Re: scsi hard disk not correctly showing up

Hi Theo!

thanks for the reply, I am honored. It took a moment to answer, because 
I had to physically power down a system, put in a "test" Linux disk and 
connect to it. I don't have other SCA-80 systems available beyond SPARC 
themselves and Linux (or probably, GRUB) doesn't like some of them.

I write a hort summary for those who might encounter such a disk.
Theo de Raadt wrote:
> There are SCA disks, particularily in the Hitachi / IBM chassis configuration
> which have a different sector size more than 512.  That is another possible
> problem.
>

Didn't know that, but now that you told me, searching with that IBM part 
number, lots of questions come about it. The official IBM FAQ is to put 
the Disk in the original system - which I don't have - and see what it 
says. But a hint got that Linux can at least see it and spit out the 
message with the unsupported size. Strange values like 520 or more can 
be seen,
apparently the extra byte per sector are used for RAID extra information.



[   22.641838] scsi target2:0:1: FAST-40 WIDE SCSI 80.0 MB/s ST (25 ns, 
offset 31)
[   22.649047] scsi 2:0:1:0: Direct-Access     IBM-SSG H01D073          
RSQP PQ: 0 ANSI: 3
[   22.649075] scsi target2:0:1: tagged command queuing enabled, command 
queue depth 16.
[   22.649113] scsi target2:0:1: Beginning Domain Validation
[   22.707639] scsi target2:0:1: Domain Validation skipping write tests
[   22.707657] scsi target2:0:1: Ending Domain Validation
[   27.492186] sd 2:0:1:0: Power-on or device reset occurred
[   27.492516] sd 2:0:0:0: [sda] Write Protect is off
[   27.492545] sd 2:0:0:0: [sda] Mode Sense: a7 00 10 08
[   27.495414] sd 2:0:1:0: [sdb] Unsupported sector size 524.
[   27.495453] sd 2:0:1:0: [sdb] 0 512-byte logical blocks: (0 B/0 B)
[   27.495471] sd 2:0:1:0: [sdb] 524-byte physical blocks
[   27.496564] sd 2:0:0:0: [sda] Write cache: disabled, read cache: 
enabled, supports DPO and FUA
[   27.514621] sd 2:0:1:0: [sdb] Write Protect is off
[   27.514719] sd 2:0:1:0: [sdb] Mode Sense: d3 00 10 08

"Unsupported sector size 524" says it all. It would be nice if OpenBSD 
could also spit out a similar warning message, but it is what it is.


I tried sg3-utils

  sg_format --format --size=512 /dev/sg2

     IBM-SSG   H01D073           RSQP   peripheral_type: disk [0x0]
Mode Sense (block descriptor) data, prior to changes:
   Number of blocks=137036592 [0x82b0330]
   Block size=524 [0x20c]

which confirms also the firmware is modified to 524 standard block size.

after the time for a cup of tea,

FORMAT UNIT Complete

At reboot, linux sees the disk as:
[   26.615042] sd 0:0:1:0: [sdb] 142368624 512-byte logical blocks: 
(72.9 GB/67.9 GiB)

which is for sure "progress"..
I could fdisk & create an ext4 file system on it.

How does OpenBSD do now?

sd1 at scsibus2 targ 1 lun 0: <IBM-SSG, H01D073, RSQP> 
t10.HITACHI_HUS103073FL3800_V3VLHBWA
sd1: 69515MB, 512 bytes/sector, 142368624 sectors

disklabel & newfs  /dev/rsd1a

spits out:

/dev/rsd1a: 69515.7MB in 142368084 sectors of 512 bytes

mount it and looks working! further tests next week.

A couple of GBytes seem to have disappeared, but it proves that these 
Hard Disks which can be gotten for cheap can be reformatted into 512 
bytes and at least this example is also very silent, many of these old 
SCA drives are noisy!

Cheers,

Riccardo