|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: iSCSI: Unsolicited data PDU retryBill, I do not quite follow what "could lead to a mess"....let me address what I think is your question. I did not say that target "drives the data-out sequencing" for unsolicited data, only that the target drives the *error recovery* for all data-out (assuming it received the command, so has the ITT context). Let me attempt to rephrase. A digest error as an example, taken on one of the unsolicited data PDUs, would have to be recovered by the target sending a recovery R2T - IOW, target driving the Data-Out PDU recovery is true even for unsolicited data. If I understand you correctly, you seem to suggest that retries may carry different header flags (F-bit and no-F-bit) - it's illegal. Take a look at 5.2.1. I believe the protocol does work correctly even if the initiator retransmits Data-Out PDUs as it pleases, but that's not the design expectation. I believe that's all Tony's question was about. -- Mallikarjun Mallikarjun Chadalapaka Networked Storage Architecture Network Storage Solutions Hewlett-Packard MS 5668 Roseville CA 95747 cbm@rose.hp.com ----- Original Message ----- From: "Bill Studenmund" <wrstuden@wasabisystems.com> To: "Mallikarjun C." <cbm@rose.hp.com> Cc: <tonyb@cybernetics.com>; "Julian Satran" <Julian_Satran@il.ibm.com>; <ips@ece.cmu.edu> Sent: Wednesday, August 07, 2002 1:20 PM Subject: Re: Unsolicited data PDU retry > On Wed, 7 Aug 2002, Mallikarjun C. wrote: > > > Tony, > > > > In addition to what Julian said, let me add one more comment. > > > > Initiator is not expected to initiate any Data-Out recovery on its own - > > without being requested so via a recovery R2T by the target. The unsolicited > > data PDU retry is not advised. This follows from the design goal (e) > > in section 5.1.2 (top of page 80 in rev15) - only one side drives error > > recovery for a given class of PDUs, and target does it for Data-Out. > > > > Hope that clarifies. > > Mallikarjun, > > How can that be correct, though? > > Maybe I'm mis-understanding something (or thinking you're side-stepping > something you're not), but it seems like that could lead to a mess. > > First off, while you're right that the target drives data-out sequencing > with R2Ts, initial data is a little different. The initiator drives that, > subject to negotiation and spec rules. It decides if it wants to use > Immediate Data or unsolicited PDUs (assuming negotiations permit it). > > >From Julian's comment: > > > As for the data our assumption has been that target will drop data for > > non-instantiated tasks (not a known ITT). > > If we are retransmitting the command because we think it didn't get > received, then doesn't it seem sensible to conclude that the above text > triggered all of the unsolicited data getting dropped? > > Also, at a basic level, regardless of retransmission or not, if the > initiator (re)sends a command PDU that indicates it will use unsolicited > data (F bit not set in command PDU), then doesn't it always have to send > the unsolicited data it just promised?? > > I could see an initiator choosing to retransmit a command without > immediate and unsolicited data (no data in PDU and F-bit set) and that > normal R2Ts would take over from there. But we don't require it to do so > at present, and if the initiator chooses to indicate the presence of > unsolicited PDUs, then I'd expect it to have to send them. > > Take care, > > Bill > >
Home Last updated: Wed Aug 07 21:18:52 2002 11562 messages in chronological order |