Giles Lean said:
> 
> 
> On Fri, 12 Jan 1996 14:13:25 +0100 (MET)  Manuel Bouyer wrote:
> 
> [List changed to port-i386 rather than current users -- I think only
>  the i386 port is using IDE.  Someone can move it back if I'm wrong.]

For now only i386 uses  atapi. But other architectures (like some macs and
dec alpha i think) uses ide drive, but whith a different controller. I don't
know if atapi devices are supported on theses.

> 
> The diffs don't create the atapi files in quite the right place --
> I had to create the /usr/src/sys/atapi directory by hand and move
> the files from /usr/src.

You have to create /usr/src/sys/atapi before applying the diffs. I should
have mentioned that in README.

> 
> To compile I had to add a scsi controller to the configuration
> file.  I presume there is a way to tell config that we need the
> SCSI files? (Any config gurus out there?)

Can you send me your config file ? I didn't have any problems compiling
a kernel whithout any scsi controllers.

> 
> > - Create the devices in /dev, with major 17 for block device and 21 for
> > characters. The minors are as for other disk devices.
> 
> Lets see if I've got this right:
> 
> brw-r-----  1 root  operator   17,   0 Jan 13 18:21 /dev/acd0a
> brw-r-----  1 root  operator   17,   1 Jan 13 18:21 /dev/acd0b
> brw-r-----  1 root  operator   17,   2 Jan 13 18:21 /dev/acd0c
> brw-r-----  1 root  operator   17,   3 Jan 13 18:21 /dev/acd0d
> brw-r-----  1 root  operator   17,   4 Jan 13 18:21 /dev/acd0e
> brw-r-----  1 root  operator   17,   5 Jan 13 18:21 /dev/acd0f
> brw-r-----  1 root  operator   17,   6 Jan 13 18:21 /dev/acd0g
> brw-r-----  1 root  operator   17,   7 Jan 13 18:21 /dev/acd0h

That's rigth, but you need only /dev/acd0a.
> 
> > After reboot you should be able to mount your cdrom. If the mount fails
> > with a 'unit attention' message in the console, retry the mount.
> 
> I got a unit attention probe during boot.  (Full dmesg output
> below.)  Note that the drive was identified correctly.  After boot
> I just get the following error:
> 
> wdc0: lost interrupt
> wdc0: lost interrupt
> wdc0: lost interrupt
> wdc0: lost interrupt
> wdc0: lost interrupt
> acd0(wdc0:1): illegal length indication 
> 
> Same deal trying to read the raw partitions.  (The machine's "disk
> light" does go on, so I think the IDE interface is being used. The
> CD ROM's light doesn't go on.)
> 
> Thoughts and suggestions about what to do next are welcome; I plan
> to try another IDE disk as the master, just in case the disk is
> the problem.
> 
> The machine is crash and burn, by the way (hence the IDE disks :-)
> and so available for testing.
> 

I've two different pc here, but with the same atapi cdrom. I don't have any
problems. Perhaps you could #define ATAPI_DEBUG, ATAPI_DEBUG_WDC,
WDCNDELAY_DEBUG and DIAGNOSTIC in sys/dev/isa/wdc.c, recompile your kernel
and send me the ouput ? This could help.

--
Manuel Bouyer, Ecole Nationale Superieure de Techniques Avancees, Paris
email: bouyer@ensta.fr
--