A list of issues regarding White Rabbit Specification v2.0
- Invalid timestamps
- Description: When master (i.e. WR switch working as Grandmaster) is adjusting it's clock the timestamps of Sync Messages are invalid. If used (the timestamps) by the slave (i.e. WR switch or WR Node), the synchronization is improper
- Current solution: If a timestamp is invalid, the Follow_UP message is not sent. Instead, the Sync Message is re-sent.
- Autoconfiguration of WR Network using BMC
- Description: We want to precisely control the configuration of a WRPTP network so that a misbehaving device (i.e.: WR node, WR switch) will not break the timing hierarchy of a system. This means that some ports of a boundary clock (i.e. WR Switch) shall never become master ports, while the other will never become slaves. In PTP, it is only possible to set slave-only mode in an ordinary clock.
- Current solution: Each port has a configuration flag (WR_SLAVE_ONLY, WR_MASTER_ONLY and WR_MASTER_AND_SLAVE). This flag restricts (in WRSpec.v2.0) WRPTP protocol only, so a port with configuration WR_SLAVE_ONLY will never become WR Slave but can become "standard PTP slave". Now, the WR_* flag affect also functioning of PTP and the modifiedBMC is used only if WR_MASTER_AND_SLAVE flag is set for a port. Otherwise, the modified BMC is skipped
- Single fiber for two ways communication vs two fibers
- Description: Can WR be used with two separate fibers for Rx and Tx ? Yes, but with performance degradation for links longer then few meters
- Change in WR spec: State explicitly that sub-ns accuracy can be achieved only with single fiber, however two fibers will work as well but with worse performance