T10/07-435r0 Voting Results on T10 Letter Ballot 07-434r0 on Forwarding SAM-4 to First Public Review Ballot closed: 2007/10/29 12:00 noon MDT Organization Name S Vote Add'l Info --------------------------------- -------------------- - ---- ---------- Adaptec, Inc. DNV AMCC Paul von Stamwitz P Yes Amphenol Interconnect Gregory McSorley P Yes ATL Technology Jaremy Flake P Yes Brocade David Peterson P Yes Cmnts Dell, Inc. Kevin Marks P Yes EMC Corp. David Black A Yes Emulex William Martin P Yes Cmnts ENDL Ralph O. Weber P Yes FCI Douglas Wagner P Yes Finisar Corp. David Freeman P Yes Foxconn Electronics Elwood Parsons P Yes Fujitsu Mike Fitzpatrick P Yes General Dynamics Nathan Hastad P Yes Hewlett Packard Co. Rob Elliott P No Cmnts Hitachi Global Storage Tech. Dan Colegrove P Yes IBM Corp. George O. Penokie A Yes Intel Corp. Mark Seidel P Yes Iomega Corp. Robert Payne P Yes Kawasaki Microelectronics Am Joel Silverman P Yes KnowledgeTek, Inc. Dennis Moore P Yes Lexar Media, Inc. Martin Furuhjelm A Yes LSI Corp. John Lohmeyer P Yes Marvell Semiconductor, Inc. Paul Wassenberg A Yes Maxim Integrated Products Gregory Tabor P Yes Microsoft Corp. Mark Benedikt P Yes Molex Inc. Jay Neer P Yes NeoScale Systems Inc. Landon Noll A Yes Network Appliance Frederick Knight P Yes Cmnts Nvidia Corp. Mark Overby P Yes Panasonic Technologies, Inc Takaharu Ai A Yes PMC-Sierra Tim Symons P Yes Quantum Corp. Paul Suhler P Yes Cmnts Samsung Michael Rogers A Yes SanDisk Corporation Avraham Shimor P Yes Seagate Technology Gerald Houlder P Yes STMicroelectronics, Inc. Stephen Finch P Yes Sun Microsystems, Inc. Erich Oetting P Yes Symantec Roger Cummings P Yes TycoElectronics Ashlie Fan P Yes Vitesse Semiconductor Mahbubul Bari A Yes Western Digital Mark Evans P No Cmnts Ballot totals: (39:2:0:1=42) 39 Yes 2 No 0 Abstain 1 Organization(s) did not vote 42 Total voting organizations 6 Ballot(s) included comments This 2/3rds majority ballot passed. 39 Yes are more than half the membership eligible to vote [greater than 21] AND 39 Yes are at least 28 (2/3rds of those voting YES or NO [41]). Key: P Voter is principal member A Voter is alternate member Abs Abstain vote DNV Organization did not vote Cmnts Comments were included with ballot NoCmnts No comments were included with a vote that requires comments [This report prepared by LB2 v2.2.] ************************************************************** Comments attached to Yes ballot from David Peterson of Brocade: Brocade-001 The text in SAM-4 rev 13, clause 4.7.2 "The task identifier (i.e., the Q in an I_T_L_Q nexus) represents a task, allowing many uniquely identified tasks to be outstanding at once. Each SCSI transport protocol defines the size of the task identifier, up to a maximum of 64 bytes, to be used by SCSI ports that support that SCSI transport protocol." Has been changed to read something like: "The Task Identifier (i.e., the Q in an I_T_L_Q nexus) uniquely identifies a task,..." I do not believe that is precisely correct. It only identifies the task uniquely within the context of a particular I_T_L nexus. As an example, see SPI-3's use of the Message Out and Message In to provide the Q value, which is only valid for a particular I_T_L nexus. As a second example, consider FCP, that uses X_ID between a single initiator and target as the identifier, but where the same X_ID may appear on other commands from a different initiator to the same target. As a result, the proper wording would be something like: "The Task Identifier (i.e., the Q in an I_T_L_Q nexus) uniquely identifies a task in the context of a particular I_T_L nexus, ..." ************************************************************** Comments attached to Yes ballot from William Martin of Emulex: Emulex-001 Page: 2 first sentence below Figure 2 - "roadmap" s/b "document structure" Emulex-002 Page: 3 Device-Type Specific Command Sets: - second sentence "is" s/b "are" Emulex-003 Page: 5 3.1.10 "When in this state" s/b "A state in which" Emulex-004 Page: 6 3.1.28 "supplier" s/b "server" Emulex-005 Page: 6 3.1.35 and 3.1.36 "When in this state" s/b "A state in which" Emulex-006 Page: 11 3.1.121 This definition is a partial sentence. It needs to specify: The portion (i.e., Q) of an I_T_L nexus (see 3.1.49) in a task set that uniquely identifies each task. Emulex-007 Page: 12 ADC-2, FCP-4, SAS-2, SBC-3, SBP-3 and SPC-4 make hyphenation consistent in all full standard names. Some have a space before hyphen some do not. ADC-2 has a space after the hyphen. Emulex-008 Page: 13 3.3.8 last sentence: For backward compatibility in future standards, shouldn't this be "Recipients shall not check ..."