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

Re: dd bug or cdrom driver bug???



On 27 Jul 2000, Alan P. Kennedy, Sr wrote:

> First thing that I found out is that the i/o input error was correct. 
> The CDROM disk being used was damaged and the i/o input error was real.

Hmmm...is it possible that the disc in question is a rewritable?  I've run
into problems similar to this in the past with older CD-ROMs reading
rewritable disks (older lasers can't read them).

> I then got a good cdrom disk and found that the scsi cdrom worked
> like a charm with all the following commands:

That's good news :-)

> I then tried the same cdrom disk in the eide drive with the same
> commands (of course I changed /dev/scd0 to /dev/hda). 
> The eide cdrom produced an i/o input error and the following errors
> were in the /var/log/massages file.
> 
> ** Start error **
> Jul 27 19:04:35 alpha kernel: ATAPI device hda:
> Jul 27 19:04:35 alpha kernel:   Error: Illegal request -- (Sense key=0x05)
> Jul 27 19:04:35 alpha kernel:   Logical block address out of range -- (asc=0x21,
>  ascq=0x00)
> Jul 27 19:04:35 alpha kernel: hda: command error: status=0x51 { DriveReady SeekC
> omplete Error }
> Jul 27 19:04:35 alpha kernel: hda: command error: error=0x52
> Jul 27 19:04:35 alpha kernel: end_request: I/O error, dev 03:00 (hda), sector 17
> 3870
> Jul 27 19:04:35 alpha kernel: ATAPI device hda:
> Jul 27 19:04:35 alpha kernel:   Error: Illegal request -- (Sense key=0x05)
> Jul 27 19:04:35 alpha kernel:   Logical block address out of range -- (asc=0x21,
>  ascq=0x00)
> Jul 27 19:13:27 alpha kernel: set_program_attributes(12000000 d8c000 14000000 45
> d1d0)
> Jul 27 19:13:29 alpha kernel: <sc 0(84,1f4,11ffff940)><sc 53(8,1f4,11ffff940)><s
> c 0(17,1f4,11ffff940)><sc 53(8,1f4,11ffff940)>set_program_attributes(12000000 d8
> c000 14000000 45d1d0)
> Jul 27 19:27:36 alpha kernel: <sc 0(84,1f4,11ffff960)><sc 53(8,1f4,11ffff960)><s
> c 0(17,1f4,11ffff960)><sc 53(8,1f4,11ffff960)>set_program_attributes(120000000 2
> 3e000 140000000 56b50)
> ** End error **

Hmmm...I'll have to look at the kernel diffs a bit between 2.2.15 and
2.2.16 (which I use) since this error didn't pop up for me.  I know there
were some problems in the 2.4.0-test kernels WRT reading from CD-ROMs and
DVDs that may have been present in some of the 2.2.x kernels too (again,
need to verify this...just a guess for now).  Very odd.

> Now the interesting thing even with all the errors "dd" worked fine
> and when I compared all the images produced with "cmp" all the images
> were identical. The scsi iso images were identical to the eide cdrom
> images. I guess these errors really don't mean much. To be very frank
> I don't have a clue what these errors mean. Interesting enough even
> with a different cdrom disk (that is known good) I get the exact same
> errors. Note this is the same eide drive that would not work with
> cdparanoia. 

Yeah, I recognised the model right off as being a problematic drive that
I've run into in the past.  The weird thing is, while diagnosing this, I
realised that my old CD-ROM no longer worked, so I had to switch it with a
newer one that I had laying around :-P

> So I guess one should make sure that the cdrom disk is good in the
> first place. Especially before posting. :-)

No worries :-)

> So why did the i386 machine read the bad cdrom disk correctly? I
> suspect the reason is the i386 machine has a brand new scsi cdrom
> reader/writer that was able to read the damaged disk. That is of
> course a guess. If I have some time I may change cdrom drives and see.  

Again, newer CD-ROMs, in my understanding, have changed their laser
colour/type to enable them to read other types of discs (namely
rewritables burned in ISO9660 format), but older ones cannot.  If it was a
factory-burned disc, however, it shouldn't have had any problems at
all.

> Does anyone know what these hda cdrom errors mean?

I'll look into it.  Looks like it *might* have been trying to seek beyond
the end of the disc.  This was the problem, in my understanding, in the
2.4.0-test kernels until recently, so maybe it manifested itself earlier
as well.

C



Reply to: