Seagate Ultra 320 - Manual (Page 172)

158 Parallel SCSI Interface Product Manual, Rev. A
8.3.3.1 Protection code transmission
SCSI devices supporting this protection code transmit the protection code check data during all Command,
Message, and STATUS phases. The protection code byte is transferred on the upper eight bits of a wide bus
simultaneously with the information day byte on the lower eight bits of the bus using the same clock for the
transfer. Thus the transfer of the information byte and the protection code byte is performed exactly like a nor-
mal wide transfer. The check data is transmitted even if detection is not enabled.
8.3.3.2 Enabling protection code checking
A SCSI device enables protection code checking for an I_T nexus when it detects that valid protection code
data is being transmitted on the upper byte of the SCSI bus. The frequency that a SCSI device will try to enable
protection code checking and the number of valid protection code bytes required is vendor-specific. The follow-
ing are some possible times when a SCSI device could try to enable protection code checking:
a. During the first Command, Message, or STATUS phase after a power cycle, after a hard reset, after a
SCSI target port Reset message, or after a change in the transceiver mode.
b. Any time that removal and insertion of a SCSI device is possible, i.e. after a Unit Attention condition.
c. During the MESSAGE phases of a negotiation.
8.3.3.3 Disabling protection code checking
The removal and insertion of a SCSI device could require that protection code checking be disabled for a pre-
viously enabled I_T nexus. A SCSI device disables protection code checking when it detects that no protection
code data is being transmitted on the upper byte. The determination that no protection code data is being
transmitted is vendor-specific. The following are some possible ways that a SCSI device could determine that
no protection code data is being transmitted:
a. The DB(15-8) and DB(P1) signals are continuously deasserted while there is good parity on DB(7-0)
and DB(P_CRCA).
b. The protection code has a consistent error while there is good parity on DB(7-0) and DB(P_CRCA).
8.3.4 Parity
When protection code checking is enabled normal wide parity is used during a protected transfer of Command,
Message, or Status information. DB(P_CRCA) contains the parity for DB(7-0), and DB(P1) contains the parity
for DB(15-8).
8.3.5 Error handling
Protection code errors are handled exactly like parity errors during Command, Message, or STATUS phases
as defined in the relevant subclauses on exception condition handling in clause 11.1 of SPI-5.
Some examples of how API works are given in SPI-5 Annex M, subsection M.6.
8.4 Removal and insertion of SCSI devices (popularly known as “hot plugging”)
8.4.1 Removal and insertion of SCSI devices overview
This section defines the physical requirements for removal and insertion of SCSI devices on the SCSI bus. The
issues related to the logical configuration of the SCSI bus and characteristics of the SCSI devices when a
replacement occurs are beyond the scope of this standard. It should also be noted that the cases listed are dis-
tinguished for compatibility reasons and in most cases describe a system environment independent of this
standard.
Four cases are addressed. The cases are differentiated by the state of the SCSI bus when the removal or
insertion occurs.