Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 9 Apr 1999 15:54:46 -0600 (MDT)
From:      "Kenneth D. Merry" <ken@plutotech.com>
To:        wilko@yedi.iaf.nl (Wilko Bulte)
Cc:        freebsd-scsi@freebsd.org
Subject:   Re: Getting Pioneer CD changer to work
Message-ID:  <199904092154.PAA18063@panzer.plutotech.com>
In-Reply-To: <199904091815.UAA13726@yedi.iaf.nl> from Wilko Bulte at "Apr 9, 1999  8:15:35 pm"

next in thread | previous in thread | raw e-mail | index | archive | help
Wilko Bulte wrote...
> As Kenneth D. Merry wrote ...
> > Wilko Bulte wrote...
> > > As Kenneth D. Merry wrote ...
> > > > Wilko Bulte wrote...
> > > > > As Kenneth D. Merry wrote ...
> > > > > > Wilko Bulte wrote...
> > > > > > > As Kenneth D. Merry wrote ...
> > > > > > > > Wilko Bulte wrote...
> > > > > > > > > As Kenneth D. Merry wrote ...
> > > > > > > > > > Wilko Bulte wrote...
> > > > > > > > > > > As Kenneth D. Merry wrote ...
> > > > > > > > > > > > Wilko Bulte wrote...
> > > > > panic as far as I can see. I'll try to reproduce.. Just the DD seems
> > > > > to work just fine.
> > > > 
> > > > Hmm, sounds strange indeed.
> > > 
> > > In the meantime I ran the test again. No machine lockup this time. But 
> > > already twice a solid lockup of the drive.. It goes completely comatose,
> > > not even a SCSI reset from a FreeBSD reboot awakens it. The only thing
> > > what works is a whack over the head with the powerswitch. This definitely
> > > is a brilliant piece of firmware I have.
> > > 
> > > Console messages:
> > > 
> > > (cd0:isp0:0:5:0): cddone: got error 0x6 back
> > > (cd4:isp0:0:5:4): cddone: got error 0x6 back
> > > (cd3:isp0:0:5:3): cddone: got error 0x6 back
> > > (cd1:isp0:0:5:1): cddone: got error 0x6 back
> > 
> > Hmm, looks somewhat familiar.  Error 6 is ENXIO.  If you boot with -v you
> > might get more info.  You can probably fix it by increasing the minimum
> > and possibly the maximum changer timeouts.  The defaults are 2 seconds and
> > 10 seconds, respectively.
> 
> The boot messages are OK, so is the probe. I've diagnosed it as
> follows:
> 
> - probes ok
> - works ok for some time (running my test script)
> - locks up it's own firmware 
> - kernel throws away the devices associated with the changer,
>   thus cddone 0x6 and 'no such device' errors for 'dd'

Right, sounds familiar.

> > Try increasing the minimum to 8 seconds and the maximum to 20:
> > 
> > options		"CHANGER_MIN_BUSY_SECONDS=8"
> > options		"CHANGER_MAX_BUSY_SECONDS=20"
> > 
> > I had similar problems with Lars Fredriksen's Nakamichi changer that I
> > borrowed to do the changer functionality in the CD driver.  The default
> > timeouts may be a little too low, but I think they work okay for that sort
> > of changer at least.
> 
> In this case it is not the changer. The lockup that happened with me 
> present was during reads by two parallel operating 'dd's like:
> 
> 	while :
> 
> 		dd <bla>
> 		sleep 5
> 		dd <bla>
> 	done
> 
> So, the drive is seeking quite a bit (you can hear that, and observe it
> as the case is still off). My guess: the firmware looses track of the
> parallel I/Os going on. I'm retesting with a single 'dd' in the loop.

I'm confused here.  When you say "dd <bla>" above, do you mean that each dd
is going to a *separate* CD on the changer, or are you just reading with
two different processes from the same CD?  (so the changer isn't switching
CDs)

With the Nakamichi changer I had, I found that if it got pounded too often
to switch back and forth (i.e. with I/O going to a number of different LUNs
at the same time) it would eventually get confused and stop responding on
various LUNs.

If you increase the minimum timeout, it'll prevent that from happening.

> > The BDR functionality in the kernel doesn't work quite right yet, so if you
> > try something just try the bus reset.
> > 
> > > Sidetracking a bit: man xpt mentions /dev/xpt. /dev/MAKEDEV does not
> > > have it though. Is it supposed to be a manual operation with mknod? 
> > > Or is my /dev/MAKEDEV out of date?
> > 
> > The xpt(4) man pages mentions /dev/xpt0.  If your MAKEDEV doesn't have it,
> > it's out of date:
> > 
> > {roadwarrior:/usr/home/ken/perforce/cam/etc/etc.i386:26:0} grep xpt MAKEDEV
> > 
> >         sh MAKEDEV pass4 xpt2                           # cdev, CAM
> > xpt*)
> >         name=xpt
> >         units=`expr $i : 'xpt\(.*\)'`
> > 
> > 
> > That's from -current as of about 3am MDT today.  All you need to do is grab
> > a new copy of MAKEDEV, and then:
> 
> Right. It proved to be in my -current source tree alright. Apparantly it
> did not get installed by the 'installworld' for some reason ?? My testbox
> started off as a 2.2.6 CD install. Ah well, it is there now. camcontrol
> is happy.
> 
> If we ever meet, I definitely owe you a couple of beers ;-)

Coke (the kind that comes in red cans) would be preferrable. :)

Ken
-- 
Kenneth Merry
ken@plutotech.com


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-scsi" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199904092154.PAA18063>