|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] RE: sector alignment for DataOut PDUs?It reffers explicitely (and with good reason) to unsolicited data ONLY. See the archives for a reason. Solicited bursts are what the target asks through R2T. Julo
>>>>> "Julian" == Julian Satran <Julian_Satran@il.ibm.com> writes: Julian> Where is the hint you are alluding to? >> .... (It does >> hint that targets may object if you don't do it that way -- >> which makes no sense at all unless there's a MUST for >> initiators to do what targets are allowed to expect.) In this section: 9.5 Unsolicited Data and Performance Unsolicited data on write are meant to reduce the effect of latency on throughput (no R2T is needed to start send- ing data). In addition, immediate data are meant to reduce the protocol overhead (both bandwidth and execu- tion time). However, negotiating an amount of unsolicited data for writes and sending less than the negotiated amount when the total data amount to be sent by a command is larger than the negotiated amount may negatively impact perfor- mance and may not be supported by all the targets. Specifically, the last sentence. That statement is not a good thing, because it explicitly permits failures to interoperate, which a protocol standard must never do. Instead, a receiver must always be required to accept anything that the standard permits the sender to send in a state that the sender can legitimately get to. There are two ways to fix this: one is to forbid the sender sending less than the negotiated unsolicited data length. The other is to require the receiver to accept unsolicited data of less than the negotiated length. Either change works because either change establishes interoperability. paul
Home Last updated: Thu Feb 28 16:18:05 2002 8942 messages in chronological order |