Get FC2 to load 1394 drivers automatically

[Mark Knecht]

On Wed, 08 Dec 2004 09:36:18 -0800, Rick Stevens
<rstevens at vitalstream.com> wrote:
<SNIP>
> > I see what you see at line 753. I don't read this stuff very well, but
> > clearly the script is expecting /proc/bus/ieee1394 and I don't have it
> > but things still work.
> >
> > Interesting!
> 
> Yeah.  I'm wondering if perhaps it appears right after loading the ieee
> modules and disappears after sbp2 is loaded.  Ok, now my curiosity is
> piqued.  I'll have to stop by Fry's on the way home and get a firewire
> drive to play with.  Sheesh!  Just what I need...another piece of
> hardware!  My office already looks like a combination of a bombed
> computer factory and a combined harvester!

Just went to Fry's. Bought one of these $199 computers that runs
Linspire for my kid for Christmas. We'll see how that works out...

I know there were changes between 2.4 where /proc/bus/ieee1394 existed
and 2.6 where it was moved somewhere else. Just looking around I don't
see it though.

Looking under /proc/scsi I see the 1394 drive though:

[mark at Godzilla mark]$ cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
  Vendor: IC35L080 Model: AVVA07-0         Rev:
  Type:   Direct-Access                    ANSI SCSI revision: 06
[mark at Godzilla mark]$

This is not the FC2 kernel BTW. This one is from PlanetCCRMA. However
I do not think it changes the 1394 stuff at all.
> 
> >>I should have caught this sooner, Mark, but I don't have a firewire
> >>device so I wasn't familiar with the mechanism used.  I should have
> >>guessed that an alias line would trip it.  I'm embarrassed.  :-]
> >
> >
> > Please don't be. I'm grateful to have such attention and fast responses!
> 
> [blush!]  Still embarrassed.
> 
> > I was able over the last couple of days to do about 8 hours of audio
> > recording to 1394 drives using FC2 with not a single problem.
> > Everything worked really well. After that nice experience I'm trying
> > to get the machine into the same state at boot time. This requires the
> > 1394 stuff that you just helped me with, and also echoing a value to
> > /proc/irq/... to set up interrupt threading in this kernel. Off to
> > work on that one now.
> 
> Really?  /proc/irq?  Weird.  

It's very new stuff. Ingo has been doing kernel patches that allow you
to make individual IRQ handlers threaded or non-threaded. For low
latency audio work I make the sound card non-threaded which gives it
lower response times from the kernel when an interrupt happens. So far
I haven't had any need to make 1394 non-threaded, but I have the
option.

> Well, I'd say take a look at sysctl...but
> it controls stuff in /proc/sys.  Playing with /proc/irq smells like
> a classic /etc/rc.d/rc.local hack to me.  If you want to risk it and
> it's sbp2 specific, you could tack on
> 
>         && echo "value" >/proc/irq/whatever

IIRC on Gentoo I did it in local.start. Could I do it in rc.local in FC2?

Thanks,
Mark