? Emulex-009 Page: 34 4.5.10 second paragraph: This sentence fragment seems to be out of place. Emulex-010 Page: 35 Paragraph after second a-b list "The interactions between the task manager, or a task router, &" remove the first comma. Emulex-011 Page: 35 4.5.11.3 "The Function Identifier attribute contains function identifier" s/b "The Function Identifier attribute contains a function identifier" Emulex-012 Page: 72 task set full end of second paragraph there is an extraneous "c" in sentence Emulex-013 Page: 79 Receive Data-Out Input argument list: Put these arguments in the same order as in the service request above. Emulex-014 Page: 122 A.2 first sentence "attribute" s/b "attributes" Emulex-015 Page: 122 A.2 last two sentences of first paragraph: To clarify that the tables do not specify which standard makes requirements, change to: "Table A.1 also lists whether SCSI transport protocols and logical units are required to support identifier attributes by either this standard or SPC-4. Table A.2 also lists whether SCSI transport protocols and logical units are required to support name attributes by either this standard or SPC-4." Emulex-016 Page: 127 A.3.4: This seems to be a self reference. Should this be "SAS-2 Serial SCSI Protocol (see SAS-2)." ************************************************************** Comments attached to No ballot from Rob Elliott of Hewlett Packard Co.: HPQ comment number 1 Page=1 Subtype=Text Subj=Note Author=relliott Comment= move right-justified text on page i right by .2 inches to line up with the horizontal lines --- HPQ comment number 2 Page=3 Subtype=Highlight Subj=Highlight Author=relliott Comment= SCSI Architecture Model - 4 s/b SCSI Architecture Model - 4 (SAM-4) --- HPQ comment number 3 Page=18 Subtype=Highlight Subj=Highlight Author=relliott Comment= The SCSI Architecture Model - 4 standard s/b This standard --- HPQ comment number 4 Page=18 Subtype=Highlight Subj=Highlight Author=relliott Comment= Lists s/b lists --- HPQ comment number 5 Page=18 Subtype=Highlight Subj=Highlight Author=relliott Comment= SAM-3 and this standard s/b this standard and previous versions of this standard --- HPQ comment number 6 Page=19 Subtype=Highlight Subj=Highlight Author=relliott Comment= SCSI Architecture Model - 4 s/b SCSI Architecture Model - 4 (SAM-4) (page 1 footer differs from the other pages) --- HPQ comment number 7 Page=19 Subtype=Highlight Subj=Highlight Author=relliott Comment= architecture model concepts s/b concepts --- HPQ comment number 8 Page=20 Subtype=Highlight Subj=Highlight Author=relliott Comment= Architecture Model s/b SCSI Architecture Model --- HPQ comment number 9 Page=21 Subtype=Highlight Subj=Highlight Author=relliott Comment= Architecture Model s/b SCSI Architecture Model --- HPQ comment number 10 Page=22 Subtype=Highlight Subj=Highlight Author=relliott Comment= References s/b references --- HPQ comment number 11 Page=22 Subtype=StrikeOut Subj=Cross-Out Author=relliott Comment= Delete "IEC 60027-2:2000, Letter symbols to be used in electrical technology - Part 2: Telecommunications and electronics" This is the reference that defines Ki, Mi, etc. prefixes for powers-of-two units to avoid misusing the SI powers-of-ten units. These prefixes are not used in this standard. --- HPQ comment number 12 Page=23 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA task attribute s/b ACA (smallcaps lowercase) task attribute --- HPQ comment number 13 Page=23 Subtype=Text Subj=Note