New subject: [linux-audio-user] Re: 2.6.15 alsa problem SOLVED

31 Jan 2006

On Tuesday 31 January 2006 20:28, linux-audio-user-request(a)music.columbia.edu 
wrote:
...
  Also a problem
in 2.6.14. I am not using udev or devfs but have
 explicitely created /dev nodes (old fashioned way and far far too many
 entries, but ... always worked before).

Newer kernels boot with many undefines. They seem to correspond to stuff
 in the sound core modules. Do I now need to explicitely load these or
 soundcore.ko on boot up (i.e. placing them in /etc/modules--somewhere
 along the line, snd_seq_midi had to be placed there!). Anybody doing
 this? (Guess it wouldn't hurt to try it.)

Easy enough -- didn't work. It is as if these modules are not exporting
 their symbols correctly. 
 Made a kernel with alsa sound support compiled in! I also compiled in the
 "generic" devices. Result: NO UNDEFINED. My three (a set ncards = 3) became
 dummy, virmidi and the ensonic. The ensonic did not play, however, but the
 alsa restore on bootup failed because of the 1st two. Kmixer settings are
 restored but have no effect.

 So I am trying to build it with just the alsa core compiled in and
 hopefully the three cards will be the mpu401, usb-audio and ensonic loaded
 as modules as before. Wish this did not take so long to try out :- 
Was not so bad. No more undefines and get my UM1 usb and the ens1371 just 
fine. snd=mu401 does not load with same error as in 2.6.12 kernels!

The audio plays through the ensonic just fine.
I used to be able to "cat" the midi from the UM1 cat /dev/midi1 but this did 
not work.
Since I had no midi out, I could not try playing MIDI or "thru"ing the 
keyboard using jackqtl to connect the UM1 to the mpu401.

So ... 
1. Debian stock kernels are not built with alsa's core. I needed to roll my 
own for this.
2. The mpu401 problem remains from 2.6.12.

[linux-audio-user] RE: 2.6.15 alsa problem