Date: November 10, 1988 To: X3T9.2 Membership From: John Lohmeyer, X3T9.2 Chairman Subject: November 1988 SCSI-2 Working Group Meeting Minutes Bill Spence of Texas Instruments hosted a working group meeting November 7-9 in Austin, TX at the Stouffer Austin Hotel. The working group thanks Bill and Texas Instruments for hosting this productive meeting and for making copies of SCSI-2 Rev 6 for the attendees. The final agenda was: 1. Detailed Review of SCSI-2 Revision 6 in preparation for forwarding. 1.1 88-114 & 88-135 Comments on SCSI-2 [Tom Wicklund] 1.2 88-152 Save Data Pointer rules [John Lohmeyer] 1.3 88-103R1 Generic Error Recovery Page [Dan Davies] 1.4 88-122 CHANGE DEFINITION vs INQUIRY Inconsistency [Skip Jones] 1.5 Rotational Position Locking [Bob Mortensen] 1.6 88-138 Section 9 comments [Jim Semenak] 1.7 88-146 Helical Scan device considerations [Pete Bramhall] 1.8 Medium Changers [Paul Boulay] 1.9 Which appendixes should be retained? 1.10 Move shielded connectors to the body of the document? 1.11 88-143 Comments on SCSI-2 Rev. 5 [Bob Pentecost] 1.12 88-137 Reference to IT8 in appendix [W. Smythe] 1.13 88-149 Emulex comments on SCSI-2 document 1.14 88-121 Disconnect mode page, re-visited [Paul Boulay] 2. 88-142 & 88-144 Proposed changes to the single-ended terminator. [Christian Mollard and Paul Boulay] 3. 88-116 WDTR Disable Residue function [Jeff Stai] 4. TERMINATE IMMEDIATE and REPORT BLOCKS TRANSFERRED messages [Steve Goldman] 5. 88-124 Format Progress Reporting [Steve Cornaby] 6. 88-132 Parameter Pointer for LOG SENSE [Bob Pentecost] 7. 88-145 Clarification of Appendix H [Joe Lawlor] 8. 88-147 Transmission Line issues 9. 88-115 Autoconfiguration issues for SCSI-2 [Jim McGrath] 10. 88-114p3 Should a "detect B cable" function be added? [Tom Wicklund] SCSI-3 Items [Not Addressed by the Working Group] 50. 87-186 SEND DIAGNOSTIC Pages [Bill Spence] 51. 87-203 & 87-217 LOAD SKIP MASK command proposal [Greg Floryance, Dave McIntyre] 52. 87-206 More that 8 devices on wide SCSI [David Harms] 53. 88-002 Search Command modifications [Jeff Stai] 54. 88-007 Expanded RelAdr Bit Definition [Paul Boulay] 55. 88-69R1, 88-92, & 88-100 Autoconfiguration SSWG [Jerry Marazas, Paul Nitza, Jim McGrath, Paul Boulay] 56. 88-127 Error Handling Action Codes [George Penokie] The following people attended the meeting: Name Status Organization ------------------------------ ------ ------------------------------ Mr. Robert C. Herron A 3M Company Mr. Robert Plummer V Adaptec, Inc. Mr. Charles Brill P AMP, Inc. Mr. Bob Whiteman A AMP, Inc. Mr. Jerry Katzung O Apple Computer Mr. Matt Gulick O Apple Computer Mr. James J. Semenak P AT&T Mr. Joe Lawlor O AT&T Mr. Dan Davies P Cipher Data Products, Inc. Mr. Paul Hanmann A Emulex Corp. Ms. Jean Kodama S Emulex Corp. Mr. I. Dal Allan P ENDL Mr. Bill Duran O Exabyte Corp. Mr. Robert Liu P Fujitsu America, Inc. Mr. Bob Pentecost P Hewlett Packard Co. Mr. Kurt Chan A Hewlett Packard Co. Mr. Gerald Marazas A IBM Corp. Mr. David A. Buesing O IBM Corp. Mr. Giles Frazier V IBM Corp. Mr. Gerald Houlder A IMPRIMIS/CDC Mr. Rick Donihoo O Interphase Corp. Mr. Paul Boulay A LMS-OSD Mr. William Homans P LMS-TSD Mr. Robert Johnston S LMS-TSD Mr. Kim Ryal A Miniscribe Corp. Mr. John Lohmeyer P NCR Corp. Mr. James McGrath P Quantum Corp. Mr. Dexter Anderson P Scientific Micro Systems Mr. D. W. Spence P Texas Instruments Mr. Anil Gupta A Western Digital Status Key: P - Principal A - Alternate O - Observer S - Special Interest (Does not get mailings) V - Visitor The following documents were distributed at the meeting: Document Doc Date Author Description of Document ------------- -------- --------------- --------------------------------------- X3T9.2/88-138 10/20/88 J. Semenak SCSI-2 Revision 5, Section 9 comments X3T9.2/88-140 10/25/88 P. Michels Communications Device Model X3T9.2/88-143 B. Pentecost Comments on SCSI-2 Rev 5 X3T9.2/88-144 11/7/88 P. Boulay Discussion of the C. Mollard termination scheme X3T9.2/88-145 10/31/88 J. Lawlor Clarification to Appendix H. X3T9.2/88-146 10/24/88 P. Bramhall SCSI-2 Chapter 9 and DDS-Format Helical Scan Tape Drives X3T9.2/88-147 11/4/88 Maurer, Kodama, SCSI Bus Transmission Line Analysis & Muething X3T9.2/88-149 11/4/88 M. Aarons General Problems with Specification Implementation X3T9.2/88-150 11/7/88 K. Chan SCSI Single-ended physical layer testing proposal X3T9.2/88-151 E. Reyner More Single-ended Signal Simulation Results X3T9.2/88-152 11/8/88 J. Lohmeyer Proposed revision to the SAVE DATA POINTER message 1. Detailed Review of SCSI-2 Revision 6 in preparation for forwarding. An extensive review of SCSI-2 Rev 6 was conducted. Most of the non- controversial corrections were marked in the editor's copies of Rev. 6 and will be included in Rev. 7. Significant changes and those deemed as controversial are documented in these minutes or were added to the agenda for the San Diego meeting. Working Group recommendations are highlighted in Bold type. In addition to the following agenda items, the working group spent some time on Wednesday afternoon resolving [] notes in Rev 6. Committee members should review the [] note on page 5-8 concerning bit and byte ordering for wide data transfers. This item was added to the agenda for the plenary meeting in San Diego. 1.1 88-114 & 88-135 Comments on SCSI-2 [Tom Wicklund] A significant amount of time was devoted to reviewing these comments, with most accepted. One item was sufficiently technical to warrant assigning a separate agenda item (see agenda item 10). The issue of whether the soft RESET alternative should be retained in SCSI-2 was added to the San Diego meeting agenda. It was agreed that the EXTENDED IDENTIFY message should be deleted from SCSI-2 since it is not well documented and for most devices is not really necessary. The message code should be reserved with a note to indicate that it was used in SCSI-1. The 88-135 document was delegated to the document editors for consideration in that most of the comments are editorial. The technical comments were resolved as follows: a) CD-ROM devices should not include the READ LONG command. b) The WRITE SAME command should be added to Section 15, but not to 9. 1.2 88-152 Save Data Pointer rules [John Lohmeyer] The 88-120 document was accepted at the end of the Boston plenary meeting with only about 15 people remaining. It had the effect of adding a new paragraph of requirements to the SAVE DATA POINTER message. John Lohmeyer stated that the committee had acted too hastily in adding these new rules as requirements, but he was not opposed to adding them as recommendations. During the meeting, he drafted a revised SAVE DATA POINTER section (88-152) that casts these rules as recommendations. Some attendees objected to even recommending the rules as it would add unnecessary overhead. After even more discussion under agenda item 1.14, the working group agreed that these rules should be deleted under the SAVE DATA POINTER message and instead created a SDPC bit in the Device Disconnect-Reconnect Parameters page. This bit controls whether these rules are to be followed or not. The working group recommends that these changes be made in S2R7. 1.3 88-103R1 Generic Error Recovery Page [Dan Davies] The working group recommends that this proposal be accepted with some modifications to make it more generic. The page should be included in section 7 with an explanation of the individual bits. Each device type may include additional explanation of the device type's interpretation of the combinations of the bits. 1.4 88-122 CHANGE DEFINITION vs. INQUIRY Inconsistency [Skip Jones] The purported inconsistency does not exist with the VPD page that returns the current operating definition information (page code 81h). The working group recommends that no change be made. 1.5 Rotational Position Locking [Bob Mortensen] This section was significantly revised. Four changes identified by the working group were: a) If a target reports UNSYNCHRONIZED SPINDLE to one initiator, it must also generate a unit attention condition to the other initiators that may be on the system. b) Delete the whole paragraph with the [] note. c) Delete the last sentence in 8.3.5. d) Break the RPL code value 00b into two lines, one showing that 00b in the MODE SENSE command means RPL is not supported or is disabled. The other should show that 00b in the MODE SELECT command should be ignored. 1.6 88-138 Section 9 comments [Jim Semenak] Jim Semenak and Dan Davies met off-line to discuss this item and they reported back to the working group on their recommendations to solve the issues in 88-138. a) The document should be corrected to reflect that the REW bit affects the unbuffered mode and the SEW bit does not affect the unbuffered mode. b) If the SILI bit is one and a CHECK CONDITION status results for some other reason, then all REQUEST SENSE data should be valid, including the residue information. c) Jim Semenak will write an implementors note dealing with the possible need for the initiator to restore the mode parameters following a third- party release. 1.7 88-146 Helical Scan device considerations [Pete Bramhall] There was some concern expressed that the Sony/HP DDS is not yet complete and may change as a result of a meeting being held at the end of the month. Most of these changes/additions should be addressed in SCSI-3. The only changes recommended to be accepted from this document were items 1, 2, 3, (a variation of) 4, 5, and 6. 1.8 Medium Changers [Paul Boulay] Paul Boulay reviewed the changes made to section 16 in SCSI-2 Rev 6. Paul plans to add an explanation to the device model saying that this model is a primitive medium changer. Other (more sophisticated) models also exist but are not supported by this command set. The working group recommended that a 32-byte Volume Tag field be added to the storage element descriptor (16.2.5.3). They also asked that a means to translate the Volume Tag to an element address be added. The diagnostic translate address function was suggested as a prototype. 1.9 Which appendixes should be retained? Bill Spence volunteered to revise appendixes A through C to bring them up to date. The working group recommended Appendix E be deleted except for the information in E.1, which should be moved to section 4. The working group recommended that the editors be permitted to deal with Appendix F as necessary. Jim McGrath volunteered to revise Appendix G to bring it up to date. Appendix H needs to be revised to include information concerning AEN and to delete section H.5. The working group recommended that Appendix I be deleted. The working group recommended that Appendix J be retained. 1.10 Move shielded connectors to the body of the document? The working group recommended that the shielded connectors in Appendix D, except the EUROCARD Boxes (which should be deleted), be moved to section 4. 1.11 88-143 Comments on SCSI-2 Rev. 5 [Bob Pentecost] Bob Pentecost's comments were reviewed and appropriate editorial changes were noted by the section editors for inclusion in S2R7. 1.12 88-137 Reference to IT8 in appendix [W. Smythe] The working group recommended that the editors add an appropriate reference to the IT8 project somewhere in the document. Possible locations are the foreword, section 1 or 2, or an appendix. 1.13 88-149 Emulex comments on SCSI-2 document Some of these comments were either already addressed or rendered obsolete in Rev 6. Others were editorial and were marked in the editor's copies of SCSI-2 Rev 6 for inclusion in Rev 7. Still others were considered rhetorical in that comments like, "MODE SELECT/SENSE is hard to read" may be true, but little advise was offered on how to improve it. 1.14 88-121 Disconnect mode page, re-visited [Paul Boulay] Document 88-121 was accepted at the Boston plenary meeting and was included in S2R6. Subsequent to the Boston meeting, Dave Filpus suggested that the DDTD bit be expanded into a two-bit field, giving more control over when the target is permitted to disconnect. Paul Boulay presented the suggested change to the working group, which after several refinements was accepted by the working group. While discussing this issue, the working group also found a way to deal with agenda item 1.2, by creating an SDPC bit, which controls whether the SAVE DATA POINTER message must be sent prior to the final DISCONNECT message and prior to a RESTORE POINTERS message to restore the status pointer. The working group recommended that these changes be made to the Device Disconnect-Reconnect Parameters page. The working group recommended that implementors note under the DISCONNECT message be revised to make it a requirement as follows: "Targets which break data transfers into multiple connections shall end each successful connection (except possibly the last) with a SAVE DATA POINTER - DISCONNECT message sequence." 