Disconnect/MessageReject

Gerry Houlder Gerry_Houlder at notes.seagate.com
Mon Jun 5 06:24:13 PDT 1995


>What do you think the proper response for a target should be if
>an initiator issues a MESSAGE REJECT in response to a DISCONNECT
>message? 
>I should think that the target should simply stay connected as if
>it did not have the DISCONNECT privilege. Should it attempt 
>subsequent DISCONNECT messages during the current nexus? 

I agree that the target shouldn't disconnect -- it should continue processing 
the command as if the disconnect privilege had not been granted. No, I don't 
think the target should send subsequent disconnect messages during the current 
nexus. The initiator has made its desire pretty clear -- why should the target 
want to get hit by the 2 by 4 again?

You didn't say whether the command was an untagged or a tagged command. The 
above answer is definite for an untagged command, but there is wording in the 
standard saying that a target is not supposed to accept a tagged command that 
doesn't allow disconnection. The target is supposed to end such a command with 
BUSY status. Should a target end the command with BUSY status if an initiator 
says it allows disconnection but later rejects a disconnect message? This 
doesn't seem very appropriate, but if an initiator wants to be weird it should 
expect weirdness from its targets also. This is the only response that can be 
inferred from real wording in the standard today.




More information about the T10 mailing list