From mboxrd@z Thu Jan 1 00:00:00 1970 From: Brian King Subject: Re: [PATCH 0/6] libata: Support SATA attached via SAS Date: Wed, 07 Jun 2006 12:56:47 -0500 Message-ID: <4487135F.3050908@us.ibm.com> References: <4486FDBD.9030200@us.ibm.com> <4487084E.7090307@torque.net> Reply-To: brking@us.ibm.com Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4487084E.7090307@torque.net> Sender: linux-scsi-owner@vger.kernel.org To: dougg@torque.net Cc: jgarzik@pobox.com, "linux-ide@vger.kernel.org" , 'SCSI Mailing List' , James Bottomley List-Id: linux-ide@vger.kernel.org Douglas Gilbert wrote: > Brian King wrote: >> Jeff, >> >> Here is a refresh of my patchset to allow SAS HBAs to use libata to >> control attached SATA devices. It should apply cleanly to #upstream. >> >> James, >> >> After our discussion at the storage summit I think I have >> a better idea what you are looking for with a SATA transport class. >> Unfortunately, the problem I am facing is that I am dealing with a >> firmware stack (ipr) which does not expose the physical transport >> to the host. The ipr firmware does all the device discovery and reports >> a configuration table describing the attached devices, but does not >> export any physical topology information. At some point in the future >> I hope to be able to extract this information from the ipr firmware >> stack and convert ipr to be able to utilize the sas/sata transport class, >> but, unfortunately, the firmware interfaces to accomplish this >> aren't available at this time. > > Brian, > Does that mean that it doesn't have an SMP pass through? > SMP and the SAS addresses attached to the phys on the HBA > is enough to do discovery in the user space. Doug, That is correct. There is no SMP passthru interface. The interface I am working with basically gives me some very basic information regarding all attached end devices (which does not include SMP devices). The information I get for each device includes: 1. The LUN's SAS address 2. The type of device to aid in HLDD attachment. This could be either a SAS device, a SATA device, or a RAID logical device. 3. A bus/target/lun tuple which can be used to map into SPI if the host dd wants to. If you look in drivers/scsi/ipr.h at struct ipr_config_table_entry, that might give you a better idea as to what I get back for each device. In a future firmware release I should have the firmware interfaces I need to be able to extract both adapter SAS/SATA phy information and SAS expander information such that I can move ipr to fit into the sas transport model. Unfortunately, I don't have this today. Brian -- Brian King eServer Storage I/O IBM Linux Technology Center