Author=relliott Comment=At the end of 3.1.14 class diagram, add "See 3.6.2." --- HPQ comment number 14 Page=23 Subtype=StrikeOut Subj=Cross-Out Author=relliott Comment=performed by a task --- HPQ comment number 15 Page=24 Subtype=Highlight Subj=Highlight Author=relliott Comment= (see 3.1.45), also s/b (see 3.1.45). Also, --- HPQ comment number 16 Page=24 Subtype=Highlight Subj=Highlight Author=relliott Comment= returning a status and service response of TASK COMPLETE s/b returning a service response of TASK COMPLETE The status is of secondary importance and doesn't need to be mentioned. As worded, it sounds like TASK COMPLETE could be a status value. --- HPQ comment number 17 Page=24 Subtype=StrikeOut Subj=Cross-Out Author=relliott Comment= 3.1.22 completed task: A task that has ended by returning a status and service response of TASK COMPLETE. Delete this definition and replace all uses of "completed task" with "completed command." Their definitions are identical now that task=command. --- HPQ comment number 18 Page=24 Subtype=Text Subj=Note Author=relliott Comment= The relationship of command and task is unclear now that linked commands are gone. The standard uses a mix of the terms with no apparent reason (if read without remembering the history). --- HPQ comment number 19 Page=26 Subtype=Text Subj=Note Author=relliott Comment=At the end of 3.1.72 object diagram, add "See 3.6.3." --- HPQ comment number 20 Page=26 Subtype=Text Subj=Note Author=relliott Comment= The phrase "logical unit number" is used many times in the standard where the acronym LUN could/should be used instead. --- HPQ comment number 21 Page=27 Subtype=Text Subj=Note Author=relliott Comment=At the end of 3.1.80 procedure call, add "See 3.6.4." --- HPQ comment number 22 Page=27 Subtype=Text Subj=Note Author=relliott Comment= Add: 3.1.xx power loss expected: A condition resulting from a power loss expected event in which the logical unit performs the power loss expected operations described in 6.3.5, SPC-4, and the appropriate transport protocol and command standards. --- HPQ comment number 23 Page=27 Subtype=Text Subj=Note Author=relliott Comment= Add: 3.1.xx power loss expected event: An event that results in a power loss expected condition (see 3.1.xx) as described in 6.3.5. --- HPQ comment number 24 Page=29 Subtype=Highlight Subj=Highlight Author=relliott Comment= (global) task identifier s/b task tag There is no good justification for making this change from SAM-3 to SAM-4. Every transport protocol uses the name "tag" now and will have to unnecessarily change. This is reminiscent of changing "queue" to "task set" from SCSI-2 to SCSI-3. This helps make the ingredients in I_T_L_Q nexus have similar names, but "logical unit number" is not being renamed to "logical unit identifier" to make them all consistent. --- HPQ comment number 25 Page=29 Subtype=Highlight Subj=Highlight Author=relliott Comment= "assume the SCSI initiator device role" The logical unit doesn't become an SCSI initiator device itself. The command forces the SCSI device containing that logical unit to assume the SCSI initiator device role. --- HPQ comment number 26 Page=29 Subtype=Highlight Subj=Highlight Author=relliott Comment= whose objects are, or an object that is, within the logical unit representing s/b within the logical unit whose objects represent, or an object that represents, --- HPQ comment number 27 Page=29 Subtype=Highlight Subj=Highlight Author=relliott Comment= whose objects are, or an object that is, within a logical unit that controls s/b within a logical unit whose objects control, or an object that controls --- HPQ comment number 28 Page=29 Subtype=Highlight Subj=Highlight Author=relliott Comment= are s/b are each to match "is" later --- HPQ comment number 29 Page=29 Subtype=Highlight Subj=Highlight Author=relliott Comment= whose objects are, or an object that is, within a SCSI target port that routes s/b within a SCSI target port whose objects route, or an object that routes --- HPQ comment number 30 Page=31 Subtype=Highlight Subj=Highlight