RE: RE: iwarp-tech Conformance Question
"Culley, Paul" <[email protected]> Sun, 14 Nov 2004 18:46:28 -0600
| Newsgroups | gmane.ietf.rddp |
|---|---|
| Message-ID | <4D027986353D1341ADA4F2A4F8A170762F08A689@cceexc18.americas.cpqcorp.net> |
If the requestor has set his to 1, then it does not matter what the reply frame has in it, CRCs must be generated by both ends, and to be compliant, must also be checked by both ends. I suppose that a non-compliant implementation could not check, or could tear down in this case, but then it is not an IETF RNIC. Paul R. Culley HP Fellow 281-514-5543 > -----Original Message----- > From: [email protected] [mailto:[email protected]] On > Behalf Of Carrier, John > Sent: Saturday, November 13, 2004 8:18 AM > To: Barry Reinhold > Cc: [email protected]; RDDP > Subject: [rddp] RE: iwarp-tech Conformance Question > > Hi Barry, > > my understanding is that CRCs are required unless both sides > set the CRC bit to '0'. If the reply frame from the > responder has the C bit set to zero after the request frame > had it set to one, then CRCs are required. The responder may > not be using them on receive, but it has to accept them from > the initiator and it has to generate them on transmit > > --jc > > > -----Original Message----- > > From: [email protected] > > [mailto:[email protected]]On Behalf Of Barry Reinhold > > Sent: Friday, November 12, 2004 2:43 PM > > To: RDDP > > Cc: [email protected] > > Subject: iwarp-tech Conformance Question > > > > > > Working on an interoperability issue related to section > > 3.6.1.1 (CRC bit > > paragraph) of MPA, which states: > > > > This bit (CRC bit) declares an endpoint's preferred CRC usage. When > > this field is '0' in the "MPA Request Frame" and the "MPA Reply > > Frame", CRCs MUST not be checked and need not be generated > by either > > endpoint. When this bit is '1' in either the "MPA Request Frame" or > > "MPA Reply Frame", CRCs MUST be generated and checked by both > > endpoints. > > > > The question: > > > > If an MPA request frame has set the CRC bit to 1, can the > MPA replay > > frame's CRC bit be set to 0 without violating the draft. > > > > >From a strict reading of the draft there does not seem to be a > > requirement that a "1" in the request results in a "1" in > the reply, > > only that both devices must use CRCs. > > > > So is the meaning of a "0" in the reply frame "I'm not > using CRCs, or > > is it I didn't want to use CRCs". Further if the requester > sees a 0 in > > the reply frame after setting the CRC bit to 1 in the request frame > > should it reset the connection? > > > > Barry Reinhold > > Lamprey Networks > > [email protected] > > (603) 868-8411 > > > > > > > > > > > > _______________________________________________ > rddp mailing list > [email protected] > https://www1.ietf.org/mailman/listinfo/rddp >