This message is from the T13 list server.

Pat,

Thanks for fixing the title. In fumbling with my Blackberry I replied to
the wrong e-mail.

RE: difficult to test - do you know anybody in the industry who actually
tests every optional feature and every possible value of every
configurable field in the SCSI standards? Finding devices that support
all the options would be quite a chore in itself. I don't think the
purpose of a standard is to define product testing requirements - only
to provide the baseline to define interoperability if/when optional
functionality is implemented. Just as there are a great many SCSI
initiators that don't work with anything but 512-byte blocks, I imagine
there will be a great many SCSI/ATA Translation products that only work
with ATA disks with 512-byte sector sizes. That's not a specification
problem, it's a purchase requisition problem. As you said, it's not easy
to find an ATA disk that's not 512-bytes, so I don't think we'll see
widespread problems with inadvertant use of other blocksizes in SAT
implementations that don't know what to do with it.

Furthermore, there's no edict that the blocksize presented from the SCSI
view has to match the sector size reported by the ATA device. I think
there's a need to leave room for value-add in the SAT layer at the
discretion of various implementers.

Bob 

-----Original Message-----
From: Pat LaVarre [mailto:[EMAIL PROTECTED] 
Sent: Wednesday, June 09, 2004 9:29 AM
To: Sheffield, Robert L
Cc: [EMAIL PROTECTED]
Subject: bytes per lba in SCSI to ATA

> Neither the SCSI standard nor ATAPI-7 restrict the blocksize to be 512
> bytes,

There is theory and there is practice.

I likewise now mis/remember hearing ATA/PI 7 changed vs. earlier work,
maybe specifically vs. ATA/PI 6, to try to make room for N * 512
bytes/LBA.  SCSI I know better:

The SCSI PDT x05 MMC = DVD/CD standard requires 2048 bytes/LBA, in
theory as in practice.

The SCSI PDT x0E/07/04/00 SBC/RBC = HDD/Flash standard theoretically
allows bytes/LBA other than 512, aye.  In practice, hosts go boom,
especially when bytes/LBA exceeds a physical page e.g. 4 KiB.

> there is a wide variety of applications that use other blocksizes. I

Aye.

> see no reason to place an artificial constraint on SCSI/ATA
> translation when it's a simple matter to forward the capacity reported
> by Identify Device (when it's valid).

Is it simple?

Not simple to test - we'll have to find such a device?

> I have no problem with reporting 512 bytes if the ATA device doesn't
> report the blocksize and if it isn't known by some other means.

I agree default to 512 isn't in dispute.

> ReplyTo: '[EMAIL PROTECTED]'
> Re: [t13] Web site ...
> Sent from my BlackBerry Handheld.

I first took the liberty of retitling.  Transmission failed.  I then
took the liberty of replying to [EMAIL PROTECTED] instead.  Transmission
failed.  I then took the liberty of crafting a from scratch message
containing only the body of this e-mail.  Maybe this third try will
work.

Pat LaVarre

Reply via email to