[TriLUG] Linux causes damage to LG CD-ROM drives (Dells) ?

Tanner Lovelace lovelace at wayfarer.org
Mon Oct 27 14:01:22 EST 2003


Jeremy Portzer wrote:
> On Fri, 2003-10-24 at 11:46, Brian Daniels wrote:
> 
>>On Fri, Oct 24, 2003 at 11:02:16AM -0400, Jeremy Portzer wrote:
>>
>>>Noticed something interesting on the Mandrake web site:
>>>http://www.mandrakelinux.com/en/errata.php3#badlg
>>>
>>>"Error scenario: Installing 9.2 and being told unable to install the
>>>base system and subsequent reboot reveals that CD-ROM is physically
>>>dead.
>>>"Why: According to LG Electronics, their ODD (Optical Disc Drive)
>>>products do not support Linux nor do they test with Linux.
>>>Unfortunately, many Dell computers (possibly others) come with these
>>>CD-ROM drives. 
>>
>>Very odd.  I have a LG cd burner (Model=HL-DT-ST GCE-8320B according to 
>>hdparm) in my box and have had no problems.
>>
>>My machine here at work has a HL-DT-STDVD-ROM GDR8161B LG DVD drive, again 
>>no problems.
>>
>>I'm using RedHat 9...I can't imagine what Mandrake is doing during the 
>>install to have that effect!
> 
> 
> Turns out it's a kernel patch that Mandrake applied that causes this
> problem: 
> http://archives.mandrakelinux.com/cooker/2003-10/msg02437.php

Actually, it's the fact that the LG engineers are apparently brain
dead that caused the problem.  Mandrake is only the first distro
to ship the patch that caused the problem.  Other than that, there's
nothing specific to Mandrake here.

Here's the problem.  In trying to determine if a given CD drive is
a reader or a writer the code in question (which adds packet writing
to linux) issues the "flush" command.  Normally this command is only
used for writers, telling the writer to flush the data out of the caches
and to the drive.  If a reader receives the command it's supposed to
do nothing or perhaps return an error indicating it's a reader, not
a writer.  Instead, for these drives, the LG engineers redefined that
particular instruction to mean "upload firmware".  So, what happens is
the kernel issues what it thinks is a flush command, expecting an
error if it's a reader (so it can distinguish between reader and writer).
The CD, however, gets what it thinks is an "upload firmware" command.
Since no firmware is forthcomming, the drive's firmware is effectively
overwritten with giberish and the drive is toast.  This is MOST DEFINITELY
a problem with the drives in question.  That said, however, the software
will have to be rewritten to take care of this brain damage.

So, the only Mandrake specific thing is that they were the unfortunate
ones to release the kernel first.  For those of you who are currently
saying Mandrake should test more I have this to say.  Mandrake *did*
test on these very drives.  Unfortunately, this error is only present
in certain versions of the firmware.  Saying that Mandrake should test
on every single firmware version of every sigle drive doesn't face
the reality of the resulting combinatoric explosion.  There's just simply
no way to test all those possible combinations.  This could have happened
to any other linux distributor, even Red Hat.  It's just unfortunate
that it had to happen at all.

Tanner
-- 
Tanner Lovelace | lovelace(at)wayfarer.org | http://wtl.wayfarer.org/
--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--
GPG Fingerprint = A66C 8660 924F 5F8C 71DA  BDD0 CE09 4F8E DE76 39D4
GPG Key can be found at http://wtl.wayfarer.org/lovelace.gpg.asc
--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--*--
    He who receives an idea from me, receives instruction himself
    without lessening mine; as he who lights his taper at mine,
    receives light without darkening me.  --  Thomas Jefferson
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 222 bytes
Desc: not available
URL: <http://www.trilug.org/pipermail/trilug/attachments/20031027/f09d7ceb/attachment.pgp>


More information about the TriLUG mailing list