My understanding... The control-plane CPU will initially exchange control packets with the remote peer to determine proper discriminators and settle on negotiable session parameters. The negotiable session parameters include the interval timer and multiplier, which multiplied together create the detect timer. The CPU exchanges control packets with the peer until its own discriminator value is received in the peer's control packet, at which point the BFD session is established. In platforms that support HWO, the CPU can then "offload" the session to the NPU, which typically can scale to hundreds of BFD sessions with sub-second intervals (some NPUs can support single-digit millisecond interval timers).
It appears that %BFD_HWO-3-BFD_HWO_ERR_MSGS: % BFD_HWO: DETECT Timer Expired is an alarm from the NPU to the CPU that no control packet for this session was received before the detect timer (interval x multiplier) expired. The session would then be taken down, with the CPU attempting to re-establish. [The CPU generates syslog messages, not the NPU].
If there is chronic flapping of this BFD session, it could indicate that there is a high packet loss rate, high bit error rate, or that the interval timer is too short for the remote peer (I am assuming the local HWO device is operating within its BFD scale limits).
Disclaimers: I am long in CSCO. Bad answers are my own fault as they are not AI generated.
