SPL3R04 text in conflict with respect to Initiate Full Duplex Close

Craig Stoops Craig.Stoops at synopsys.com
Mon Oct 28 15:11:44 PDT 2013


Formatted message: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r1310282_f.htm">HTML-formatted message</a>

Hi George,
I believe two sections of text are in conflict in SPL3r04 regarding Initiate
Full Duplex Close. I believe it's a result of a hold over from the original
proposal in one section and an update to the other, and they got out of sync.
See the spec for the TF1 state, and look at the argument passed on transition
to TF2 when BOTH the PL is requesting a new frame transmission AND a begin
close message has been received. DONE (CLOSE) takes priority regardless of
whether the PL is requesting a new frame to be sent or not. Then in TF2, the
TF will transition to sending DONE regardless of whether there is a pending
frames transmission request because a transmit frame argument was not passed
|from TF1. (6.18.9.6.3.3 Transition SSP_TF2:Tx_Wait to SSP_TF4:Transmit_DONE)
The conflicting text left over from the original proposal is located in
6.18.8 Closing an SSP connection, where it states
In response to receiving an RRDY (CLOSE) or an EXTEND_CONNECTION (CLOSE) the
SSP link layer
requests the transmission of a DONE (CLOSE). The transmission of a DONE
(CLOSE) occurs:
a) if there are no SSP frames waiting to be transmitted by the SSP link layer
(see 6.18.9.6.3);
b) if there are no SSP frames being transmitted by the SSP link layer (see
6.18.9.6.4); and
c) ACK/NAK count is balanced (see 6.18.9.3).
In the above, item a) is in error, but b) is correct because the TF would be
in TF3 if a frame was being transmitted. In the original proposal, the TF
could transmit 1 more frame after the begin close request, but that is no
longer true.
Please confirm that my interpretation is correct and that this will be
removed.
Thanks,
Craig



More information about the T10 mailing list