Valid SCAM Selection?
tdelacru at QNTM.COM
Wed Jun 18 11:11:54 PDT 1997
* From the SCSI Reflector (scsi at symbios.com), posted by:
* tdelacru at qntm.com (Tony DeLaCruz)
Would the reference to the I/O signal in SPI r15a section 10.3.1 under "To
perform a selection the initiator PIA shall:" "a) negate the I/O signal so that
this can be distinguished from reselection;" be used in the definition of SCAM
Tony De La Cruz
500 McCarthy Blvd.
Milpitas, CA 95035
e-mail: tdelacru at qntm.com
From: John Lohmeyer <John.Lohmeyer at Symbios.com>
Sent: Wednesday, June 18, 1997 11:47 AM
To: tdelacru at mail.qntm.com (Tony DeLaCruz)
Cc: SCSI Reflector <scsi at Symbios.com>
Subject: Re: Valid SCAM Selection?
The SPI Amendment #1 says the following about SCAM selections:
A device initiates the SCAM protocol by first winning bus arbitration, then
performing SCAM selection. The device may arbitrate using its current ID or
it may arbitrate without an ID. After winning arbitration the device has
the BSY and SEL signals asserted. It shall release all DATA BUS signals and
assert the MSG signal, then wait at least two deskew delays and release the
BSY signal. It shall maintain this pattern of the SEL and MSG signals
asserted with the BSY signal released for a minimum of a recommended SCAM
selection response time, then release the MSG signal. After releasing the
MSG signal the device shall wait, using wired-OR glitch filtering (see
table 1), until the MSG signal has been released by all other devices.
Level 2 SCAM initiators and SCAM targets that have not yet been assigned an
ID recognize SCAM selection if a pattern of the SEL and MSG true and the
BSY signal false is detected. After a variable delay, devices responding to
SCAM selection release the MSG signal, then wait, using wired-OR glitch
filtering, until the MSG signal has been released by all devices. SCAM
targets should release the MSG signal quickly, perhaps never asserting it
at all. SCAM initiators should wait a SCAM selection response time before
releasing the MSG signal."
I'm sure you've noticed that nothing is said about the I/O signal.
Therefore, the I/O signal should not be asserted during a SCAM selection
and the I/O signal should not be tested to detect a SCAM selection. As a
general rule in protocol descriptions (I'm not sure if we actually said
this in SPI), signals not mentioned should remain in their last state, in
this case, false.
It sounds to me like you've run into two implementations that are both
'flawed'. If either were fixed, it would work with the other
implementation. By 'flawed', I mean not optimal. I'm not sure that either
implementation actually violates the standards (although I think driving
I/O true should be illegal if it is not already!).
John Lohmeyer E-Mail: john.lohmeyer at symbios.com
Symbios Logic Inc. Voice: 719-533-7560
4420 ArrowsWest Dr. Fax: 719-533-7036
Colo Spgs, CO 80907-3444 SCSI BBS: 719-533-7950 300--28800 baud
* For SCSI Reflector information, send a message with
* 'info scsi' (no quotes) in the message body to majordomo at symbios.com
More information about the T10