On Fri, 2009-11-20 at 11:19 -0800, Joe Eykholt wrote:
> As there's been some discussion on the linux-scsi alias
> about making FC a "bus" that upper-level drivers can use,
> I thought I'd send out what I've been working on, as an
> RFC in order to collect comments on the concept.
> 
> It may not be complete in that other hooks may be needed for
> a full target implementation, but it is a start.
> 
> The first patch just rearranges code in the PRLI handling.
> The second patch is the bulk of the FC-4 hooks.
> The last patch adds a place in the lport for FC-4 providers
> to hang their private data.
> 
> I should probably rename FC-4 providers to FC-4 drivers or
> upper-level drivers.  Any suggestions?
> 
> I'm very interested in getting feedback.
> ---

Hey Joe... I'll take a better look at this next week, but I wanted you
to know what I have. I'm trying to build a fc_sysfs interface to build
the device tree as discussed.

Right now I have a very crude device tree API and I've been hooking it
into libfc. (Stuff like fcport_add(), fcfport_add(), etc...) I've moved
the scsi_host_add() to later in the stack and I'm now trying to move the
scsi_host_alloc(). This effort is mostly an exercise in breaking libfc's
dependency on the fc_host/fc_host_attrs and scsi_host objects.

I chose to focus on stack changes first before really flushing out the
fc_sysfs APIs since I want to have working code to test as I start
moving attributes out of fc_host/fc_host_attrs and into these other
devices.

I hope to publish something (early RFC) soon. Hopefully, after I move
the scsi_host_alloc() I can clean up the patches a bit so that they're
readable.

//Rob

_______________________________________________
devel mailing list
[email protected]
http://www.open-fcoe.org/mailman/listinfo/devel

Reply via email to