cdrom mount 2.0.27 vs 2.0.34 kernels
I have run across a problem getting my Debian 2.0.34 install to
mount my cdrom. I posted this message onto the debian-user list
and Helge Hafting responded with a lot of help. However, we did
not come to a final understanding of my problem, nor a solution.
I send three of these emails which contain some insight into
what my problem is, and some of my trials and results. However,
the solution to this cdrom mount seems to us to be within the
programing of the kernel or the device drivers and may require
the assistance of you developer types. Basically, I can get
older kernels to mount the cdrom driver, but the newer ones
wont. Something changed between kernels 2.0.27 and 2.0.34, and I
now can't mount my Hitachi CDR-1750S cdrom at /dev/scd0. If I
touch that file with a mount or cp command, the system goes into
the following endless series of one line commands.
Aiee scheduling in interrupt 001260b1
I hoping someone in debian-devep will recognize what my cdrom
problem (either from the intro info above or the more detailed
info stated in my emails below) is and how to solve the problem.
Sincerely, Wes Jennings
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* We must be land stewards. . . . for our future depends on it! *
* Wes Jennings, soil scientist wes@blarg.net 1-253-474-5432 *
* Puget Land Consultants P.O. Box 9635 Tacoma, WA 98409 *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
=============================================================
Date: Wed, 4 Nov 1998 04:03:59 -0800 (PST)
From: Wes Jennings <wes@blarg.net>
To: debian-user@lists.debian.org
Subject: Mounting problem on cdrom
Hope someone can determine what this problem is. Outlined
below is information about my system, observations and error
messages.
1. I have had a Slackware Linux system, kernel 2.0.27, running
for about 1 year. All my hardware functions fine. I decided to
upgrade my system and wanted to try the Debian 2.0 release,
kernel 2.0.34. Instead of upgrading my Slackware system, disk
partition sdb1, I decided to split my dos disk in half and use
sda1 for dos and sda2 for a Debian partition. I want to play
around with a KISS Linux setup (non-windows) for command line
applications and emergency use. I figured I could learn the
Debian install routine on this new partition before tackling the
conversion of my main Linux partition from Slackware to Debian.
Here is some additional info that may be needed.
Amd 5x86, 133 processor
AMIBIOS, release 5/16/1996
Adaptec 1520 SCSI card
Hitachi CDR-1750S external SCSI cdrom
2. After backup my existing partitions and reading the install
documentation I started the Debian install using the floppy
disks for boot/rescue, drivers, and the basic disk sets
base144-1 thru 5. The install seemed to go alright. I was able
to setup LILO to boot from the hard drive. I also setup a boot
floppy. The system boots properly from both the hard drive and
boot floppy.
3. I did some more reading about dselect to be able to install
more of the Debian system. I then fired up dselect. Here,
however, I run into a serious problem getting dselect to mount
my SCSI cdrom. From dselect I choose Access, cdrom source, and
then filled out the device as /dev/scd0. The result was an
endless series of lines with the following message.
Aiee scheduling in interrupt 001260b1
I could not find a way to escape out of this endless message.
4. I next hit the reset button, Ctrl-Alt-Del did nothing, and
got the system back up. I next tried to manually mount the SCSI
cdrom in the Debian partition with the following command.
mount -t iso9660 /dev/scd0 /mnt
The result was the same as in 3.
5. I then used a Slackware floppy to boot up the Slackware
partition. I tried to mount the cdrom with the exact same
command as in 4. and the cdrom mounted properly.
6. Next I wondered if the Debian distribution was using some
other system to refer to cdrom drives: like using /dev/scd1
instead of /dev/scd0. I tried all 7 files /dev/scd1-7 with the
manual mount command. None mounted, but there was no endless
messages as when I tried to manually mount /dev/scd0. So I
conclude that /dev/scd0 is the proper device file.
Now I'm getting desperate, and I tried some wild stuff to see if
I could shed some light on my problem and find a way to fix
things.
7. Next I copied a Slackware kernel image that I knew would
mount the cdrom drive over to the Debian partition and made a
changed in a symbolic link as follows.
