[ALSA - driver 0001748]: sound driver will not load (CS4237B chip)

[bugtrack@alsa-project.org]

A NOTE has been added to this issue.
======================================================================
<https://bugtrack.alsa-project.org/alsa-bug/view.php?id=1748> 
======================================================================
Reported By:                briand
Assigned To:                
======================================================================
Project:                    ALSA - driver
Issue ID:                   1748
Category:                   ISA - cs4236
Reproducibility:            always
Severity:                   major
Priority:                   normal
Status:                     new
Distribution:               Fedora Core 3
Kernel Version:             2.6.9-1.667
======================================================================
Date Submitted:             01-14-2006 23:47 CET
Last Modified:              01-22-2006 10:18 CET
======================================================================
Summary:                    sound driver will not load (CS4237B chip)
Description: 
with BIOS set to "PnPOS" for sound chip:
  `/sbin/modprobe snd-cs4236` produces the following on stdout:
  "FATAL: Error inserting snd_cs4236 \
   (/lib/modules/2.6.9-1.667/kernel/sound/isa/cs423x/snd-cs4236.ko): \
   No such device"

   -- occurs with 1.0.10, 1.0.11rc1, and (currently) 1.0.11rc2
   alsaconf reports I have no sound card.
   cat /proc/asound/cards reports I have no sound card.


with BIOS set to "Enable" for sound chip:

 `/sbin/modprobe snd-cs4236 port=0x530 cport=0x538 isapnp=0 dma1=1 dma2=0
irq=5`

  ..produces nothing on stdout/stderr, but the laptop speakers click once,
and dmesg shows:

 "ALSA /path/to/alsa/source/isa/cs423x/cs4236_lib.c:305: CS4236+ chip
detected, but control port 0x538 is not valid"

  - that message shows up in dmesg regardless of what i specify for
'cport=' parameter (with, of course, the corresponding port value in the
error message).

  alsaconf seems to find the card, test the card, and reports all is fine;
offers to modify /etc/modprobe.conf (with parameters shown above), and
congratulates me.
  cat /proc/asound/cards reports I have no sound card.


In looking at cs4236_lib.c, lines 301-307:

        ver1 = snd_cs4236_ctrl_in(chip, 1);
        ver2 = snd_cs4236_ext_in(chip, CS4236_VERSION);
        snd_printdd("CS4236: [0x%lx] C1 (version) = 0x%x, ext = 0x%x\n",
cport, ver1, ver2);
        if (ver1 != ver2) {
                snd_printk("CS4236+ chip detected, but control port 0x%lx
is not valid\n", cport);
                snd_device_free(card, chip);
                return -ENODEV;
        }

  ...it looks like ver1 and ver2 aren't agreeing.  ;)   Okay, so I
experimented a bit.  I commented out (#if 0 ... #endif) the if (ver1 !=
ver2) {...} block, and added:

   snd_printk("CS4236: [0x%lx] C1 (version) = 0x%x, ext = 0x%x\n", cport,
ver1, ver2);

 ...recompiled, reinstalled... and then:

  `modprobe snd-cs4236 port=0x530 cport=0x538 isapnp=0 dma1=1 dma2=0
irq=5`

  results in:

   /dev/dsp, /dev/audio, /dev/snd/* all appear, laptop speakers click once
(as before), followed quickly by a <blart>, and we're on our way...
almost.

  my shorthand dmesg (printk) shows me that the driver was unloading
because 0xFF did not equal 0xE8. (I'm going from memory here.  I'm
positive that the value of "ver1" was 0xFF -- not quite sure about "ver2",
but it's close.  ..it definitely wasn't 0xFF (obviously!) which is why the
driver load aborted).

  cat /proc/asound/cards reports card 0 is a CS4237B at IRQ 5, DMA 1,0

  in KDE, kmix loads and allows me to play with buttons/sliders - I can
get CD audio to play out the laptop speakers, but other audio does not
work.  xmms will load an mp3 file, but when I click the play button, the
display shows 0:00, and nothing plays.  No sound, and the "progress bar"
and "time remaining" stay at 0.

  in terminal, `alsamixer` results in a segmentation fault.

  ..incidentally, in this "hacked" module, it doesn't really matter what
(reasonable) value I specify for cport= in /etc/modprobe.conf (0x538,
0x120, 0x110 ...).  The result is the same.  kmix allows some adjustment,
cd audio plays, no other sound, and alsamixer seg faults.  No doubt, if I
were to work at it, I could generate a kernel panic with the right
value... ;)




======================================================================

----------------------------------------------------------------------
 briand - 01-21-06 14:41 
----------------------------------------------------------------------
re: the spec sheet:  It is no longer available on the manufacturer's
(Cirrus Logic) web site.  I did find one place (an IBM laptop site,
concerned with making drivers for OS/2 and OS/2 Warp) that had a link to a
"local copy" of the PDF file.  So, the file is not easily found, but out
there.  If it is  needed/wanted, I have a copy I can send via email.

I have not tried installing any earlier kernel on this machine,
personally, but have read anecdotal evidence on the web (google search)
that sound had not worked with the OSS driver(s), and everyone (FC2 era
and earlier) that wrote about success with ALSA drivers did not get it to
work as a PnP device, ever.

I think the most recent modification to the driver (snd-cs4236) added the
section outlined above (in cs4236_lib.c) to solve some (other, unknown to
me) problems with actual cs4236 and cs4236b chips, but it appears (at that
point) to have broken support for the cs4237b chip at that time.  My
opinion here is bolstered by the fact that alsaconf purports to find the
chip (albeit, not when it's in PnP mode) and "goes through the motions" to
probe it, install the driver (and it always picks snd-cs4236), and even
claims to have !tested! it (admittedly, this test goes by pretty quick,
and no audio is heard outside of a couple clicks).  The actual driver,
however, doesn't load ("CS4236+ chip detected, but control port 0x538 is
not valid"), even when alsaconf is in the midst of
loading/testing/installing it.

I have not read anywhere online, nor talked to anybody (live or online)
that has had success loading the driver (recent vintage) in "explicit"
mode (it now aborts loading at the code snippet posted above), nor ever
having loaded it successfully in PnP mode.

My opinion (at this point) is that:

a) PnP has never worked for this chip, using any version of any driver,
and that 

b) those who have had success in the past (using snd-cs4232 and later,
snd-4236) did so by bypassing PnP (isapnp=0) and explicitly passing
parameters, using an ALSA driver created before the most recent changes to
the cs4236 library.

Perhaps all that is needed is for this chip to be properly "identified" so
that it passes the (ver1 == ver2) test [above].  Maybe it needs a whole
new driver; I don't know.  I am certainly willing to work with the
developers to help pinpoint the problem(s), and get sound working on this
laptop.  :)

----------------------------------------------------------------------
 perex - 01-22-06 10:18 
----------------------------------------------------------------------
What's output from this command on your machine (try to toggle the BIOS PnP
OS option as well)?

" for i in /sys/bus/pnp/devices/* ; do cat $i/id ; cat $i/resources ; done
"

Issue History
Date Modified  Username       Field                    Change              
======================================================================
01-14-06 23:47 briand         New Issue                                    
01-14-06 23:47 briand         Distribution              => Fedora Core 3   
01-14-06 23:47 briand         Kernel Version            => 2.6.9-1.667     
01-14-06 23:51 briand         Issue Monitored: briand                      
01-21-06 03:48 briand         Note Added: 0007733                          
01-21-06 04:13 rlrevell       Note Added: 0007734                          
01-21-06 14:41 briand         Note Added: 0007739                          
01-22-06 10:18 perex          Note Added: 0007744                          
======================================================================




-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642