RE: Question about Link Fault Signalling
As you recall, I thought we would eventually end up with an RS state machine
for this. Some with emotional attachment to a stateless RS will object. If
we choose to go to what you propose, then it must be in a state machine, as
many other proposals would have.
I must admit, I need to get up to speed better on the implication to the PCS
types. I don't like receive windows like you propose below. When they are
big, the tendancy is to want to create a rolling window which is a pain for
implementation. If the window for receiving the specified number of status
messages is arbitrary (an easier implementation), then it takes longer to
recognize a fault condition. I don't see the need for your 512 window, but
could think I might accept everything else.
I think we should probably lobby for a breakout dedicated to fault protocol
issues on Wednesday and at a minimum make sure we are all in sync on the
solution. A small group should then wordsmith a new 46.2.6 based on the
chosen protocol rules for approval by the larger group on Thursday. We then
have Friday morning to work the resolutions on dependent comments.
--Bob
-----Original Message-----
From: pat_thaler@xxxxxxxxxxx [mailto:pat_thaler@xxxxxxxxxxx]
Sent: Friday, January 05, 2001 1:27 PM
To: bob.grow@xxxxxxxxx; dgross@xxxxxxxxxxxxxxxxxx;
rtaborek@xxxxxxxxxxxxx
Cc: stds-802-3-hssg@xxxxxxxx
Subject: RE: Question about Link Fault Signalling
Bob,
I have problems with these two parts of the proposed text.
The 10GBASE-X PCS adds 10 states to its receive state diagram inorder to
restore the sending of Pulse ordered sets every other set frequency. This is
an unnecessary function. The Fault signalling would work fine with the
10GBASE-X PCS receiver just sending on the Pulse ordered sets as it receives
them - one per every 17 to 64 columns. Sending them more frequently doesn't
even speed up fault detection because it will they will be only sent every
other column after 3 sparsely place P's have been received.
Furthermore, the PHY doesn't necessarily deliver the messages to the RS
alternating with idles because of insertion and deletion of idles.
It is unnecessary to describe the PHY repeating of the fault message here
and the statement is not entirely accurate. A PHY detecting a local fault
doesn't "overwrite" RF messages. It has no idea whether RFs or anything else
is being received because it can't decode the input signal.
The description of detection needs to say the time period over which the
four status messages shall be received. Three status messages should be
enough protection on setting the fault condition.
Eight consecutive idles is too low a threshold for clearing the fault
condition even if clause 48 is left as it is because of idle insertion.
Also, it should require more than one missed status message to clear the
indication just as it took more than one message to set it.
Therefore, the text below should be changed to:
"A PHY indicates Local Fault conditions to the Reconciliation sublayer by
sending the corresponding status message at least once every 256 characters
on
RXC<3:0> and RXD<31:0>. The Reconciliation sublayer sends the Remote Fault
indication to the remote DTE by alternating the Remote Fault message with
Idle control characters on TXC<3:0> and TXD<31:0>.
"The Reconciliation sublayer continuously monitors RXC<3:0> and RXD<31:0>
for status messages. The reception of three status messages of the same type
without intervening status messages of another type within 512 characters
shall indicate that the corresponding fault condition has occurred. Once a
fault condition has been detected, it shall be cleared upon not receiving
fault messages of the same type for 1024 characters or upon statisfying the
detection condition for another status message type."
We also need to clarify whether transmission of MAC frames means stopping a
frame in progress or whether the RS waits until the end of the current
frame. If it is to terminate a frame in progress, it should send the /E/
before doing so.
I really think the details are getting to the point where a simple state
machine would be helpful to the reader.
Regards,
Pat
-----Original Message-----
From: Grow, Bob [mailto:bob.grow@xxxxxxxxx]
Subject: RE: Question about Link Fault Signalling
"A PHY indicates Local Fault conditions to the Reconciliation sublayer by
alternating the corresponding status message with Idle control characters on
RXC<3:0> and RXD<31:0>. The Reconciliation sublayer sends the Remote Fault
indication to the remote DTE by alternating the Remote Fault message with
Idle control characters on TXC<3:0> and TXD<31:0>.
"The PHY repeats a Remote Fault indication received from the remote DTE
unless a Local Fault condition is detected resulting in the PHY over writing
the received data with the Local Fault indication.
"The Reconciliation sublayer continuously monitors RXC<3:0> and RXD<31:0>
for status messages. The reception of four status messages of the same type
shall indicate that the corresponding fault condition has occurred. The
reception of four Idle control characters on successive RX_CLK edges (eight
consecutive Idle control characters) shall clear all fault conditions.