Question on use of SEL during SCAM

[Chris Burns X3119]

* From the SCSI Reflector, posted by:
* chrisb at eng.adaptec.com (Chris Burns X3119)
*

I am trying to figure out whether or not a SCAM target
or SCAM-2 master is supposed to assert SEL in response
to SCAM selection by some other device.

In one document I've seen (Edward Gardner's Oct 11, 1993
memo proposing the SCAM protocol) both sections 3.2 and 4.3
state that both MSG and SEL should be asserted in responding
to SCAM selection. The relevant lines from section 4.3 read
as follows: "Level 2 SCAM master devices and SCAM slave devices
that have not yet been assigned an ID recognize SCAM selection
and assert SEL and MSG. After a variable delay (see below) they
release MSG, then wait until MSG has been released by all
devices, using wired-or glitch filtering. SCAM slave devices
should release MSG quickly, perhaps never asserting it at all.
SCAM master devices should wait a SCAM selection time before
releasing MSG." The section goes on to describe how all devices
should release SEL later in the SCAM selection process, then
reassert SEL at the very end of SCAM selection.

My trouble is, the document I'm using as an "official" spec
of the SCAM protocol is revision 15 of Annex B of the SCSI-3
Parallel Interface spec, and this doesn't explicitly talk about
SEL getting asserted when devices recognize a SCAM selection in
process. Section B.4.1 is the part of the document that addresses
SCAM selection. It looks to me like the statement that devices
should assert MSG and SEL upon recognizing SCAM selection was
inadvertently dropped.

Anyway, when the SCAM-2 master sees some other device initiate
SCAM, should it assert SEL right away, or not?

Thanks in advance to whoever can clear this up for me.

Chris Burns
Adaptec
408-945-8600