Re: Logical Block Addresses in SCSI

[Bill Studenmund <wrstuden@wasabisystems.com>]

On Tue, 3 Sep 2002, shesha bhushan wrote:

> Hi All,
>
> I have mounted a remote directory and created a file. So this will generate
> SCSI WRITE CDBs that contain Logical Block Addresses (LBA) which is
> encapsulated in an ISCSI command PDU. How did SCSI at the local system
> generate a valid LBA for the remote disk?

This question is off-topic for this list. As others have said, a layer
above the iSCSI layer will have handed us the LBAs.

Is your question, "How did SCSI at the local system generate a[n] LBA" or
"... a valid LBA"?

The first question has to do with the file system. The file system knew it
had a given amount of space, and generated write operations (say wrote
blocks in the buffer cache) at specific offsets in its partition. It is a
very simple transform to go from the offsets of those blocks in the
partition to blocks on the disk (you add the partition offset). So writing
blocks in the buffer cache (or pages in the VM system, depends on kernel
design) readily turns into writing blocks on the disk. This question
really is about file systems and buffer caches.

How you get "valid" LBAs is either you have a partition map that says you
can access up to a given maximum, or you just try the command and see what
happens. This question really is about disk labels.

Take care,

Bill