Vincent Lefevre <[EMAIL PROTECTED]> wrote:
> On 2007-12-24 12:17:42 +0100, Joerg Schilling wrote:
> > Apple may or may not follow the standards here.......
>
> Since isoinfo detect errors (or do you mean that isoinfo may be
isoinfo is based in mkisofs technology.
The original Author of the Apple extensions did missinterpret the
way the Apple extensions are expected to fit into the ISO-9660 and Rock Ridge
framework.
Somebugs are fixed when the problem in the code is obvious and others are not
fixed before a testcase exists....
> buggy?), it seems that Apple does not follow the standards here
> (unless their interpretation is ambiguous). But the important
> point is that this can be detected, but it isn't detected when
> using mount + {ls or cp or whatever software that tries to read
> the file}.
Whether or not Apple follows the standard cannot be said before I see the
filesystem image or before you report results from a mount attempt on Solaris.
> > Your primary problem seems to be Linux kernel related and
>
> This is what I thought first. The bug is in the code that reads
> the RR index but doesn't check that it is valid. But I don't know
> if it is "mount" that reads the data and passes it to the kernel,
> or if it is the kernel that reads the data directly (according to
> Maximilian Attems, the bug is not in the kernel).
It is unrelated to "mount" but if the filesystem follows the standard, it is
a result of a bug in the Linux filesystem code.
> Now, if an error can be detected on the Linux side (and isoinfo, if
> not buggy, shows that this can be detected), the software should at
> least output a warning or an error message when the RR index is read.
Whether there is a bug or not in the filesystem produced by Apple depends
on the strucure inside the Rock Ridge part of the filesystem.
> I won't have the Linux machine in question in front of me until
> early January... There's another Linux machine here with a working
> CD drive, I'll try it.
OK
> > The old mkisofs code definitely did not handle Apple extensions
> > correctly.
>
> Could this explain the errors returned by isoinfo?
Yes
> > Unfortunately I am not sure whether I fixed all related problems
> > already. I know that the Solaris "hsfs" kernel filesystem driver
> > correctly ignores Apple extensions, so you may like to test the CD
> > on a recent Solaris version to verify whether there is a Linux
> > kernel problem.
>
> Unfortunately I don't have physical access to Solaris machines.
You don't need to, you may just scp a small filesystem image.
Well, you need to have root privileges or someone who helps.
It is also possible to run a life CD and check from this code.
> > I am interested in the filesystem image to check mkisofs and fix it
> > if needed. Could you please send a small example?
>
> This iso image contains private data, but I could try to build another
> one when I come back home.
Please do.
Jörg
--
EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin
[EMAIL PROTECTED] (uni)
[EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/
URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily
