Soft Errors

The 802.5 Soft Error Types

• Burst Error
• Line Error
• Lost Frame Error
• Token Error
• Internal Error
• Frequency Error
• AC Error
• FC Error
• Abort Delimiter
• Receive Congestion

Analysis Of Soft Error Reports

When you examine a Protocol Analyzer Decode of a Soft Error Report frame you will see one or more of the following errors being reported. When an error occurs, the Token-Ring adapter begins what, by default, is a 2 second timer. During that 2 seconds the adapter keeps track of all other errors that it may observe. At the end of the 2 second period the adapter sends a Soft Error Report frame to the functional (group) destination address for the Ring Error Monitor. There may, or may not, be an Error Monitor on the ring to actually record or take action on the Soft Error Report.

Note: The numbers (like 3.8.3) at the beginning of each description refer to the section numbering in the IEEE 802.5 standards document.

The 802.5 spec defines a Burst Error as:

3.8.3 Burst Error. This counter is incremented when a station detects the absence of transitions for five half-bit times (burst-five error). Note that only one station detects a burst-five error because the first station to detect it converts it to a burst-four.

Five half-bit times implies that three bits were detected with no clock in the middle.
A Burst Error Is The Result Of Noise On The Line

• This could be environmental background noise
• This could be noise due to a single point of noise injection into the cable
• This could be noise as a result of faulty hardware

The 802.5 spec defines a Line Error as:

3.8.1 Line Error. This counter is incremented when a frame or token is copied or repeated by a station, the E bit is zero in the frame or token and one of the following conditions exists:

• There is a non-data bit (j or K bit ) between the SD and the ED of the frame or token.
• There is an FCS error in a frame.

The first station detecting a line error increments its appropriate error counter and sets E=1 in the Ed of the frame; this prevents other stations from also logging the error and isolates the source of the disturbance to the proper error domain.

"A Line Error Is a Baby Burst Error That Hasn't Grown Up"

Expect a ratio of roughly 1 Line Error for every 10 Burst Errors

The 802.5 spec defines a Lost Frame Error as:

3.8.6 Lost Frame Error (LOST_FR). This counter is incremented when a station is transmitting and its TRR timer expires. This counts how often frames transmitted by a particular station fail to return to it (thus causing the active monitor to issue a new token).

Noise can cause lost frames.

The 802.5 spec defines a Token Error as:

3.8.10 Token Error. This counter is incremented when a station acting as the active monitor recognized an error condition that needs a token transmitted. This occurs when the TVX timer expires (see transition (03) of the Active Monitor FSM).

Noise can corrupt tokens.

The 802.5 spec defines an Internal Error as:

3.8.2 Internal Error. This counter is incremented when a station recognizes a recoverable internal error. This can be used for detecting a station in marginal operating condition.

A parity error was detected during DMA (Direct Memory Access) between the adapter and the attached system. Like any other parity check error that might be encountered the machine should be powered down and restarted. If the error returns, move the adapter. If the error moves with the adapter then the problem is in the on-board adapter RAM. If the error stays with the host system then the problem is with the system-board memory.

The 802.5 spec defines a Frequency Error as:

3.8.9 Frequency Error (FREQ). This counter is incremented when the frequency of the incoming signal differs by more than that specified in Section 7 form the expected frequency.

The Active Monitor is responsible for maintaining a constant clock signal on the ring. If a Standby Monitor station detects faulty clock then a Frequency Error is generated.

Make a different station into the Active Monitor. If the problem goes away then the original Active Monitor adapter is defective.

The 802.5 spec defines an AC Error as:

3.8.4 AC Error. This counter is incremented when a station receives an AMP or SMP frame in which A and C are both equal to 0, and then receives another SMP frame with A and C are both Equal to 0 without first receiving an AMP frame.

Someone got more than one AMP or SMP with address recognized and frame copied equal to zero in the same ring-poll cycle. This means that the station immediately upstream from the reporter didn’t properly set the ARI and FCI bits.

The 802.5 spec defines an FC Error as:

3.8.8 Frame Copied Error (FR_COPIED). This counter is incremented when a station recognizes a MAC frame addressed to its specific address and detects that the FS field A bits are set to 1 indicating a possible line hit or duplicate address.

Remembering The Difference Between AC and FC Errors:

..hmm... one of these errors is the fault of the immediate upstream neighbor failing to set the ARI/FCI bits. One of these is the result of a duplicate address setting the bits.

• The ‘immediate upstream neighbor’ situation is ISOLATING. The address problem is NON-ISOLATING.
• ‘A’ comes before ‘F’ in the alphabet; as in ‘AC’ error and ‘FC’ error.
• ‘I’ comes before ‘N’ in the alphabet; as in ‘Isolating’ and ‘Non-Isolating’.
• The AC Error is Isolating; the FC Error is Non-Isolating.
• The Isolating error points to the immediate upstream neighbor; the AC Error points to the immediate upstream neighbor as being the one who didn’t set the bits properly in the AMP or SMP frame.
• The Non-Isolating error is the result of the Duplicate Address situation; The FC error.

The 802.5 spec defines an Abort Delimiter transmitted error as:

3.8.5 Abort Delimiter Transmitted (AD_TRANS). This counter is incremented when a station transmits an abort delimiter while transmitting.

After a station converts the SDEL and AC bytes of a token into a frame the receive state machine expects to see the EDEL in the token followed by idle clock until the SDEL of the new frame comes around the ring to be stripped.

If, during transmission, the station detects either an invalid ending delimiter, a Claim Token, or a Beacon frame the transmission stops and the partial frame is ended with an Abort Delimiter.

The 802.5 spec defines a Receive Congestion error as:

3.8.7 Receive Congestion Error (RCV_CON. This counter is incremented when a station recognizes a frame addressed to its specific address, but has no available buffer space indicating the station is congested.