/vmlinuz > /boot/vmlinuz-slackware
I then booted the Debian partition, and things went along just
fine. Well until I tried to manually mount the cdrom. The result
was exactly the same as point 3. Oh well, I now know, from
experience, we can mix and match kernels if need be.
8. Please don't ask why, but I next tried to see if I could copy
device files. Could I possible copy the /dev/scd0 from Slackware
to my Debian partition? Well, I learned some crazy stuff can
happen. The following command cp /dev/sda1 /dev/sda1.0 copied
the contents of my dos partition to file /dev/sda1.0. And this
command cp /dev/scd1 /dev/scd1.0 returned the message
cp: scd1: Device not configured. But, cp /dev/scd0 dev/scd0.0
did the unexpected. It sent the system into the same endless loop
of lines and message as when I attempted to mount the cdrom drive!
CONCLUSION - Well, my conclusion from all this, is something
must be wrong with file /dev/scd0. Well, it seems that this is
the case to this novice-intermediate Linux user.
QUESTION: If I have identified the problem correctly, what
should I do to get a good copy of the /dev/scd0 file onto my
Debian partition? And if I have mis-identified the problem, or
need to do some further testing (hacking?), please advise what
else I should do or consider.
A bit discouraged, but learning from this experience,
Wes Jennings
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* We must be land stewards. . . . for our future depends on it! *
* Wes Jennings, soil scientist wes@blarg.net 1-253-474-5432 *
* Puget Land Consultants P.O. Box 9635 Tacoma, WA 98409 *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
==================================================================
>From wes@blarg.net Thu Nov 12 19:32:47 1998
Date: Thu, 5 Nov 1998 15:13:57 -0800 (PST)
From: Wes Jennings <wes@blarg.net>
To: Helge Hafting <helge.hafting@daldata.no>
Subject: Re: SCSI cdrom mounting, #2
Helge,
Thanks for the response. I have quoted your entire email as it
contains my second email on my cdrom mount problem. I have
answered your questions at the end of this message. I will also
forward to you a draft of my first email which has some other
valuable info. I too think the problem is in the device driver,
but I don't understand my options to be able to correct the
situation. For example, can I simply copy the Slackware device
driver /dev/scd0 onto my Debian partition? Is this something to
bring to the attention of the development team?
Wes
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* We must be land stewards. . . . for our future depends on it! *
* Wes Jennings, soil scientist wes@blarg.net 1-253-474-5432 *
* Puget Land Consultants P.O. Box 9635 Tacoma, WA 98409 *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
On Fri, 6 Nov 1998, Helge Hafting wrote:
> In <Pine.LNX.3.95.981104152650.177B-100000@affco>, on 11/04/98
> at 03:34 PM, Wes Jennings <wes@blarg.net> said:
>
> >To: The Debian Gang - A howdy to all:
>
> >Since I posted my message yesterday on debian-user about having problems
> >mounting my Hitachi CDR-1750S cdrom I have found out some additional
> >information. As a review, the problems is that when I give the following
> >command
>
> >mount -t iso9660 /dev/scd0 /mnt
>
> >my Debian system starts a continous one line message loop that can't be
> >broken out of: not even Ctrl-Alt-Del. Please see my previous message for
> >more details of what I did. Now the additional information.
>
> >1. I decided to try out the possibility that there was a termination
> >problem, or bad cable, in my external SCSI line. Therefore I put a
> >termination plug on the outside of my SCSI card and tried to mount
> >something with the above mount command. Of course, nothing would be able
> >to mount, but I wanted to see if the endless one line message loop
> >happened. It did not, and returned the following error message.
>
> >mount: the kernel does not recognize /dev/scd0 as a block device
> > (maybe `insmod driver'?)
>
> >2. Therefore, I moved the termination plug so that there was a cable from
> >the SCSI card to the Hitachi cdrom, and the termination plug was used to
> >terminate the SCSI line after the cdrom. I left the Hitachi cdrom off,
> >and issued the same mount command. There was no endless one line message
> >loop and the follow error message received.
>
> >SCSI error: host 0 id 3 lun 0 return code = 27010000
> >o(very small o)Sense class 0, sense error 0, extended sense 0
>
> >These two lines were repeated many times with a few additional lines
> >thrown in.
>
> >3. I next turned the Hitchi cdrom on and ran the same mount command
> >listed above. Under these conditions, Debian went into the endless one
> >line loop bit again. By the way, for those who have not seen my first
> >post on this, the endless one line message is as follows.
>
> >Aiee scheduling in interrupt 001260b1
>
> >4. Lastly, again as a check, I boot my main Slackware partition and again
> >run the same mount command, and the Hitachi cdrom is mounted and
> >functional.
>
>
> >CONCLUSION - Well, I'm still stumped for a solution to this
> >problem. Does anyone have any suggestions as to where I go from here
> >toward a solution?
>
> >Wes Jennings
>
> The thing works fine with slackware and fails with an "Aiee" in debian?
> Looks like a device driver problem to me.
> Are the kernels used in slackware and debain compiled from different
> sources?
The Slackware is an earlier compile. It is kernel 2.0.27. The
Debian Kernel is 2.0.34.
> Possibly different versions or at least different options?
> Which kernel is the newest one, and what scsi controller do
> you use?
> Helge Hafting >
There were different options built into the kernels, but see my
first email to debian-user where I describe running the
slackware kernel under Debian.
If you have any suggestions what I should do here it would be
appreciated. To get more of the Debian Distribution loaded onto
my Debian partition, I have put my old Sound Blaster cdrom card
and drive back into my computer. It works, but I don't want to
leave it permanently in my computer as the power supply is
nearly max. out. The Hitachi cdrom is an external SCSI. Again I
will forward my earlier email that you may not have seen.
Thanks again, Wes Jennings
==================================================================
>From wes@blarg.net Thu Nov 12 19:33:34 1998
Date: Mon, 9 Nov 1998 12:36:23 -0800 (PST)
From: Wes Jennings <wes@blarg.net>
To: Helge Hafting <helge.hafting@daldata.no>
Subject: Re: cdrom mount problem, message 1
Helge,
Thanks for your insight into my cdrom mount problem. Here are my
results.
1. I did as you requested and added a section into lilo.conf for
the slackware kernel. Here is a copy of lilo.conf.
boot=/dev/sda2
root=/dev/sda2
install=/boot/boot.b
map=/boot/map
vga=normal
delay=20
image=/vmlinuz
label=Linux
read-only
image=/boot/vmlinuz.scsi-only
label=slackware
read-only
Note that the first boot position gives the boot image at
/vmlinuz. vmlinuz is a symbolic link to /boot/vmlinuz-2.0.34.
This is the system kernel written during the Debian
installation. I added to lilo.conf the second boot option with
reference to the slackware kernel which I copied onto the Debian
partition as /boot/vmlinuz.scsi-only. I then ran the lilo
command. Now the results.
2. During the boot-up process I allowed the first boot position
to load by default. I then tried to mount my SCSI Hitachi cdrom
at /dev/scd0 with the mount command
mount -t iso9660 /dev/scd0 /mnt
The result was the system went into the endless one line loop
with message Aiee scheduling in interrupt 001260b1
3. I hit the reset button on my the computer, waited for fsck to
clean-up my Debian partition, /dev/sda2, and then manually sent
lilo to boot the slackware kernel. I ran the exact same mount
command as in point 2. The slackware kernel mounted the Hitachi
cdrom properly. WALLA! It works. You were right Helge.
4. Now I have another question to help me understand whats going
on. Does the slackware kernel, when properly setup with lilo,
utilize the device drivers on the Debian or Slackware partitions
on my computers? Slackware is on a 2.1Gig partition: /dev/scb1.
Debian is on a 100Meg partition: /dev/sca2. There is a 100Meg
dos partition: /dev/sca1. The boot strap is on the /dev/sca
drive. Finally fstab on the Debian /dev/sda2 partition lists
only mounting references for proc and /dev/sda2.
5. A side point to something you mentioned. You said I would
benefit from converting to Debian because it has more depth or
flexibility, or something of that nature. Do you mind sharing
what you meant by this comment?