Author=relliott Comment= as error s/b as an error --- HPQ comment number 31 Page=32 Subtype=Highlight Subj=Highlight Author=relliott Comment= Within class diagrams and object diagrams there may be constraints which specify requirements and notes which are informative. s/b Class diagrams and object diagrams may include constraints, which specify requirements, and notes, which are informative. --- HPQ comment number 32 Page=41 Subtype=StrikeOut Subj=Cross-Out Author=relliott Comment= Delete "(e.g., a task identifier)" None of the classes include that as their single attribute. --- HPQ comment number 33 Page=41 Subtype=Highlight Subj=Highlight Author=relliott Comment= The model does not address other requirements that may be essential to some I/O system implementations s/b The model does not address other requirements that are essential to some I/O system implementations --- HPQ comment number 34 Page=42 Subtype=Highlight Subj=Highlight Author=relliott Comment= A client may only originate requests for service. A server may only respond to such requests. s/b A client only originates requests for service. A server only responds to such requests. --- HPQ comment number 35 Page=43 Subtype=Text Subj=Note Author=relliott Comment= The last paragraph in 4.3.1 discusses commands and device server requests. It should also discuss TMFs and task management requests, to cover everything shown in figure 12. --- HPQ comment number 36 Page=43 Subtype=Highlight Subj=Highlight Author=relliott Comment= "SCSI initiator device" is not quite right. The SCSI initiator device might deduce that a command was received by the target by noticing data transfer requests for that command. Some protocols explicitly mention that "implicit ACK." It might be better to word this sentence with "application client", since although it is part of the initiator device it is not involved in the data transfer protocol services. --- HPQ comment number 37 Page=43 Subtype=Highlight Subj=Highlight Author=relliott Comment= architecture model s/b SCSI architecture model --- HPQ comment number 38 Page=44 Subtype=Highlight Subj=Highlight Author=relliott Comment= SCSI domain s/b SCSI Domain class --- HPQ comment number 39 Page=46 Subtype=Highlight Subj=Highlight Author=relliott Comment= SCSI Domain class (figure 15) s/b SCSI Domain class (see figure 15) --- HPQ comment number 40 Page=46 Subtype=Text Subj=Note Author=relliott Comment= The implication that a logical unit is contained within only one SCSI target device may be too tight. 1. The way hierarchical logical units are currently modeled, a logical unit is part of its real SCSI target device and also part of each of the SCSI target devices that route to it. (A separate comment suggests removing hierarchical logical units from the model) 2. With virtualization (e.g., RAID), some logical units (e.g., the physical disk drives) are used by a "higher level" logical unit (e.g., a RAID-5 volume). There is some interaction between the states of the higher and lower level logical units. 3. With remote replication, a logical unit can be in two different places at the same time. The media is essentially synchronized (writes to New York are immediately picked up by reads in Los Angeles). However, the task set states are not (an ABORT TASK SET in New York doesn't abort commands pending in Los Angeles). To acknowledge these oddities, perhaps add a statement like "The medium accessed by a logical unit may not be exclusively accessible through that logical unit." An optional containment relationship from logical unit to logical unit might represent the more complex interactions. --- HPQ comment number 41 Page=46 Subtype=Text Subj=Note Author=relliott Comment= The SCSI Device to SCSI Target Device/SCSI Initiator Device relationship should be an inheritance relationship, not an aggregation relationship. Same for SCSI Port to SCSI Initiator Port/SCSI Target Port. This would be a proper use of multiple inheritance. Comment from someone experienced with UML at HP: "When they start showing that ports "contain" target