Re: Bonobo Activation Server goes to sleep

[Geoff Hacker <geoffreyhacker bigpond com>]

Shmyrev wrote:

> Well, there are known some problems in 2.4.20 kernel with esd and via chipset. 
> What is your hardware and your kernel version?
>  

Say no more! I'm using version 2.4.20-8 in a Red Hat 9.0 distro. I'm 
running an AMD Athlon 2000XP CPU on a Gigabyte GA-7VM400M motherboard, 
with 256Mb PC2700 DDR RAM and a Seagate 40Gv IDE 7200RPM HDD. Running a 
sound card detection on my machine, I get a VT8233 AC97 Audio Controller 
from Via Technologies. I don't know whether the program is right in 
saying that, but it does play the test sound correctly. The module is 
via82cxxx_audio.

> For example look for:
>
> https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=76603
>  

The summary says it all: (AUDIO VIA82CXXX_AUDIO) via82cxxx_audio + esd 
== bad news! Esd doesn't seem to be able to work with that particular 
kernel driver.

> and
> http://kerneltrap.org/node/view/779
>  

In other words, this sounds like my problems would be fixed if I updated 
to version 2.4.22 of the kernel. Of course, the latest stable version of 
the kernel these days is 2.6.0, but I'd probably have to update a lot of 
other software as well on top of that.

> Does you have the same problems with your gnome 2.2 installation, as I 
> remember you have one? 

I sure do, pretty much as soon as esd tries to play a sound.

> There are some other thing that I don't understand: 
> 1. Who writes /bin/sh: /usr/local/bin/esd?

If you are referring to this line from startx2.log:

/bin/sh: line 1: /usr/local/bin/esd: No such file or directory


Then the answer is that I do not know, though I wish I did. I have not 
found anywhere where this is recorded, except in cases where esd is not 
actually running at the time, which might have been the case here, I do 
not know.

> 2. Did you manually add output "close command executed", because I can't find 
> it place in esound/gnome?

That's correct, I manually added it, because it was my hunch that that 
was where the problem lay. I would have used a proper ESDBG_TRACE macro 
for it, but for some reason, use of those macros in audio.c seems to 
generate linker errors. In any case, here is the routine in question, 
including my debug code, which is the two simple printf statements:

/*******************************************************************/
/* close the audio device */
#ifndef ARCH_esd_audio_close
void esd_audio_close()
{
   printf("about to execute close command\n" );
   close( esd_audio_fd );
   printf("close command executed\n" );
   esd_audio_fd = -1;
   return;
}
#endif

Thanks,

Geoff