[802.3ae] RE: [802.3ae_Serial] RE: Clarification of local faults
"Local fault" in the management registers means locally detected. Therefore,
signal detect = FAIL should set the PMD receive local fault and PMD local
fault bits. PMD local fault should be the OR of PMD receive local fault and
PMD transmit local fault. Any time a device detects a fault, it could be at
the transmitter, in the path or in the receiver - local fault doesn't
indicate where the problem is. It indicates where the detection is.
I'm not sure what the answer to your items 3 and 4 should be. In one sense,
disabling the transmitter on purpose is not a fault. However, the link is
down and it is nice to be able to check one bit to determine whether the
link is up and then if it is down to start checking other bits to determine
why the link is down. I think the latter is a stronger argument.
The 802.3ae draft says about PMD_transmit_local_fault: "The faults detected
by this function are implementation specific." (52.4.8 ) and "If the MDIO is
implemented, and the PMD has detected a local fault on any transmit lane,
the PMD shall set the PMD_transmit_local_fault variable to ONE." (53.4.10).
Since 53 doesn't define what things are detected as a local fault and 52
clearly says it is implementation dependent, it is up to the implementer
whether disabling the transmitter sets the transmitter local fault. If
someone wants to change that, they will have to submit as sponsor ballot
comment.
There appears to be a potential infinite loop if PMD_transmit_local_fault is
defined as true due to the transmitter being disabled as 52.4.7 recommends
that PMD_transmit_disable_0 be asserted when PMD_transmit_local_fault is
detected. If the spec was changed to assert PMD_transmit_local_fault when
the transmitter is disabled than the text in 52.4.7 should probably modified
to say something like "If a PMD_transmit_local_fault (optional) is detected
for a reason other than disabling of the transmitter, then ...." to break
the loop.
Pat
-----Original Message-----
From: Lindsay, Tom [mailto:tlindsay@xxxxxxxxxxxxxxxxxxxx]
Sent: Wednesday, December 19, 2001 12:50 PM
To: 802. 3ae Serial PMD (E-mail); Stds-802-3-Hssg (E-mail)
Subject: [802.3ae_Serial] RE: Clarification of local faults
All - I am a little suprised there was no response to this. I (and Fibre
channel) would still appreciate some clarification on the scope of "local"
for the 3 local faults defined in 802.3ae.
Thanks, Tom
-----Original Message-----
From: Lindsay, Tom
Sent: Tuesday, December 11, 2001 1:19 PM
To: 802. 3ae Serial PMD (E-mail); Stds-802-3-Hssg (E-mail)
Subject: Clarification of local faults
In MDIO mappings, PMD local fault, PMD transmit local fault, and PMD receive
local fault functions are defined. See for example 52.4.6, .8, & .9.
Last week, Fibre channel debated the meaning of "local" without clear
conclusion. Should local mean there is something faulty strictly within the
PMD (bad component, etc.) or that the PMD is unable to successfully
communicate regardless of cause. With no intention of being exhaustive, here
are some examples:
1. Should signal detect = FAIL be considered PMD receive local fault?
2. Should signal detect = FAIL be considered PMD local fault?
Mechanisms behind signal detect = FAIL could include a disabled remote
transmitter, a broken cable, a bad photodiode, etc. All prevent successful
communication, but only the latter is a fault truly within the PMD.
3. Should assertion of transmit disable be considered PMD transmit local
fault?
4. Should assertion of transmit disable be considered PMD local fault?
Again, this will prevent communication, yet it is not due to a problem
within the PMD.
Thanks,
Tom Lindsay
Stratos NW
425/672-8035 x105