RE: iSCSI: two CmdSN gap questions

[Eddy Quicksall <eddy_quicksall@ivivity.com>]

For multiple connections, ABORT TASK should always have the I bit set.
Otherwise you can get into a dead lock.

Eddy

-----Original Message-----
From: Buck Landry [mailto:blandry@crossroads.com] 
Sent: Thursday, March 13, 2003 4:47 PM
To: ips@ece.cmu.edu
Subject: iSCSI: two CmdSN gap questions

I'll try to keep this short.

1) Is ABORT TASK the only Task Management Function Request type that can be
used to plug a CmdSN gap?  (I'd assume "yes")

2) I assume ABORT TASKs that plug a gap should be sent w/immediate bit set.
Otherwise, a target wouldn't consider the ABORT TASK for execution until
after the missing command itself (3.2.2.1 ".. the iSCSI target layer MUST
deliver the commands for execution in the order specified by CmdSN").  Do I
misunderstand this?

Thanks for any clarification,
--buck