SRP Sense Buffer Length (NO RESTRICTIONS IN PROTOCOLS)

Robert Snively rsnively at Brocade.COM
Wed Mar 13 13:02:12 PST 2002


* From the T10 Reflector (t10 at t10.org), posted by:
* Robert Snively <rsnively at brocade.com>
*
In skimming this, I thought you might like to know:

These are the requirements specified by FCP-2:

a)	9.1.2.5 requires additional CDB length to be a multiple of 4.

b)    5.6.2.12 and 9.2.2 require all data frames to 
	begin on a 4-byte boundary
	as specified by their relative offset (and the FCP_XFER_RDY
	parameters).  The last one need not end on a
	4-byte boundary, although in FC, pad bytes may
	need to be included at the frame level.  Implicitly,
	all sequences except the last need to be 0 modulo 4 in length.

c)	8.3 requires retries start on 4-byte boundaries.

A requirement NOT made by FCP-2 is that the sense information
be a multiple of 4-bytes.  There is no such requirement in
SPC-2, nor in FC-PLDA.  There is similarly no requirement that
lengths be counted by units of 4.  I could similarly find no
such limitation in SPI-4, since pad bytes are always available 
to control ending boundaries.

I think this requirement comes from some previous out-of-context
discussions and is not necessary or desirable for SRP.

Bob Snively                        e-mail:    rsnively at brocade.com
Brocade Communications Systems     phone:  408 487 8135
1745 Technology Drive
San Jose, CA 95110






> -----Original Message-----
> From: Patrick Fitzpatrick [mailto:PFitzpatrick at jni.com]
> Sent: Thursday, March 07, 2002 3:46 PM
> To: 'Edward A. Gardner'; t10 at t10.org
> Subject: RE: SRP Sense Buffer Length
> 
> 
> * From the T10 Reflector (t10 at t10.org), posted by:
> * Patrick Fitzpatrick <PFitzpatrick at JNI.com>
> *
> Hi Edward,
> 	thanks for the background. I am fine with the clarification. It
> would have been nice for the SRP doc to have stated that the 
> length should
> be rounded up and unspecified padded used. There is a possibility of
> confusion because of the rules for treating sense data when 
> it will not fit
> within the IU, i.e truncate. This led me to infer that if 
> sense data was not
> multiple of 4 bytes long that I should truncate the sense data. 
> 
> 
> Patrick
> 
> -----Original Message-----
> From: Edward A. Gardner [mailto:eag at ophidian.com]
> Sent: Thursday, March 07, 2002 3:41 PM
> To: Patrick Fitzpatrick; t10 at t10.org
> Subject: Re: SRP Sense Buffer Length
> 
> 
> > STP T10 Rev 10 Page 37 Line 30 states
> >SRP RSP IU: "The SENSE DATA LIST LENGTH field shall only 
> contain lengths
> >that are a multiple of 4."
> >
> >I am confused as to the reasoning for this because it seems 
> that most sense
> >data is 18 bytes long. There is nothing to say what should 
> be done is the
> >actual sense data is not a multiple of 4 bytes long. Should 
> the data be
> >trunacated down to the lowest length multiple of 4 or should 
> it be rounded
> >up. What happens to the (1 to 3) undefined bytes that are 
> now in the sense
> >buffer if rounding up is the solution.
> >
> >I do not see the need for this restriction, other than for 
> alignment, but
> as
> >sense data is at the end of the IU, alignment should not be an issue.
> 
> The SENSE DATA LIST LENGTH field specifies the length of the 
> SENSE DATA
> field, *not* the length of the sense data.  The sense data is
> self-describing, including its own length field.  When 18 
> bytes of sense
> data are returned, the SENSE DATA field contains 18 bytes of 
> sense data and
> two bytes of padding.  SRP doesn't specify the contents of 
> the padding.
> They are not part of the sense data.
> 
> FCP/FCP-2 and SPI-n IUs share this "feature" and use almost identical
> wording.  It arises from a requirement to keep FCP / SPI-n / 
> SRP IUs 32-bit
> aligned.  It is just unfortunate (and a historical accident) 
> that standard
> sense data is 18 bytes, which is not 32-bit aligned.
> 
> If you want to pursue this further, there is an SRP 
> conference call at 9 AM
> PST tomorrow (Friday Mar 8).
> 
> Edward A. Gardner               eag at ophidian.com
> Ophidian Designs                719 593-8866 voice
> 1262 Hofstead Terrace           719 593-8989 fax
> Colorado Springs, CO  80907     719 210-7200 cell
> 
> *
> * For T10 Reflector information, send a message with
> * 'info t10' (no quotes) in the message body to majordomo at t10.org
> 
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org




More information about the T10 mailing list