2. 88-142 & 88-144 Proposed changes to the single-ended terminator. [Christian Mollard and Paul Boulay] (See agenda item 8, below.) 3. 88-116 WDTR Disable Residue function [Jeff Stai] This item was deferred to the San Diego meeting so that Jeff can present his own proposal. 4. TERMINATE IMMEDIATE and REPORT BLOCKS TRANSFERRED messages [Steve Goldman] There was no work on this agenda item since Steve Goldman was not present. 5. 88-124 Format Progress Reporting [Steve Cornaby] There was no work on this agenda item since Steve Cornaby was not present. 6. 88-132 Parameter Pointer for LOG SENSE [Bob Pentecost] This was included in Rev 6. 7. 88-145 Clarification of Appendix H [Joe Lawlor] Part of this proposal was deemed editorial and will be included in Rev 7. Also proposed is that a power-on to selection time be defined which is a recommended maximum time that a device take from power application to being able to accept a selection. There was much discussion over what is meant by being able to accept a selection, what time should be recommended, and whether this should be in section 4 or appendix H. The working group also suggested that a RESET to selection time also be defined (250 ms). Joe Lawlor accepted an action item to revise his proposal for the San Diego meeting, including the working group's suggestions. 8. 88-147 Transmission Line issues Jean Kodama of Emulex gave a presentation on their SPICE simulations and bench verifications of single-ended SCSI signals (88-147). There are several preliminary recommendations included in this document. The working group recommends that these preliminary recommendations be included in S2R7. The statement saying that fast SCSI has not been tested with single-ended systems has managed to slip to the end of Appendix J. A stronger statement should be included in section 4 saying that initial testing indicates fast SCSI does not work with single-ended systems. Bob Whiteman presented further single-ended simulation results with close clustering and device capacitance (assumed 25 pF.) added (88-151). These results showed some serious problems, even with cable impedances of 105 ohms. Kurt Chan discussed his document (88-150) on a plan for testing the physical layer of SCSI. Paul Boulay presented his document (88-144) which comments on the proposal from Christian Mollard (88-142). He suggests some improvements that avoid drawing excess current under worst case conditions and that avoid changes in the high level set point depending on where terminator power is sourced. Paul recommends that we consider changing the terminator specification to the actual requirements rather than requiring the 220/330 ohm circuit currently specified. This would improve the single-ended signal quality, but is not necessarily a complete solution. Paul accepted an action item to develop such a proposal. 9. 88-115 Autoconfiguration issues for SCSI-2 [Jim McGrath] Some of this proposal is related to item 7, above. The working group recommends that the proposal be accepted, except that only the SEL signal be re-defined as OR-tied. 10. 88-114p3 Should a "detect B cable" function be added? [Tom Wicklund] The working group recommends that a "detect B cable" function be added. The direction is for devices capable of wide data transfers to drive the REQB signal during any message phase if it is in the target role and to drive the ACKB signal during any message phase if it is in the initiator role. During wide negotiation, the device should negotiate for 8-bit operation if it detects that the other device's signal is false. John Lohmeyer accepted an action item to develop a proposal on this. None of the following agenda items were addressed: 50. 87-186 SEND DIAGNOSTIC Pages [Bill Spence] 51. 87-203 & 87-217 LOAD SKIP MASK command proposal [Greg Floryance, Dave McIntyre] 52. 87-206 More that 8 devices on wide SCSI [David Harms] 53. 88-002 Search Command modifications [Jeff Stai] 54. 88-007 Expanded RelAdr Bit Definition [Paul Boulay] 55. 88-69R1, 88-92, & 88-100 Autoconfiguration SSWG [Jerry Marazas, Paul Nitza, Jim McGrath, Paul Boulay] 56. 88-127 Error Handling Action Codes [George Penokie]