Re: [PATCH 2.6.13 14/14] sas-class: SCSI Host glue

[Luben Tuikov <luben_tuikov@adaptec.com>]

On 09/11/05 05:40, Christoph Hellwig wrote:
> 
> Yes, absolutely.  This discussion is driving far off right now, no one
> is asking Adaptec to add support for competing products here, we're just
> asking to not declare the host_template in the common code, and supporting
> limited controllers is one of the reasons.

Hi Christoph,

I cannot make something to be, something that it is not.

That is, the SAS LLDD is not a "scsi host" and it will never be,
because it is just not a "scsi host".

What it is, is an access point to the transport, this is its
sole function and existance.  E.g. it doesn't know about max_luns,
etc, which are purely SCSI Core-into-scsi_host concepts.
It does know about Execute Command SCSI RPC and TMFs.

The "scsi_host" template is a SCSI Core concept, which
mixes a software component (what it is) and a hardware component
(what you're talking about).

Now, see the layering infrastructure: SATA support:
The sas_sata_host.c file would also declare a scsi_host, where
it will define its own queuecommand(), eh_timed_out(),
eh_strategy_handler(), etc. and do the _protocol_ part,
whereby the SAS LLDD does the transport part.

	Luben