and initiator ports, it seems to me that they really are describing inheritance. If that is true, they are mixing inheritance (a port shouldn'treally be a SCSI device so much as a Network device anyway so that SCSI can go over any topology) and containment concepts in the same UML by overloading the aggregation symbol to include inheritance. It is probably going to be hard for UML people to decipher. It is not at all consistent with UML for SCSI management (read SMI-S here), so if they are planning on representing topologies or developing a data model for management or as part of the protocol with this then I am really concerned." --- HPQ comment number 42 Page=47 Subtype=Highlight Subj=Highlight Author=relliott Comment= A Service Delivery Subsystem class s/b The Service Delivery Subsystem class --- HPQ comment number 43 Page=48 Subtype=Highlight Subj=Highlight Author=relliott Comment= "See figure 17 for the SCSI Device class diagram. The SCSI Device class" s/b "The SCSI Device class (see figure 17)" --- HPQ comment number 44 Page=48 Subtype=Highlight Subj=Highlight Author=relliott Comment= architecture model s/b SCSI architecture model --- HPQ comment number 45 Page=50 Subtype=Highlight Subj=Highlight Author=relliott Comment= SCSI Initiator Port class s/b SCSI Initiator Port class (see figure 18) --- HPQ comment number 46 Page=50 Subtype=Highlight Subj=Highlight Author=relliott Comment= SCSI Initiator Port (see figure 18) class s/b SCSI Initiator Port class --- HPQ comment number 47 Page=50 Subtype=Highlight Subj=Highlight Author=relliott Comment=; s/b . --- HPQ comment number 48 Page=50 Subtype=Highlight Subj=Highlight Author=relliott Comment= Relative port identifiers may be retrieved through the Device Identification VPD page (see SPC-4) and the SCSI Ports VPD page (see SPC-4). s/b The Device Identification VPD page (see SPC-4) and the SCSI Ports VPD page (see SPC-4) report relative port identifiers. --- HPQ comment number 49 Page=50 Subtype=Highlight Subj=Highlight Author=relliott Comment= A SCSI target port may have at most one name. s/b A SCSI target port shall have at most one name. --- HPQ comment number 50 Page=51 Subtype=Highlight Subj=Highlight Author=relliott Comment= A SCSI initiator port may have at most one name. s/b A SCSI initiator port shall have at most one name. --- HPQ comment number 51 Page=51 Subtype=Highlight Subj=Highlight Author=relliott Comment= incorrect logical unit There is no such thing as an incorrect logical unit, just incorrect logical unit numbers. --- HPQ comment number 52 Page=51 Subtype=Highlight Subj=Highlight Author=relliott Comment= incorrect logical unit There is no such thing as an incorrect logical unit, just incorrect logical unit numbers. --- HPQ comment number 53 Page=52 Subtype=Text Subj=Note Author=relliott Comment= Nexus and Function Identifier should be swapped in figure 19 and in sections 4.5.11.2 and 4.5.11.3 to match the SCSI Target Device side in figure 22 and following sections. (or the target side should be swapped to follow this) --- HPQ comment number 54 Page=52 Subtype=Highlight Subj=Highlight Author=relliott Comment= Service Response[0..1] s/b Service Response[1] like in Application Client Task (also see comment on text removing "if any") --- HPQ comment number 55 Page=53 Subtype=Highlight Subj=Highlight Author=relliott Comment= Application Client Task Management Function class s/b Application Client Task Management Function class (see figure 19) --- HPQ comment number 56 Page=53 Subtype=Highlight Subj=Highlight Author=relliott Comment= Service Response attribute, if any, s/b Service Response attribute (also see comment on table changing [0..1] to [1]) --- HPQ comment number 57 Page=54 Subtype=Highlight Subj=Highlight Author=relliott Comment= Application Client Task class s/b Application Client Task class (see figure 19) --- HPQ comment number 58 Page=54 Subtype=Highlight Subj=Highlight Author=relliott Comment= (e.g., SIMPLE task attribute, ORDERED task attribute, HEAD OF QUEUE task attribute, ACA task attribute) s/b (e.g., SIMPLE, ORDERED, HEAD OF QUEUE, or ACA) --- HPQ comment number 59 Page=54 Subtype=Text Subj=Note Author=relliott Comment= All the "output" attribute sections 4.5.13.2 to 4.5.13.12 and 4.5.13.19 should cross reference 5.4.2.2 (Send SCSI Command), just like the "input" attribute sections (4.5.13.13 to 4.5.13.18) reference 5.4.2.5 (Command Complete Received). --- HPQ comment number 60 Page=54 Subtype=Highlight Subj=Highlight Author=relliott Comment= SPC-3 s/b SPC-4 or delete this reference and just refer to 5.2 alone --- HPQ comment number 61 Page=55 Subtype=Highlight Subj=Highlight Author=relliott Comment=; s/b : --- HPQ comment number 62 Page=55 Subtype=Text Subj=Note Author=relliott Comment= Delete the hierarchical logical unit classes from the UML model. UML should just model the logical units that are contained in the SCSI target device (i.e., the level 1 hierarchical logical units). Some LUN values address those logical units; others address logical units in other SCSI target devices. They should not be considered part of the same SCSI target device. The Task Router class should own the rules about parsing a LUN field (e.g. deciding where to send a task or TMF - send it to a logical unit in this target, or forward it elsewhere) --- HPQ comment number 63 Page=55 Subtype=Highlight Subj=Highlight Author=relliott Comment= the retry delay time for the completed command (see 5.4.2.5) s/b the additional status information for the completed command (see 5.3.2 and 5.4.2.5). --- HPQ comment number 64 Page=55 Subtype=Highlight Subj=Highlight Author=relliott Comment= command (see 5.4.2.5). s/b command (see 5.8.6 and 5.4.2.5) --- HPQ comment number 65 Page=55 Subtype=Highlight Subj=Highlight Author=relliott Comment= command (see 5.4.2.5). command (see 5.8.6 and 5.4.2.5). --- HPQ comment number 66 Page=55 Subtype=Text Subj=Note Author=relliott Comment= Move 4.5.13.16 First Burst Enabled attribute up after 4.5.13.12 Task Priority attribute so the outputs are all ahead of the inputs. Also move it higher in the UML diagram attribute list. --- HPQ comment number 67 Page=58 Subtype=Highlight Subj=Highlight Author=relliott Comment= Hierarchical Logical Unit class There is no class with that name, and it doesn't appear in figure 22. --- HPQ comment number 68 Page=59 Subtype=Highlight Subj=Highlight Author=relliott Comment= I_T_L_Q nexus s/b I_T_L_Q Nexus --- HPQ comment number 69 Page=60 Subtype=Highlight Subj=Highlight Author=relliott Comment= in the logical unit s/b if the logical unit --- HPQ comment number 70 Page=61 Subtype=Highlight Subj=Highlight Author=relliott Comment= Task class s/b Task class (see figure 22) --- HPQ comment number 71 Page=61 Subtype=Highlight Subj=Highlight Author=relliott Comment= (see 8.6) contains Delete the blue underline after the ")" and before "contains" --- HPQ comment number 72 Page=61 Subtype=Highlight Subj=Highlight Author=relliott Comment= SPC-3 s/b SPC-4 or delete this reference and just refer to 5.2 alone --- HPQ comment number 73 Page=63 Subtype=Highlight Subj=Highlight Author=relliott Comment= 4.5.24.3 should be swapped with 4.5.24.2 to match the order the attributes are listed in figure 22. --- HPQ comment number 74 Page=63 Subtype=Highlight Subj=Highlight Author=relliott Comment= The name of the well known logical unit may be determined by issuing an INQUIRY command requesting the Device Identification VPD page (see SPC-4). s/b The Device Identification VPD page (see SPC-4) reports the names of the SCSI target device (i.e., the names of the well-known logical unit). --- HPQ comment number 75 Page=63 Subtype=Highlight Subj=Highlight Author=relliott Comment= "If a well known logical unit is supported within a SCSI target device, then that logical unit shall support all the commands defined for it." s/b "A well known logical unit shall support all the commands defined for it." --- HPQ comment number 76 Page=63 Subtype=Highlight Subj=Highlight Author=relliott Comment= selection of incorrect logical units s/b incorrect logical unit numbers --- HPQ comment number 77 Page=66 Subtype=Highlight Subj=Highlight Author=relliott Comment= 8 9 10 11 s/b 4 5 6 7 --- HPQ comment number 78 Page=68 Subtype=Text Subj=Note Author=relliott Comment= The field name is ADDRESS METHOD, but the descriptions all use "addressing method" --- HPQ comment number 79 Page=69 Subtype=Highlight Subj=Highlight Author=relliott Comment= selection of incorrect logical units s/b incorrect logical unit numbers --- HPQ comment number 80 Page=69 Subtype=Text Subj=Note Author=relliott Comment= In 4.6.7 Peripheral device addressing method and 4.6.9 Logical unit addressing method, add a figure showing one level of hierarchy to illustrate the relay concept and how those addressing methods parse the fields. --- HPQ comment number 81 Page=70 Subtype=Highlight Subj=Highlight Author=relliott Comment= by a BUS IDENTIFIER field d/n nu a BUS IDENTIFIER field of zero --- HPQ comment number 82 Page=70 Subtype=Highlight Subj=Highlight Author=relliott Comment= may be addressed s/b is addressed --- HPQ comment number 83 Page=70 Subtype=Highlight Subj=Highlight Author=relliott Comment= selection of incorrect logical units s/b incorrect logical unit numbers --- HPQ comment number 84 Page=71 Subtype=Highlight Subj=Highlight Author=relliott Comment= m should be two rows : n+1 m sharing the EXTENDED ADDRESS METHOD SPEIFIC field --- HPQ comment number 85 Page=72 Subtype=Text Subj=Note Author=relliott Comment= Earlier addressing format tables used n-1 to n. Tables 13, 15, 16, 17, 20, 21 should end in n, not start with n. --- HPQ comment number 86 Page=72 Subtype=StrikeOut Subj=Cross-Out Author=relliott Comment= Delete (MSB)/(LSB) from the EXTENDED ADDRESS METHOD SPECIFIC field in table 16, 17, 18 --- HPQ comment number 87 Page=72 Subtype=StrikeOut Subj=Cross-Out Author=relliott Comment= Delete (MSB)/(LSB) from the EXTENDED ADDRESS METHOD SPECIFIC field in table 16, 17, 18 --- HPQ comment number 88 Page=72 Subtype=Text Subj=Note Author=relliott Comment= A LUN that includes a LENGTH field value that goes beyond the LUN field length supported by the transport protocol (2 bytes or 8 bytes) is invalid. For example, if the protocol supports 8-byte LUNs, and a LUN contains: bytes 0-1: logical unit addressing format byte 2: address method 11b, length 11b bytes 3-7: ... that L:UN must be treated as an invalid LUN (two bytes are being truncated). For example, if the protocol only supports 2-byte LUNs, then a LUN containing anything longer must be considered invalid: byte 0: address method 11b, length 01b, 10b, or 11b byte 1: ... --- HPQ comment number 89 Page=73 Subtype=StrikeOut Subj=Cross-Out Author=relliott Comment= Delete "A SCSI target device may support zero or more well known logical units (see 4.5.25)." Since this allows 0 through infinity, it is not stating a requirement or allowance. Could replace with ""A SCSI target device supports zero or more well known logical units (see 4.5.25)." --- HPQ comment number 90 Page=73 Subtype=StrikeOut Subj=Cross-Out Author=relliott Comment= Delete (MSB)/(LSB) from the EXTENDED ADDRESS METHOD SPECIFIC field in table 16, 17, 18 --- HPQ comment number 91 Page=73 Subtype=Text Subj=Note Author=relliott Comment= Code Fh length 00b Code Fh length 01b Code Fh length 10b should each also be a variant of "Logical unit not specified", used for hierarchical LUN situations where the lowest level logical unit receives one of these incoming LUN values, not all FFs: FFFF0000_00000000h FFFFFFFF_00000000h FFFFFFFF_FFFF0000h --- HPQ comment number 92 Page=75 Subtype=Highlight Subj=Highlight Author=relliott Comment= A SCSI device may contain only SCSI target ports, only SCSI initiator ports, or any combination of ports. s/b A SCSI device shall contain only SCSI target ports, only SCSI initiator ports, or any combination of ports. --- HPQ comment number 93 Page=76 Subtype=Highlight Subj=Highlight Author=relliott Comment=Move "a single" into each of a) b) and c) to improve readability --- HPQ comment number 94 Page=80 Subtype=Highlight Subj=Highlight Author=relliott Comment= application clients may not be able to