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

Re: WRITE@LBA=3e0c30h failed with SK=3h/WRITE ERROR]: Input/output error (DL)



Hi,

> I would like to just try to burn one big file into a DVD+R DL (about
> 8Gb), what command using cdrskin would you recommend ?

In general it should work on DVD with the
same options as cdrecord accepts for DVD
and some options known from cdrecord with
CD.
Well, meanwhile you found out yourself:

> COMCD="-speed=$3 -v dev=2,0,0 -dao fs=1000m
> ...
> mkisofs $OPTS -V $1 $2|cdrskin $COMCD tsize="$SIZE"s -

This should be ok.


> Track 01: 4211 of 8138 MB written (fifo 100%) [buf  99%]   4.0x.cdrskin:
> FATAL : SCSI error on write(2156064,16): key=3 asc=0Ch ascq=00h

That's the same write error as with growisofs.
Only at a different sector address.
>From the specs:
  3 0C 00 WRITE ERROR
More is not explained by this code.

The address is suspiciously near the layer
break address. The error occurs at 
  2156064 * 2048 = 4415619072 bytes

Media capacity of both layers together is
reported as:
> > >  Free Blocks:           4173824*2KB

The error spot is 69152 blocks = 140 MB
_after_ the middle of this size. That is
much more than any buffer delay in the drive
could cause.


So i can see no systematic correlation between
layer break and error spot. (The layer break is
what mainly distinguishes DVD+R DL from all
other widely used DVD media.)

To me it looks like bad media. If only Nero would
be so nice to produce a misburn too ...

(The cdrecord error which you reported on May 9
happens quite exactly at half image size and issues
an error code 0 00 03 which is not listed in
MMC-5 specs. So that error is still suspicious
of systematic correlation with the layer break.)


> ata3: limiting SATA link speed to 1.5 Gbps
> ata3: hard resetting link
> ata3: COMRESET failed (errno=-16)
> ata3: reset failed, giving up
> ata3.00: disabled

I am not a kernel expert. 
Hard to say whether this documents the cause or
the consequences of the write error.


> Any other ideas ?

cdrskin announces the size to the drive in advance
if -dao is given or the size is known.
With -tao or with unknown size, it would start a write
run with unannounced size. (Here it deviates much
from cdrecord on DVD.)
So you could try this:

  COMCD="-speed=$3 -v dev=2,0,0 -tao fs=1000m
  ...
  mkisofs $OPTS -V $1 $2|cdrskin $COMCD -

But i do not really expect that this is the decisive
difference. If it works, i would suspect it to
be a lucky incident, as i suspect with Nero.

You may check whether my prediction is correct that the
next error will happen at a different sector address
or that some write runs even succeed completely.
(I believe your growisofs run nearly made it.)

You should verify that Nero really reproducibly can
write the full media without errors.
Verify whether the image file on disk is really identical
to the image on media:
  diff /dev/sr0 disk_file

If you can get different media, try them.


Have a nice day :)

Thomas


Reply to: