FCP_RSP Residual Checking.

Robert Snively rsnively at Brocade.COM
Thu Jul 12 10:50:16 PDT 2001


* From the T10 Reflector (t10 at t10.org), posted by:
* Robert Snively <rsnively at brocade.com>
*
The intent of the statement was to indicate that commands
like TEST UNIT READY and other commands with a known zero
data length shall have the FCP_DL set to 0 and shall have
the FCP_RESID_UNDER/OVER also set to zero in the response.

However, if FCP_DL is non-zero, then FCP-2 indicates that
the FCP_RESID_UNDER/OVER should be set to reflect the
residue (if any) and the FCP_RESID should be set to the
correct value.

If neither FCP_RESID_UNDER nor FCP_RESID_OVER are set, then
FCP_RESID is not defined, according to FCP-2.

The FC-PLDA information is thus all properly contained in
FCP-2 with the traditional meaning.  However, FC-PLDA did not
have the advantage of the clarifying text of FCP-2, and 
in trying to clarify the issue itself, only confused the issue.

To your questions directly:

	"During commands that do not contain an FCP_DATA IU, 
	FCP_RESID_UNDER and FCP_RESID_OVER shall both be set 
	to 0, and the value of FCP_RESID is undefined."

1)  What is the intent of the above ? 

	See the intent above

2)  Does it cover for commands which may have got a CHECK CONDITION and
    performed no data transfer ?

	It covers commands with FCP_DL = 0 only.

3)  i.e. When a command requesting FCP_DL bytes of data xfer is 
    errored back by the target with a CHECK CONDITION and no 
    data transfer has occurred, should the target initialize 
    the FCP_RESID information to indicate that no data was 
    transferred , or is the FCP_RESID set to 0 ?

	Less than the expected number of transfers occurred and the
	residual value is set accordingly.

4)  If the behaviour is the latter, then, initiators cannot check FCP_RESID
    regardlesss of SCSI STATUS.

	Right.  That is why it is not done that way.

*
* 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