[linux-audio-user] asfxload and 2.6 kernel

[Erik Steffl]

Jeremiah Benham wrote:
> I have noticed since I switched to the 2.6 kernel tree I have been having troubles with asfxload and sfxload. When I try to launch it hangs.
> The program just sits their running and can't be killed. It does not use to many clock cycles but I can't seem to kill it no matter what I do.
> I tried killall asfxload and other tricks. I also tried to kill it with kde's process management program but it still did not die. A few times it complained that 
> /dev/sequencer was not available no such device. I was wondering if there is something I am doing wrong or if I should file a bug report. Anyone else experience 
> this?
> 
> sound card = SB live
> Kernel = 2.6.2 (same problem with other versions of 2.6)
> asfxload =  ver.0.5.0c

   me: kernel 2.6.5, sb live! platinum, asfxload 0.5.0b

> launching asfxload by itself complains:
> No Emux synth hwdep device is found
> 
> so I modprobe snd-hwdep and snd-emux-synth but it still complains

   I tried asfxload and it looks like it's dead (waiting for some system 
call) but it actually works, at least on my machine. However it takes 
about 10+ hours to load soundfonts (yes, more then ten hours). Attach 
strace to see what it's doing  (it loads a bit and waits forever, loads 
a bit and waits forever).

   btw I think you need snd-seq module (maybe some other, I don't 
remember, just look at what snd-* modules you have and load some:-)

   I posted a message to alsa-user and here's what I've got (only one 
response):

> At Wed, 22 Sep 2004 05:32:38 -0700,
> Erik Steffl wrote:
> 
>>> 
>>>    I used to use sfxload to load soundfonts before and it worked (same 
>>> soundfont file, same alsa install). I just tried the asfxload and it 
>>> takes, so far, several hours to load the soundfont file.
> 
> 
> Internally there is no big difference between sfxload and asfxload.
> The latest CVS version has the __NORETRY flag indicating that the
> driver doesn't wait when the page isn't available, so it won't take
> hours but will lack some data.
> 
> Anyway, it's likely a kernel issue.  There was a bugreport telling
> that 2.6.9rc2-mm1 fixed the memory handling.
> 
> 
> Takashi
>