[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: DVD+RW kernel patch - scsi error (help!)

> The drive
> that I have is an 'Optorite DD0203', based on a Sanyo engine,

Just a reminder. For any recording other than DVD+RW, dvd+rw-tools
5.12.x is a requirement for this [and similar] units.

> I have tried two types of media with the unit:  KHypermedia DVD+RW and
> now HP DVD+RW media.
> With both, my experience runs like this:
> 1 - dvd+rw-format /dev/scd0
> this generally works without error, returning the following in the logs:
> ...
> 2 - mkudffs --spartable=2 --media-type=cdrw /dev/scd0
> this generally works fine, returning something in the logs like:

None of the logged messages indicate an error condition.

> 3 - mount -o rw,noatime /dev/scd0 /mnt/cdrom
> this generally is where the error occurs -- but this is only visible in
> the logs (the command line returns without error).  Note that at this
> stage, it is very often it is this particular sector -- 9180032 -- with
> both types of media (KHypermedia and HP).
> Sep  2 15:51:42 www kernel: scsi0: ERROR on channel 0, id 0, lun 0, CDB:
> Request Sense 00 00 00 40 00
> Sep  2 15:51:42 www kernel: Current sd0b:00: sense key Medium Error
> Sep  2 15:51:42 www kernel: Additional sense indicates No seek complete
> Sep  2 15:51:42 www kernel:  I/O error:
> dev 0b:00, sector 9180032

It's utterly strange that you see the "request sense" CDB (command data
block)... You should see some other command such as read or write... One
possible explanation could be that kernel asks for too much sense data
(0x40=64 bytes as you can see) and unit "complains" that it doesn't have
that much... Please note that I'm not implying that if we figure out and
maybe work around this "why request sense cdb" issue, we solve the
problem. On the contrary it's most likely *not* directly relevant to the
"no seek complete" itself...

> 4 - Any subsequent attempt to copy to this media is futile, eventually
> leading to a 'No space left on device' error on the command line and
> something like the following in the logs:
> Sep  2 16:11:28 www kernel: sr0: injecting "SYNCHRONIZE CACHE"
> Sep  2 16:12:40 www last message repeated 2 times
> Sep  2 16:13:10 www kernel: scsi : aborting command due to timeout : pid
> 1608, scsi0, channel 0, id 0, lun 0 Synchronize Cache 00 00 00 00 00 00
> 00 00 00

So that SYNCHRONIZE CACHE just hung... One can argue that time-out value
might be insufficient, but I don't think so...

> Sep  2 16:13:10 www kernel: sr0: unable to "SYNCHRONIZE CACHE"
> [02/04/07]?
   ^^^^^^^^ Read "non-essential rant" at
http://fy.chalmers.se/~appro/linux/DVD+RW/ide-cd-2.5.69.+patch, which
discusses differences in behaviour between DVD-RW and DVD+RW recording
strategies implementations. Now the point is that there now are DVD+RW
units that behave in the very same manner as DVD-RW units (which was the
very reason why growisofs 5.10 exhibited poor DVD+recording performance
with Pioneer DVR-x06). The kernel patch does not expect that kind of
behavior and therefore I can't even think of recommendeding it for such
units. Note that I don't know if Optorite is such unit, but...

> When it DID work (with the KHypermedia DVD+RWs), it was to a live UDF
> system and it took 60-90 minutes to back up ~3.8 GB (~8 large files).

... 1 and a 1/2 hour for 3.8GB is hardly acceptable. And it's Linux UDF
implementation issue as was already implied in "Making UDF DVD+RW
backups" thread. I plan to overhaul kernel patch once I get DVD+MRW
capable unit and can have a second look at UDF by then, but not earlier.
Meanwhile I probably have to recommend to stick to growisofs, either as
front-end to mkisofs or as recording software for pre-mastered images.

> For a test, I did a 'growisofs -Z /dev/scd0=/dev/zero' (/dev/dvd doesn't
> exist on my system!)  This proceeded normally until:
> . . . (beyond 100%? interesting). . .
> 4873519104/4700372992 (103.7%) @2.4x, remaining 0:-52
> 4884660224/4700372992 (103.9%) @2.4x, remaining 0:-55

WOW! I'd classify this as a firmware deficiency. At least I've never
heard of *any* other DVD recorder which whould permit you to record to
the area meant for lead-out.

> sr0:  CDROM (ioctl) error, command: Write (10) 00 00 24 75 30 00 00 10 00
> Current sr00:00: sense key Hardware Error
> Additional sense indicates Tracking servo failure
                             ^^^^^^^^^^^^^^^^^^^^^^ Of course, the
groove has ended and optical pick-up can't detect reflected light

> Any suggestions?

Consider deploying growisofs instead. It might appear as unfortunate
advice, but it's practical. A.

Reply to: