Date: September 3, 1987 Revised September 7, 1987 To: X3T9.2 Membership From: Lawrence J. Lamers, General Editor, X3T9.2 011-813-458-7870 Revision 1 by John Lohmeyer based on Jeff Stai's comments Subj: SCSI-2 Ad-Hoc Minutes - Aug 31, 1987 to Sept 02, 1987 - Wichita KS A SCSI-2 working group meeting was held Aug 31 through Sept 2 in Wichita, KS. Meeting agenda 1. Connector Issues 2. Command Queuing (87-59 R3) 3. SCSI Medium Changers (87-119 & 87-144) 4. Asynchronous Event Notification (87-52 R1) 5. Deferred Error Handling (87-98 R1) 6. Printer Command Set Proposal (87-103 R1) 7. Direct Access Device Translate Page in DIAGNOSTIC commands (87-126) 8. INQUIRY Command with VPD Info (87-127) 9. SEARCH Command modifications (87-128) 10. Verify Error Recovery Page (87-129) 11. Implemented Options (87-130) 12. MODE SELECT on a per initiator basis (87-131) 13. Fast SCSI (87-138 and 87-148) 14. Additions to READ POSITION COMMAND (87-139 R1) 15. See item 12. 16. Density Code for DAT Tape (87-145) 17. Additional Sense Codes (87-158) Gerry Holder 18. Review of SCSI-2 draft document. 19. Optical Issues (87-146) 20. TEST UNIT READY Command (87-147) 21. CD-ROM proposal review (87-106) 22. Target controlled data integrity (87-156) 23. SDTR negotiation (87-154) 24. Programmable SCSI option select (87-137) 25. READ LOG UPDATE command (87-43 & 87-114) Agenda Item 25. 26. MODE SELECT command modifications (87-122 R2) 27. BUSY status (87-149) 28. Zoning information (87-157) 29. October Working Group 30. January Working Group The following people attended the meeting: Name Status Organization ------------------------------ ------ ------------------------------ Mr. Robert C. Herron A 3M Company Mr. John Knudson S 3M Company Mr. Robert N. Snively P Adaptec, Inc. Mr. Charles Brill P AMP, Inc. Mr. C. Edward Reynolds O AMP, Inc. Mr. Edward Marsh O AMP, Inc. Mr. Louis C. Domshy P Archive Corp. Mr. Paul Kellogg O AT&T Information Systems Mr. Joe Lawlor O AT&T Information Systems Mr. David Harms O AT&T Information Systems Mr. David Barnum O Burndy Corp. Mr. Heinz Piorunneck O Burndy Corp. Mr. Gene Milligan P CDC/MPI Mr. Gerald Houlder A CDC/MPI Mr. Dan Davies P Cipher Data Products, Inc. Ms. Donna Pope P Cipher/Optimem Mr. Paul Nitza P Emulex Corp. Mr. Randy Weber ST Emulex Corp. Mr. I. Dal Allan P ENDL Mr. Robert Liu P Fujitsu America, Inc. Mr. Bob Thornton O Fujitsu Components of America Mr. Dave McIntyre A Hewlett Packard Co. Mr. Bob Pentecost O Hewlett Packard Co. Mr. Greg Floryance P IBM Corp. Mr. Gerald Marazas A IBM Corp. Mr. Gary R. Stephens A IBM Corp. Mr. Alan Ebright P Laserdrive Mr. Paul Boulay O LMSI-OSI Mr. Skip Jones P Maxtor Corp. Mr. Bob Masterson P Methode Electronics, Inc. Mr. Paul Wassenberg P Micropolis Corp. Mr. Thomas Gettys P Miniscribe Corp. Mr. John Lohmeyer P NCR Corp. Mr. Terry Johnson S NCR Corp. Mr. Duane Draving ST NCR Corp. Mr. Dave Skinner ST NCR Corp. Mr. Michael Linkous S NCR Corp. Mr. James McGrath P Quantum Corp. Mr. Dexter Anderson P Scientific Micro Systems Mr. Lawrence J. Lamers A Sony Corp. of America Mr. Dennis Appleyard P Storage Technology Corp. Mr. D. W. Spence A Texas Instruments Mr. Harvey Waltersdorf P Thomas & Betts Mr. Jeff Stai P Western Digital Status Key: P - Principal, A - Alternate, L - Liaison, O - Observer S - Special Interest Group, V - Visitor (no address info) The following documents were distributed at the meeting: Document Doc Date Author Description of Document ------------- -------- --------------- --------------------------------------- X3T9.2/87-122 8/31/87 J. McGrath MODE SELECT command modifications Rev 2 X3T9.2/87-139 8/27/87 D. Appleyard Additions to READ POSITION command for Rev 1 Sequential Access Devices X3T9.2/87-146 8/31/87 J. Stai Revised Optical Sections X3T9.2/87-149 8/28/87 P. Johansson Differentiation of Controller BUSY and LUN BUSY X3T9.2/87-150 8/28/87 P. Johansson Translate Page in SEND/RECEIVE DIAGNOSTIC Commands X3T9.2/87-151 8/28/87 P. Johansson Management of the "Grown" Defect List X3T9.2/87-152 8/27/87 D. Appleyard Response to REQUEST SENSE before CHECK CONDITION has been reported X3T9.2/87-154 9/1/87 D. McIntyre Clarification of Synchronous Data Transfer Negotiation X3T9.2/87-155 8/25/87 R. Weber Ending Extended Contingent Allegiance X3T9.2/87-156 8/30/87 J. McGrath Reordering of Commands in Command Queuing X3T9.2/87-157 8/30/87 J. McGrath Providing Initiators with Zoning (or Notched Drive) Information X3T9.2/87-158 G. Houlder Proposed Additional Sense Code Table for Section 8 X3T9.2/87-159 8/27/87 S. Jones Self Configuration Errors All items from the agenda were discussed with the following results: 1. Connector Issues Gerry Marazas acted as moderator for a meeting dealing with connector issues. The charter is to pick a high-density, shielded, connector. This meeting reported back the following. A pin and socket and ribbon type connector could both be adopted. A single choice is preferable, but not absolutely necessary. A request to schedule the connector issue at the October plenary at a known time was made. Their preference was for Tuesday morning. Prior to this there would be a meeting of connector people to resolve dimensions on Monday. Do we need 26 & 28 AWG compatibility? Maybe we should have two options: short & skinny(28 AWG) and long & fat cables(26 AWG). The manner of retention of the cable connector to board attachment needs some input. A single retention scheme for each type of connector is highly desirable. The choices are baillock vs. thumbscrew vs. latching type mechanism. Having stackable connectors may dictate the retention mechanism. The center to center distance may also restrict the connector size. Daisy chaining, both internal and external requirements (like IEEE 488). Chuck Brill made some comments regarding pin & socket connectors: 1) Electro-static discharge may be problem on Champ style connector; 2) Also the pin & socket is available for surface mounting. The B-cable will have a skewed sixteen arrangement using a 68-pin micro shielded and un-shielded connector. It is also recommended that an A-cable micro connector be chosen. It is the working groups recommendation that both of these will be added to the body of the standard to replace the existing 60- pin connector and not included in the appendix. The working group will issue a separate report on the subject of Wide SCSI (X3T9.2/87-160). 2. Command Queuing (87-59 R3) Bob Snively summarized the latest revision of the SCSI Command Queuing proposal. Revision 3 of the document contains the changes discussed at the Irvine working group. Jeff Stai mentioned that we had discussed error handling at the last plenary meeting without any resolution. The main question is when to restart command execution from the queue after an error occurs. When a CHECK CONDITION is sent by the target the queue will be disabled for the LUN which detected error. The target will only accept commands from the initiator which received the CHECK CONDITION and will execute them in a strict sequential order. Two new messages will be defined, INITIATE RECOVERY MODE and RELEASE RECOVERY MODE. The INITIATE RECOVERY MODE message will be sent by the target between the status and COMMAND COMPLETE message when a CHECK CONDITION occurs. This message notifies the initiator that the target has entered the error recovery mode defined above. The initiator sends the RELEASE RECOVERY MODE message when it has completed and error recovery action. Once the target has received the RELEASE RECOVERY MODE message it will continue normal operation on that LUN. Note: This method of Error Recovery Locking is independent of command queuing and device type. The following outstanding issues are not addressed in revision 3 of the document but will be included in the next revision: (1) UNIT ATTENTION condition to other initiators when one initiator clears the command queue. (2) What happens to queued commands when a soft reset occurs. Between this item and item number 1 above a considerable amount of time was spent on deferred errors, error processing/recovery in sequential devices and error processing/recovery in direct access cache devices. This whole area needs a large amount of work and should be scheduled as a separate working group agenda item. Bob Snively accepted an action item to revise document 87-59 to include the changes discussed at this meeting. Bob Snively accepted an action item to generate a new proposal which documents the discussion on Error Recovery Locking Mode operations. Dan Davies accepted an action item to update his Deferred Error for Buffered Mode 2 proposal to include the Error Recovery Locking Mode. Bob Snively accepted an action item to update his AEN proposal to include the Error Recovery Locking Mode. Bob Snively accepted an action item to update document 87-19 to include the new Error Recovery Locking mode, deferred error returned sense code (71h) and AEN. 3. SCSI Medium Changers (87-119 & 87-144) There is some merit to the Cygnet and STC proposals. However these need to be edited. There was a discussion of the merit of keeping media changers included in the SCSI-2 document. In particular the manufacturers of media changers seem to lack commitment to participation in the working groups. An agenda item for the next plenary is to review the editorship of this section. Gary Stevens accepted an action item to develop a primitive medium changer model. 4. Asynchronous Event Notification (87-52 R1) This proposal was accepted by the working group for inclusion in the next revision on the SCSI-2 draft document. 5. Deferred Error Handling (87-98 R1) Dan Davies presented his proposal on the handling of deferred errors for the new sequential access buffered mode 2. The only area where there were problems had to do with the residual count returned by the target when an error occurs. Does the residue contain the data currently in the buffer as stated on page 9-48/50 or not as stated on page 7-28? Items 1, 3 and 4 of this proposal were accepted by the working group for inclusion in the next revision of the SCSI-2 draft document. Bob Snively accepted an action item to incorporate Item 2 of 87-98 R1 in his proposal on Error Recovery Locking Mode. Gary Stephens accepted an action item to submit a document which clarifies the residue on a deferred error. 6. Printer Command Set Proposal (87-103 R1) John Lohmeyer summarized the printer proposal document. The RESERVE/RELEASE, SEND DIAGNOSTIC and TEST UNIT READY commands should be mandatory. Gene Milligan suggested that a paragraph be added on the first page that states this document is meant to encapsulate the data sent to the printer and the data may contain device specific control sequences. This proposal was accepted by the working group for inclusion in the next revision on the SCSI-2 draft document. 7. Direct Access Device Translate Page in DIAGNOSTIC commands (87-126) A lengthy discussion took place on whether or not the physical sector format refers to the sector number relative to index (sectors from index) or the current sector number. The committee decided to leave the current wording in the document since several manufacturers have implemented the option in different ways but all are within the standard. Gene submitted proposed wording for the translate function. A vote was taken to include his statement: 8 in favor, 1 opposed. It was suggested that it might belong in the direct access model. Paul Nitza accepted an action item to find a place to incorporate the wording in the SCSI-2 document. Peter Johansson's (87-150) comments was discussed. There is a legitimate reason to return a zero value if the address does not translate to requested address format. Unresolved yet is the proper action to take when an out of bounds condition exists. Paragraph two is accepted, but the last sentence will not be struck. Paragraph three is accepted, however alternate track will be retained, and alternate sectors added. Paul Nitza accepted an action item to edit these into the SCSI-2 document. 8. INQUIRY Command with VPD Info (87-127) There was a discussion about removing the Product Serial Number from bytes 56 to 67 and make it VPD page 83. Should it be reported in both places? Paul, in his frustration with the extended discussions, made a motion to remove the Product Serial Number field. The motion failed for lack of a second. Paul Nitza accepted an action item to add a description to clarify the serial number issue to the SCSI-2 document. A long discussion ensued on the topic of target/un differentiation. Gary Stevens argues that target/LUN differentiation should be done at the identify message, so that it applies to all commands. This could be a future proposal for SCSI-3. A straw vote was taken on the question "Do you like the identification of target and LUN?": 10 in favor, 6 opposed. A straw vote was taken on the question "Is the target/LUN bit in for SCSI- 2?": 8 in favor, 6 opposed. Since this was a close vote the bit will be included on an interim basis. An agenda item for the next plenary is to have a definitive vote on the target/LUN issue. The underscored text of 87-127 was accepted for inclusion in the SCSI-2 document. 9. SEARCH Command modifications (87-128) Jeff Stai made a friendly amendment to extend the LBA list format to eight bytes per entry to allow a four-byte 'number of blocks' field after each LBA. This was accepted. Gary Stephens pointed out that a version of this command that allowed searching an area defined as a bit map of blocks would be useful as well. The bit map would be passed after each LBA. This was also accepted. Paul Nitza accepted an action item to generate a revised proposal. 10. Verify Error Recovery Page (87-129) This proposal was accepted by the working group for inclusion in the next revision on the SCSI-2 draft document. 11. Implemented Options (87-130) This proposal was accepted as modified by the working group for inclusion in the next revision of the SCSI-2 draft document in the INQUIRY command. Jim McGrath accepted an action item to develop a proposal for a configuration implemented options page in the MODE SENSE command. 12. MODE SELECT on a per initiator basis (87-131) There was discussion of this topic interwoven with other discussions. Paul Boulay accepted an action item to revise this proposal to encapsulate MODE SELECT data in the COPY command data. 13. Fast SCSI (87-138 and 87-148) This proposal was accepted by the working group for inclusion in the next revision on the SCSI-2 draft document. The question is "How does this get incorporated into the document?". Should it be done as a linear function or a step function for the transfer rates? Paul Nitza accepted an action item to draft fast SCSI into the document using a step function instead of a linear function for transfer rates. 14. Additions to READ POSITION COMMAND (87-139 R1) This proposal was accepted by the working group for inclusion in the next revision on the SCSI-2 draft document. The wording in the last paragraph on page 9-5 needs to be rewritten to clarify it. For buffered mode 0 and 1 the target shall discard the buffered data if the previous command from any initiator was terminated with a CHECK CONDITION status when the initiator sends a load/unload command. This same paragraph is also in other places in section 9. Note: This does not cover buffered mode 2 which gets back into deferred error handling. The ERQ bit on Page 9-19 should be deleted since reservation queuing has been removed from SCSI-2. Jim Semenak was assigned the action item to include this proposal as amended by the working group. 15. See item 12. 16. Density Code for DAT Tape (87-145) Density code 13h was assigned to NRZI 8/10 encoding for DAT. This proposal was accepted by the working group for inclusion in the next revision on the SCSI-2 draft document. 17. Additional Sense Codes (87-158) Gerry Holder Paul Boulay's codes need to get incorporated. The proposal should be reviewed by the members of the committee. This turns out to be a religious argument which cannot be resolved. A bit of gospel will precede each sense code table to expound the correct faith to believe in for that device type. Paul Nitza accepted an action item to write a letter to the Ephesians which will precede the Additional Sense Code Tables. 18. Review of SCSI-2 draft document. This task was not started. An agenda item for the next plenary is to review the optical memory devices section. 19. Optical Issues (87-146) There is a potential conflict of the erase bypass bit and the write same bit in the command descriptor block of WRITE commands in sections 8 and 15. The working group felt that the erase bypass bit was more useful than write same for eraseable optical devices, but that having both bits in the same location might be dangerous. Donna Pope accepted an action item to draft a proposal to the write same function to the SEND DIAGNOSTIC command. Paul Boulay and Jeff Stai accepted an action item to reword the ASA bit. A disable block descriptor bit is to be added to the MODE SENSE command descriptor block. The block descriptor page is deleted. This bit is being considered for inclusion in the other device type's MODE SENSE commands. Paul Nitza accepted an action item to develop a proposal for a new MODE SELECT MODE SENSE command using ten-byte command descriptors. Jeff Stai accepted an action item to generate a proposal for Page Code 09h to be updated and extended to cover other low level items. Skip Jones and Jeff Stai accepted an action item to incorporate the 87-63 proposal into the READ UPDATED BLOCKS command. Donna Pope, Paul Boulay, and Jeff Stai accepted an action item to develop some needed additional wording for the Enable Blank Check bit regarding how it works in the UPDATE BLOCK command. Larry Lamers accepted an action item to develop an optical model. 20. TEST UNIT READY Command (87-147) This proposal specifies the responses to TEST UNIT READY if the status is not GOOD. This proposal also touches on several other proposals before the working group including 87-149, 87-158 and 87-159. The discussion went off tangentially on the Additional Sense Code topic. Should direct-access devices start using byte 13 modifiers for additional sense codes as the sequential access devices do? How about the conflicts between direct-access and sequential access codes? The proposal must consider the how it affects all device types, since it will be propagated. The restriction on not disconnecting in the proposal will be deleted. Bill Spence accepted an action item to revise the proposal based on the input received from the working group. 21. CD-ROM proposal review (87-106) The model description needs to be pulled together at the beginning of the document instead of scattered through the command descriptions. Larry Lamers accepted an action item to draft a second revision of this proposal. 22. Target controlled data integrity (87-156) Jim McGrath will modified the proposal to be included as an appendix in the document. John Lohmeyer made a comment on the number of appendices it looks like we are planning for SCSI-2 and the sheer volume of paper involved. The appendices have not be distributed yet, nor have they received much editorial review. Jim McGrath and Bob Snively accepted an action item to draft wording for the queuing section in the SCSI-2 document. 23. SDTR Negotiation (87-154) An agenda item for the next plenary is to review the this document. Dave McIntyre accepted an action item to rewrite the implied agreement portion. 24. Programmable SCSI option select (87-137) This is the proposed command to set the level of SCSI that the target operates at, effectively allowing the host to switch between SCSI-1 and SCSI-2 modes. Does it eliminate DDF in INQUIRY? This needs further consideration, since DDF is intended to set the format of the data returned. There was a request from Jim McGrath for a password feature. Data out parameters will be added for providing a password. Bob Snively accepted an action item to modify the proposal to include the password feature. 25. READ LOG UPDATE command (87-43 & 87-114) Agenda Item 25. Randy's pages had been accepted at Irvine. Paul Nitza had an action item on this. The current status of this item is a little murky. An agenda item for the next plenary is to review the status of the READ LOG UPDATE command. section. 26. MODE SELECT command modifications (87-122 R2) This proposal requests an ability to switch between current, saved, default parameters without sending out all the parameters. Jim McGrath accepted an action item to develop a proposal to put this feature into the SEND/RECEIVE DIAGNOSTIC command. Skip Jones accepted an action item to provide an example of MODE SELECT parameter changing as it should be implemented based on SCSI-2 R2. 27. BUSY status (87-149) The target/LUN bit and Peter Johansson's 87-149 document to add a CONTROLLER BUSY and LUN BUSY status were discussed together. The work in item 20 above addresses the immediate concerns of 87-149. This issue will be discussed in relation to the VPD discussion in INQUIRY at the next plenary. 28. Zoning information (87-157) Jim McGrath accepted an action item to generate a proposal based on the 87- 157 comment. 29. October Working Group The working group will be held on October 26, 27, 28, 1987 in the Orlando, Florida area at: Grosvenor Resort (305) 828-4444 Lake Buena Vista, FL (800) 624-4109 The room rate is $80.00, single or double. When making reservations mention ANSI. Please make your reservation by October 10. There is transportation from the airport via American Sightseeing for $9.50 each way. 30. January Working Group The meeting is scheduled for January 7,8,9, 1988 in San Jose, CA. Bob Snively of Adaptec will make arrangements.