> On Sun, 29 Jun 1997, Jonathan F. Freeson wrote:
>
> > Hello:
> >
> > I noticed that with some of the later 2.1.x kernels, I have been
> unable to
> > read some of my older "Shareware" CD-ROMs that I was able to read
> with
> > 2.0.30. With 2.1, I frequently got the kernel message:
>
> What are the kernel messages you get when you mount this disc?
>
> What type of cdrom drive are you using?
I get the exact same results with both of my CDRORM drives, one being a
SCSI, NEC Multipspin 3Xp, and the other being an IDE, CyberDrive 16x.
Below is a small excerpt of what messages I get. The exact same thing
happens when I use /dev/hdc instead of /dev/scd0. Am I doing anything
wrong?
[root@jonspc /mnt]# mount /dev/scd0 cdrom2
mount: block device /dev/scd0 is write-protected, mounting read-only
[root@jonspc /mnt]# cd cdrom2
[root@jonspc cdrom2]# ls
bwcc.dll* gif/ gocd.exe* readme.txt* viewers/
catalog.dir* gif2bmp.exe* gocdwin.exe* unzip.exe* vocwav/
eurodem/ gif2bmp.pif* mkgrp.exe* unzip.pif*
fli/ gocd.doc* mods/ utils/
[root@jonspc cdrom2]# cd mods
[root@jonspc mods]# ls z* (<---- I don't think you want to see
100+ error messages)
Directory entry extends past end of iso9660 block
ls: zine_sfx.zip: No such file or directory
Directory entry extends past end of iso9660 block
ls: zoltan.zip: No such file or directory
[root@jonspc mods]# mount
/dev/hda6 on / type ext2 (rw)
/dev/hda1 on /dosc type vfat (rw)
none on /proc type proc (rw)
/dev/scd0 on /mnt/cdrom2 type iso9660 (ro)
[root@jonspc mods]#
(Kernel Messages in /var/log/messages using Redhat 4.2)
Jul 4 14:07:34 jonspc kernel: Directory entry extends past end of
iso9660 block
Jul 4 14:07:34 jonspc kernel: Directory entry extends past end of
iso9660 block
>
>
> The theory behind the change is that cdrom discs have a fixed
> sector size - normally 2048 bytes, but 512 byte sector discs do come
> up
> from time to time. Directories themselves contain a series of
> directory
> entries, and the idea is that a single directory entry should never
> span
> a sector boundary.
>
> The fact that you are getting this message either indicates
> defective premastering of the disc itself, or it indicates that the
> driver
> for the cdrom drive you are using is somehow broken.
>
Based on your comments, I would think that the premastering process is
defective, since I am using two different drives on different interface
types. However, I had a chance to try this on another machine since I
don't have access to another Linux box. In DOS, I don't recall having
any problems on any other machine (I know this doesn't mean much).
> The main motivation for making the change now was that their
> is
> a lot of ugly overhead in piecing together two adjacent extents to
> form
> the directory entry that lies on a sector boundary. (This was
> originally
> added because the original cdrom drivers always gave back 1K sectors)
>
> Hope this explains things a bit.
>
> -Eric
I see. As a verification then, with my kudge of a patch as I understand
it, this extra overhead only occurs when I encounter one of these
defective disks. Otherwise, if I were to leave this "fix" in my kernel,
the only performance hit I would take with non-defective disks would be
the slightly larger code, and the execution of two additional IF
statements (if I recall the patch correctly). Otherwise, my system
should be stable (with regard to the isofs driver).
Just for reference, I have multiple CDROMs that produce this problem (I
estimate around 10). I purchased all of them from the same company. I
don't recall the exact name of the company, but here are the example
titles and other info that may be useful:
1. Mega Windows #3 --- (C) 1994 MEGA-ROM INC. (C) 1994 Profit Press
(First (C) is on the CDROM case, the other on the actual CD)
Best of Shareware, Sights & Sounds -- (C) 1994 Best of Productions
(Part of a 6 disc set)
The above error output was with the "Sights & Sounds" disk, there are
about 100 files in this directory that cause this problem.
Thanks.
---Jonathan Freeson
jfreeson@sun.science.wayne.edu