WJ COMMENTS - I have an uneasy feeling about what is happening
during bootup. I want to convert my main SCSI partition to run
Debian instead of Slackware. So if the Slackware kernel runs
because it has the right SCSI components built into it (and
accepts the Debian device driver at /dev/scd0) I can compile a
Debian kernel that will work the same. However, if the Slackware
kernel has built into it to reference the Slackware device
driver /dev/sdc0 on partition /dev/sdb1, then I still have a
problem. Helge, some insight with what is happening here would
be greatly appreciated.
I want to tell you I really do appreciate your help. When we get
to the bottom line of this problem, I want to write a message
back to the linux-users so that others will benefit from my
experiences.
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* We must be land stewards. . . . for our future depends on it! *
* Wes Jennings, soil scientist wes@blarg.net 1-253-474-5432 *
* Puget Land Consultants P.O. Box 9635 Tacoma, WA 98409 *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
On Mon, 9 Nov 1998, Helge Hafting wrote:
> In <Pine.LNX.3.95.981105152217.174T-100000@affco>, on 11/05/98
> at 03:23 PM, Wes Jennings <wes@blarg.net> said:
>
>
> >Date: Wed, 4 Nov 1998 04:03:59 -0800 (PST)
> >From: Wes Jennings <wes@blarg.net>
> >To: debian-user@lists.debian.org
> >Subject: Mounting problem on cdrom
>
> >Hope someone can determine what this problem is. Outlined
> >below is information about my system, observations and error messages.
>
> >1. I have had a Slackware Linux system, kernel 2.0.27, running for about
> >1 year. All my hardware functions fine. I decided to upgrade my system
> >and wanted to try the Debian 2.0 release, kernel 2.0.34. Instead of
> >upgrading my Slackware system, disk partition sdb1, I decided to split my
> >dos disk in half and use sda1 for dos and sda2 for a Debian partition. I
> >want to play around with a KISS Linux setup (non-windows) for command
> >line applications and emergency use. I figured I could learn the
> >Debian install routine on this new partition before tackling the
> >conversion of my main Linux partition from Slackware to Debian. Here is
> >some additional info that may be needed.
>
> > Amd 5x86, 133 processor
> > AMIBIOS, release 5/16/1996
> > Adaptec 1520 SCSI card
> > Hitachi CDR-1750S external SCSI cdrom
> I know of no particular problems with this.
>
> >2. After backup my existing partitions and reading the install
> >documentation I started the Debian install using the floppy disks for
> >boot/rescue, drivers, and the basic disk sets
> >base144-1 thru 5. The install seemed to go alright. I was able to setup
> >LILO to boot from the hard drive. I also setup a boot floppy. The system
> >boots properly from both the hard drive and boot floppy.
>
> >3. I did some more reading about dselect to be able to install more of
> >the Debian system. I then fired up dselect. Here, however, I run into a
> >serious problem getting dselect to mount my SCSI cdrom. From dselect I
> >choose Access, cdrom source, and then filled out the device as /dev/scd0.
> >The result was an
> >endless series of lines with the following message.
>
> >Aiee scheduling in interrupt 001260b1
> That look like some sort of driver problem to me.
>
> >I could not find a way to escape out of this endless message.
>
> >4. I next hit the reset button, Ctrl-Alt-Del did nothing, and got the
> >system back up. I next tried to manually mount the SCSI cdrom in the
> >Debian partition with the following command.
>
> >mount -t iso9660 /dev/scd0 /mnt
>
> >The result was the same as in 3.
> manual or automatic mount is supposed to work the same way, so this isn't
> surprising.
>
> >5. I then used a Slackware floppy to boot up the Slackware
> >partition. I tried to mount the cdrom with the exact same
> >command as in 4. and the cdrom mounted properly.
> Something is obviously broken in debian then. soemthing
> that works in slackware.
>
> >6. Next I wondered if the Debian distribution was using some other system
> >to refer to cdrom drives: like using /dev/scd1
> >instead of /dev/scd0. I tried all 7 files /dev/scd1-7 with the manual
> >mount command. None mounted, but there was no endless messages as when I
> >tried to manually mount /dev/scd0. So I
> >conclude that /dev/scd0 is the proper device file.
> Yes, I don't think there is any differences in device numbering here.
>
> >Now I'm getting desperate, and I tried some wild stuff to see if I could
> >shed some light on my problem and find a way to fix things.
