[vdr] How To use the new 160 GB hard disks under Linux.

[Carsten Koch <Carsten.Koch@icem.de>]

Hello friends of digital video recording,

we have discussed Maxtor's new 160 GB hard disks here in the past.
See http://linuxtv.org/mailinglists/vdr/2001/12-2001/msg00221.html
    http://linuxtv.org/mailinglists/vdr/2001/12-2001/msg00222.html
    http://linuxtv.org/mailinglists/vdr/2001/12-2001/msg00463.html
    http://linuxtv.org/mailinglists/vdr/2001/12-2001/msg00470.html
    http://linuxtv.org/mailinglists/vdr/2001/12-2001/msg00475.html
    http://linuxtv.org/mailinglists/vdr/2001/12-2001/msg00490.html
    http://linuxtv.org/mailinglists/vdr/2001/12-2001/msg00636.html
    http://linuxtv.org/mailinglists/vdr/2001/12-2001/msg00479.html
etc.

There were various concerns about Linux driver support for their
new 48-bit sector addressing and about booting from these disks.


Well, here is a success story:

I received one disk last week from Mindfactory, 
See http://www.mindfactory.de
Two days later I received another one from E-Bug, 
See http://www.e-bug.de



I found a patch that implements 48-bit-lba into Linux 2.4.16
at http://linuxdiskcert.org

This patch went into my 2.4.-16-xfs Kernel without any problems.
The xfs patch and the ide patch apparently do not collide, so
they can co-exist nicely (I am still 100% convinced that XFS is
a MUST for a vdr system).

I attached the new disk to my additional IDE controller PCI board
(Promise Ultra66) and ran badblocks. badblocks could write and read
all 160 GB successfully.
Of course, badblocks writes the same pattern to the whole disk,
so it cannot catch address errors. 
You may call me a bit paranoid, but I did not feel safe with the
badblocks utility, so I wrote my own test, which writes increasing
numbers to the disk and which also measures disk speed.
Sure enough, that program DID find errors.
I got in touch with the author of the patch. He suggested upgrading
to the latest controller BIOS. I upgraded to version 2.0, which still
is quite old. That did not help, my program still found errors. I
finally shuffled my other disks around, so I could attach the new
disk to the UDMA-100 controller on the (much newer) main board.
That worked. The main board's BIOS does not correctly recognize the
disk (it calls it a 135 GB disk), but still I can even boot Linux
from it! So, /dev/hda and /dev/hdc are 160 GB disks with XFS now:

/home/cko> df -hT
Filesystem    Type    Size  Used Avail Use% Mounted on
/dev/hda3      xfs    152G   70G   83G  46% /
/dev/hda1     ext2     15M  3.4M   11M  24% /boot
/dev/hdc1      xfs    153G   17G  136G  11% /video
/dev/hde1      xfs     93G   76G   18G  81% /hde
/dev/hdg1      xfs    114G   26G   88G  23% /tmp/oldvideo
shmfs          shm    251M     0  250M   0% /dev/shm


Over 300 GB should be enough for a while. 
Anyone interested to buy my old 100 GB and 120 GB disks?

Carsten.