Re: PR #40571 - NetBSD Atari install fails to prepare 16GB MTRON SSD on ACARD SCSI to IDE bridge

"David Ross" <[email protected]> Fri, 13 Feb 2009 21:32:33 -0800
Newsgroups gmane.os.netbsd.ports.atari
Message-ID <[email protected]>
Ok, David Brownlee helped diagose this and it was determined
that zero'ing the first MB of the disk and rebooting fixes this issue.  It
appears that for some reason the format operation performed by my Adaptec
SCSI BIOS leaves some junk on the disk.

It would be great if there were an advanced option during install to wipe
the disk and reload the disklabel (including doing whatever is necessary to
refresh the version of the disklabel the kernel caches).  Otherwise, the
install should be more forgiving if there is already junk on the disk -- the
current mode of failure isn't great.


I've noted this stuff in PR #40571.

I attempted a 5.0 RC2 install onto the MTRON/ACARD combo and hit the issue I 
had experienced back on NetBSD 1.6.1.  I've filed a new PR on this one:
http://www.netbsd.org/cgi-bin/query-pr-single.pl?number=40638
Screenshot: http://i43.tinypic.com/aev5ds.jpg

I'm OK with my Seagate drive, so this is not all that problematic for me. 
It would be nice to be able to use the MTRON SSD though.

David Ross
[email protected]


----- Original Message ----- 
From: "David Ross" <[email protected]>
To: "Julian Coleman" <[email protected]>; "NetBSD port-atari mailing list" 
<[email protected]>
Sent: Monday, February 09, 2009 9:35 AM
Subject: Re: PR #40571 - NetBSD Atari install fails to prepare 16GB MTRON 
SSD on ACARD SCSI to IDE bridge


>I can't be sure, but I believe it is just sending a SCSI format command.
>
> My Seagate is probably 10 yrs old at this point so it's probably not what 
> you'd consider modern.  It does appear to wipe the drive as it takes a 
> very long time for the format to complete (hours).
>
> The ACARD/MTRON returns from the format command very quickly, which would 
> normally indicate that there was no data re-write.  But on the other hand, 
> it is a very fast SSD, so it's a little unclear.
>
> In any case, David Brownlee is investigating.
>
> David Ross
> [email protected]
>
>
> ----- Original Message ----- 
> From: "Julian Coleman" <[email protected]>
> To: "NetBSD port-atari mailing list" <[email protected]>
> Sent: Monday, February 09, 2009 5:46 AM
> Subject: Re: PR #40571 - NetBSD Atari install fails to prepare 16GB MTRON 
> SSD on ACARD SCSI to IDE bridge
>
>
>> Hi,
>>
>>>                                                          Here's the
>>> relevant portion of the dmesg:
>>> http://i44.tinypic.com/ac5yl3.jpg
>>
>> This looks fine.
>>
>>> Here's the first part of the MTRON/ACARD disklabel:
>>> http://i42.tinypic.com/2lc4sgw.jpg
>>
>> Whereas this is wierd.  We must be overflowing somewhere in the disklabel
>> reading code.
>>
>>> So even though I zero'd the drive, the disklabel is wacky.  When I did 
>>> the
>>> zero'ing operation it completed super fast (a few seconds).  My 
>>> assumption
>>> was because the device is a solid state drive.  But I can imagine it's
>>> possible that the ACARD's implementation of the SCSI zero'ing command is
>>> "cheating" and not fully overwriting the disk with zeros.
>>
>> Is it sending a SCSI format command?  Most modern drives return quickly 
>> from
>> that as well and don't actually re-write data.
>>
>> Thanks,
>>
>> J
>>
>> -- 
>>  My other computer also runs NetBSD    /        Sailing at Newbiggin
>>        http://www.netbsd.org/        / 
>> http://www.newbigginsailingclub.org/
>>
>