[ This is a repost of the following article: ]
[ From: Karsten.Hoffmann@mbs-software.de ]
[ Subject: Problem with ISO9660/Rockridge-CDs ]
[ Newsgroups: qdn.public.neutrino ]
[ Message-ID: <aqntjf$diq$1@mbs-software.de> ]
Obviously the above mentioned group is quite dead …
Hello QSSL,
we’ve got some problems reading IS09660-CDs with Rockridge-Extension.
At least one thread of ‘devb-eide’ is running READY, if such an CD is
accessed. The CDs have been made with mkisofs 1.11.3 with a call like
this:
mkisofs -o /tmp/image.iso -L -l -R /cdroot
The same CD works fine with QNX4 (and other OSses).
Any ideas?
:Karsten.
P.S: We use Momentics 6.2.0 + patchA-pe
(+ fs-pkg-patch + phlocale-fix + shutdown-fix)
we’ve got some problems reading IS09660-CDs with Rockridge-Extension.
At least one thread of ‘devb-eide’ is running READY, if such an CD is
accessed. The CDs have been made with mkisofs 1.11.3 with a call like
mkisofs -o /tmp/image.iso -L -l -R /cdroot
I can’t reproduce this. Do you just mount the ISO file directly or do
you burn to CD? If the later, then can you mount the image file itself
(“mount -tcd /tmp/image.iso /cdroot2”)? Can you get the contents of
your “/cdroot” to the minimum which shows this problem and email me the
resulting “image.iso” file …
Karsten.Hoffmann@mbs-software.de > wrote:
we’ve got some problems reading IS09660-CDs with Rockridge-Extension.
At least one thread of ‘devb-eide’ is running READY, if such an CD is
accessed. The CDs have been made with mkisofs 1.11.3 with a call like
mkisofs -o /tmp/image.iso -L -l -R /cdroot
I can’t reproduce this. Do you just mount the ISO file directly or do
you burn to CD? If the later, then can you mount the image file itself
(“mount -tcd /tmp/image.iso /cdroot2”)? Can you get the contents of
your “/cdroot” to the minimum which shows this problem and email me the
resulting “image.iso” file …
I’m using a real, burned CD. A soon as I do a ‘ls /fs/cd0’ or so,
one devb-eide thread is running READY and the whole system slows down.
I also tried different samples and two different burners, but the
effect was always the same.
I tried your suggestion of mounted the file itself: that leads to a
blocking, unslayable ‘mount’ – and a READY-running devb-eide as well.
I’ll pack an archive with a small ISO-image, the ‘mkisofs’, I’m using
and the sample data and mail it to you directly.
I tried your suggestion of mounted the file itself: that leads to a
blocking, unslayable ‘mount’ – and a READY-running devb-eide as well.
Good, my remote office doesn’t have a CD burner.
I’ll pack an archive with a small ISO-image, the ‘mkisofs’, I’m using
and the sample data and mail it to you directly.
Got it, and can reproduce. Looks like an issue with the SUSP records
and a bug in stopping processing them, which I have fixed. However,
not sure if the original problem is with fs-cd.so or mkisofs … after
making the fix that image is not recognised as being RRIP (either by
QNX6 or QNX4) but if I use v1.11 or v1.15 of mkisofs (you provided
v1.10) to make an image with the contents you provided then the problem
does not happen and the image is mounted/recognised as RRIP.
I will continue looking into who is to blame for the format problem,
but in the meantime to keep you going you can: (a) pass the “norrip”
option when mounting the image to avoid the buggy code path; or (b)
upgrade to a newer version (say 1.15) of mkisofs, which appears to
construct “better” RRIP images …
Karsten.Hoffmann@mbs-software.de > wrote:
I’ll pack an archive with a small ISO-image, the ‘mkisofs’, I’m using
and the sample data and mail it to you directly.
I will continue looking into who is to blame for the format problem,
but in the meantime to keep you going you can: (a) pass the “norrip”
option when mounting the image to avoid the buggy code path; or (b)
upgrade to a newer version (say 1.15) of mkisofs, which appears to
construct “better” RRIP images …
OK, this is a bug in mkisofs v1.10. The TF record length is too big,
and it shadows/skips the CE (continuation entry) record which points to
the ER recognition block (and so the image is not recognised as RRIP).
You should upgrade your mkisofs. Now, of course, fs-cd should not run
READY on the malformed image, which I have fixed, but since there are
two workarounds I doubt this fix will make it into 6.2.1. Thanks, John.