize)
+    if (cpcm->sw_io >= cpcm->sw_bufsize)
        cpcm->sw_io -= cpcm->sw_bufsize;
    snd_cs46xx_playback_transfer(substream, 0);

I cant hear any difference in sound quality, maybe it's just not posible 
to hear the difference ...
... and now it's too late, i'll continue tomorrow ...

/Benny



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