|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: StatSN wraparound; 2 typosTony, For the difference between regular modulo arithmetic and serial number arithmetic it would be helpful to have a glimpse at the referred RFC. We sue regular modulo arithmetic whenever we don't have to define an interval (order?) and serial where we have to. Thanks for the typos. Julo
According to 2.2.2.2 (draft 15), the status sequence number space is 32-bit unsigned integers and the arithmetic operations are the regular mod(2**32) arithmetic. Why is StatSN not specified to use serial number arithmetic (like CmdSN) to handle wrap-around? For example, I am working on a target implementation that keeps the information for response PDUs around until they are acknowledged by ExpStatSN from the initiator. If the target sends two response PDUs with StatSNs equal to 2**32-1 and 0, and the initiator's next PDU has ExpStatSN equal to 1, then the target should consider both response PDUs acknowledged. Clearly, this will not happen with "regular" unsigned integer comparisons, since 2**32-1 > 1. Perhaps it should be made explicit that comparisons on StatSN and ExpStatSN should use serial number arithmetic, which would handle the wrap-around. I have also found two typos in draft 15: Section 11.12 MaxRecvDataSegmentLength Last sentence "... if immediate data where sent" change "where" to "were". Section B.3 R2TSN DataSN use Examples Bidirectional DataSN Example "... in the alter case" change "alter" to "latter". Thanks, Anthony J. Battersby Cybernetics
Home Last updated: Fri Aug 09 10:18:53 2002 11585 messages in chronological order |