>
> >7. Next I copied a Slackware kernel image that I knew would
> >mount the cdrom drive over to the Debian partition and made a changed in
> >a symbolic link as follows.
>
> >/vmlinuz > /boot/vmlinuz-slackware
>
> >I then booted the Debian partition, and things went along just fine. Well
> >until I tried to manually mount the cdrom. The result was exactly the
> >same as point 3. Oh well, I now know, from experience, we can mix and
> >match kernels if need be.
> You are sure it really was the slackware kernel booting, that is, you
> confirmed this by looking at the boot-up version number? (You can use
> "dmesg | more" or shift+pgup if boot messages scroll by too fast.)
>
> I ask because making a link isn't enough. You will also have to re-run
> lilo. This because the filesystem isn't used
> in the usual way when loading the kernel. The boot loader doesn't load a
> file, it loads a hardcoded set of disk addresses. Making a link doesn't
> change that set. Running lilo does, as lilo reads the filesystem and
> updates the disk address list.
> Maybe you knew all that, but you didn't say so, so I had to
> tell you just in case. If you really booted the slackware
> kernel and got the same problem then the problem is elsewhere, possibly in
> "mount". If so, try replacing debian's "mount" binary. I don't really
> believe in this though.
>
> >8. Please don't ask why, but I next tried to see if I could copy device
> >files. Could I possible copy the /dev/scd0 from Slackware to my Debian
> >partition? Well, I learned some crazy stuff can happen. The following
> >command cp /dev/sda1 /dev/sda1.0 copied the contents of my dos
> >partition to file /dev/sda1.0. And this command cp /dev/scd1
> >/dev/scd1.0 returned the message cp: scd1: Device not configured.
> >But, cp /dev/scd0 dev/scd0.0 did the unexpected. It sent the system into
> >the same endless loop of lines and message as when I attempted to mount
> >the cdrom drive!
> Copying device files is indeed possible, but not by a plain "cp" command.
> A plain "cp" will read data from the device, (i.e. raw tracks from a cdrom
> or harddisk. That's how your partition contents got copied.) "cp" have
> some special options for copying device
> special files. Use "man cp" for further info. I don't think copying the
> device files will help you though.
>
> Using a plain "cp" command for reading /dev/scd0 will give exactly the
> same errors as when mounting, because the cdrom is read in the same way in
> either case. This still look like a buggy driver.
>
> >CONCLUSION - Well, my conclusion from all this, is something must be
> >wrong with file /dev/scd0. Well, it seems that this is the case to this
> >novice-intermediate Linux user.
>
> Your copying from /dev/scd0 crash show that the error isn't in the
> filesystem or "mount", as you bypassed both by reading directly from the
> device. Try that with a working kernel and you'll get a cdrom image file
> in your filesystem - if there is room.
> This is standard practice for those who copy cdrom's and have to read the
> original first using the cdwriter.
>
>
> >QUESTION: If I have identified the problem correctly, what
> >should I do to get a good copy of the /dev/scd0 file onto my Debian
> >partition? And if I have mis-identified the problem, or need to do some
> >further testing (hacking?), please advise what else I should do or
> >consider.
> I think you didn't get the slackware kernel going properly with debian.
> retry that (using lilo) and see if the cdrom trouble disappear. You may
> run into some incompatibilities, I don't know what that would be though.
>
> If you use modules, make sure you use slackware modules when running the
> slackware kernel. This is necessary if scsi/cdrom drivers are in modules.
> Using a slackware kernel with faulty debian modules would give you exactly
> the same problems.
>
> You may have to compile a "conglomerate" kernel, mostly 2.0.34 but with
> cdrom drivers from slackware. Or you may simply run with the slackware
> kernel for a while. Debian is much more than the kernel.
> If you use modules, consider replacing cdrom/scsi modules in debian with
> those from slackware - just copy them. Try out one at a time.
>
> You may also want to try the latest stable kernel (2.0.35) or even a
> development kernel. 2.1.125 is supposed to be very good, I use the most
> recent wich currently is 2.1.126. I have no trouble with that one
> either.
>
> Hopes this helps!
> Helge Hafting
============================================================================
Reply to: