To: "Mignon Fernandez" <Mignon.M.Fernandez@bitmicro.com>
Cc: owner-t10@t10.org, t10@t10.org
Subject: Re: SAS: transport layer retries
From: George Penokie <gop@us.ibm.com>
Date: Thu, 27 Jul 2006 08:33:16 -0500
X-Message-Number: 7104
Formatted message: HTML-formatted message

Mignon,
The standard states: 
....the vendor-specific number of retries, if any, for the write DATA 
frame has not been reached.
and
If transport layer retries are enabled and the vendor-specific number of 
retries, if any, for the read DATA frame has not been reached,
That would suggest that even though retries are enabled you don't have to 
do the retries. But if the customer of your product enables data retries 
and you don't do it them waving the "if any" clause as your excuse for not 
have to do the retry they are expecting will not get you very far.
Bye for now,
George Penokie
Dept 9A8 030-3 A410
E-Mail:    gop@us.ibm.com
Internal:  553-5208
External: 507-253-5208
"Mignon Fernandez" <Mignon.M.Fernandez@bitmicro.com> 
Sent by: owner-t10@t10.org
07/27/2006 02:10 AM
To
<t10@t10.org>
cc
Subject
SAS: transport layer retries
Hello again,
If Transport Layer retries are enabled, and a link layer error (e.g. a NAK 
is received) occurs during a Data transfer, does a device, e.g. in SSP 
Target mode, have an option not to retry the data? Can the Target/device 
server terminate the command (send the status) without retrying? Will this 
action violate the specs? 
Thanks in advance,
Mignon Fernandez