[lug] CDROM mounting woe

D. Stimits stimits at idcomm.com
Wed Oct 31 15:28:55 MST 2001


"qqq1one @yahoo.com" wrote:
> 
> I don't know if the following problem is related to
> the CDROM install failures I mentioned in my previous
> post
> (http://archive.lug.boulder.co.us/bymonth/2001.10/msg00968.html),
> but perhaps it is, hence the reference.
> 
> I've done 3-4 fresh installs of RH7.2, and haven't
> been able to mount the CDROM with any of them.  The
> CDROM still works with my 6.2 hard drive.  Here's the
> error I get:
> 
> # mount /dev/cdrom
> mount: /dev/cdrom is not a valid block device
> 
> Here's what fstab has (the floppy works):
> 
> # cat /etc/fstab
> <snip>
> /dev/cdrom              /mnt/cdrom
> iso9660 noauto,owner,kudzu,ro 0 0
> /dev/fd0                /mnt/floppy             auto
>  noauto,owner,kudzu 0 0
> 
> Here's what the devices look like:
> 
> # ls -al /dev/cdrom
> lrwxrwxrwx    1 root     root            8 Oct 29
> 23:28 /dev/cdrom -> /dev/hdd
> #
> # ls -al /dev/hdd
> brw-------    1 install  disk      22,  64 Aug 30
> 14:30 /dev/hdd
> 
> Here's where the problem seems to be first detected:
> 
> # cat /var/log/messages
> <snip>
> Oct 29 23:29:03 localhost login(pam_unix)[1023]:
> session opened for user install by LOGIN(uid=0)
> Oct 29 23:29:03 localhost  -- install[1023]: LOGIN ON
> tty1 BY install
> Oct 29 23:29:29 localhost su(pam_unix)[1081]: session
> opened for user root by install(uid=500)
> Oct 29 23:29:33 localhost kernel: ide-floppy driver
> 0.97
> Oct 29 23:29:33 localhost insmod:
> /lib/modules/2.4.7-10/kernel/drivers/ide/ide-cd.o:
> insmod ide-cd failed

Short note: If you expect to boot by CD or rescue, you're probably
better off compiling it in directly, versus putting it in an initial
ramdisk. The errors seem to imply that your kernel module is compiled
for a different kernel version. And are you using devfs? This could
greatly complicate life since it is currently a work in progress.


> Oct 29 23:29:33 localhost kernel: hdd: driver not
> present
> Oct 29 23:31:43 localhost su(pam_unix)[1081]: session
> closed for user root
> Oct 29 23:31:47 localhost kernel: ide-floppy driver
> 0.97
> Oct 29 23:31:47 localhost insmod:
> /lib/modules/2.4.7-10/kernel/drivers/ide/ide-cd.o:
> insmod ide-cd failed
> Oct 29 23:31:47 localhost kernel: hdd: driver not
> present
> [last 3 lines repeated about 30 more times plus other
> startup stuff plus more failed attempts to load
> ide-cd...]
> 
> Here's what the drivers/ide and drivers/block
> directories look like (I noticed that the ide-cd
> driver is now in the ide directory instead of in the
> block directory like it was with RH6.2):

You probably want to add a symbolic link in /dev/ that points to the
real device in /dev/ide/, just for compatibility.

> 
> # ls -al /lib/modules/2.4.7-10/kernel/drivers/ide
> total 160
> drwxr-xr-x    2 root     root         4096 Oct 29
> 05:26 .
> drwxr-xr-x   25 root     root         4096 Oct 29
> 05:26 ..
> -rw-r--r--    1 root     root         6532 Sep  6
> 15:35 ataraid.o
> -rw-r--r--    1 root     root         7444 Sep  6
> 15:35 hptraid.o
> -rw-r--r--    1 root     root        42588 Sep  6
> 15:35 ide-cd.o
> -rw-r--r--    1 root     root         6152 Sep  6
> 15:35 ide-cs.o
> -rw-r--r--    1 root     root        70884 Sep  6
> 15:35 ide-tape.o
> -rw-r--r--    1 root     root         7692 Sep  6
> 15:35 pdcraid.o
> #
> # ls -al /lib/modules/2.4.7-10/kernel/drivers/block
> total 212
> drwxr-xr-x    3 root     root         4096 Oct 29
> 05:26 .
> drwxr-xr-x   25 root     root         4096 Oct 29
> 05:26 ..
> -rw-r--r--    1 root     root        29084 Sep  6
> 15:34 cciss.o
> -rw-r--r--    1 root     root        28440 Sep  6
> 15:34 cpqarray.o
> -rw-r--r--    1 root     root        82428 Sep  6
> 15:34 DAC960.o
> -rw-r--r--    1 root     root        15748 Sep  6
> 15:34 loop.o
> -rw-r--r--    1 root     root        11324 Sep  6
> 15:34 nbd.o
> drwxr-xr-x    2 root     root         4096 Oct 29
> 05:26 paride
> -rw-r--r--    1 root     root        23504 Sep  6
> 15:34 xd.o
> 
> Here's what happens if I try to load the same cd
> driver module that works in RH6.2:

If you mean the same module name, it might work. If you mean the very
same module, you shouldn't mix versions like that. If your kernel has
this supported directly, not as a module, then you'll also get a failed
insmod or depmod (you can't insert a module if the service is already
provided as a non-module).

> 
> # /sbin/insmod
> /lib/modules/2.4.7-10/kernel/drivers/ide/ide-cd.o
> ide/ide-cd.o: unresolved symbol
> unregister_cdrom_R94f2f104
> ide/ide-cd.o: unresolved symbol
> cdrom_mode_sense_R2001c4a1
> ide/ide-cd.o: unresolved symbol
> cdrom_get_last_written_Ra27af6fe
> ide/ide-cd.o: unresolved symbol
> cdrom_mode_select_R1f88a106
> ide/ide-cd.o: unresolved symbol
> init_cdrom_command_Rfacde1b5
> ide/ide-cd.o: unresolved symbol cdrom_fops_R5849550a
> ide/ide-cd.o: unresolved symbol
> cdrom_number_of_slots_R72c4d34f
> ide/ide-cd.o: unresolved symbol
> register_cdrom_R8898dca8
> 
> I searched on "insmod ide-cd failed" at RedHat.com and
> none of the 4 results it came back with were for this
> problem.  I also tried variations on that search, but
> still couldn't find the same problem addressed.  Does
> anyone have an idea how to fix this problem?

I'd start by creating a symbolic link in /dev/ that is /dev/cdrom, which
points to the real device (which I think you said is hdd; presumably
/dev/ide/hdd). And I'd make sure your kernel config actually had support
via module before you worry about the module insert failing. If it is
requiring modules, be sure you have one compiled for this particular
kernel. And when it comes time to mount the CD, I'd probably give it
explicit iso9660 specification: mount -t iso9660 /dev/cdrom
/mnt/wherever.

D. Stimits, stimits at idcomm.com

> 
> __________________________________________________
> Do You Yahoo!?
> Make a great connection at Yahoo! Personals.
> http://personals.yahoo.com
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug



More information about the LUG mailing list