distinguish between s/b application clients are not required to be able to distinguish between --- HPQ comment number 95 Page=81 Subtype=Text Subj=Note Author=relliott Comment= In figure 30 and its predecessors, add color to at least the logical unit boxes. --- HPQ comment number 96 Page=87 Subtype=Text Subj=Note Author=relliott Comment=Does power loss expected have any impact on CRN? --- HPQ comment number 97 Page=87 Subtype=Highlight Subj=Highlight Author=relliott Comment= A buffer containing This sentence should reference 5.4.3, where the buffer is described in more detail. --- HPQ comment number 98 Page=88 Subtype=Highlight Subj=Highlight Author=relliott Comment= output parameters s/b output arguments --- HPQ comment number 99 Page=88 Subtype=Text Subj=Note Author=relliott Comment= After first sentence in 5.2, add "CDB formats are defined in SPC-4." --- HPQ comment number 100 Page=88 Subtype=Highlight Subj=Highlight Author=relliott Comment= OPERATION CODE s/b OPERATION CODE field --- HPQ comment number 101 Page=88 Subtype=Highlight Subj=Highlight Author=relliott Comment= A buffer to contain This sentence should reference 5.4.3, where the buffer is described in more detail. --- HPQ comment number 102 Page=88 Subtype=Highlight Subj=Highlight Author=relliott Comment= Sense Data s/b Sense Data (see 5.8.6) The Data-In Buffer Size field description points to 5.4.3, so the Sense Data Length field description should point to 5.8.6. --- HPQ comment number 103 Page=88 Subtype=Highlight Subj=Highlight Author=relliott Comment= parameter s/b field --- HPQ comment number 104 Page=89 Subtype=Highlight Subj=Highlight Author=relliott Comment= An NACA bit s/b a NACA bit (last line of this paragraph uses "a naca" so it must be pronounced "a nak-ka") --- HPQ comment number 105 Page=89 Subtype=Highlight Subj=Highlight Author=relliott Comment= An NACA bit s/b A NACA bit --- HPQ comment number 106 Page=90 Subtype=Highlight Subj=Highlight Author=relliott Comment= Retry delay timer, when supported by a protocol, may provide the SCSI initiator port with more information on when the command should be retransmitted (see table 26). s/b Retry delay timer, when supported by a SCSI transport protocol, provides the SCSI initiator port with more information about when the command should be retransmitted (see table 26). (same comment on both BUSY and TASK SET FULL descriptions) --- HPQ comment number 107 Page=91 Subtype=Highlight Subj=Highlight Author=relliott Comment= Retry delay timer s/b either a) Retry delay time. Reason: the timer is the entity initialized to this value, not the value itself. or b) Additional status code. Reason: it is likely that this field will have a different meaning if used by any of the other status codes. A more generic name would be clearer. There can still be a "retry delay timer" that uses this code value for BUSY and TASK SET FULL. --- HPQ comment number 108 Page=91 Subtype=Text Subj=Note Author=relliott Comment= Some designs cannot return this information on a per I_T_L basis, but can return it on a per I_T basis. The target device should be able to return whichever scope it wants (perhaps with a "should" preferring the I_T_L scope). Either: a) Add a bit indicating scope (logical unit, target port, target device). This requires changing the transport protocols. b) redefine the code values: 0001h - 4FFFh wait for this logical unit (any I, any T, this L) 5000h - 9FFFh wait for this target port(any I, this T, any L) A000h - EFFFh wait for this target device (any I, any T, any L) F000h - FFEFh reserved The current maximum of FFEFh is 65519, so the current field supports 6551.9 seconds (109 minutes). Reducing that range by a third shouldn't overload a fabric with retries. Lack of results on a google search hints that this has not been widely implemented yet, so a change may still be viable. --- HPQ comment number 109 Page=91 Subtype=Text Subj=Note Author=relliott Comment= Replace the GOOD through TASK ABORTED rows with: All others 0000h - FFFFh Reserved That covers all the reserved status codes (table 25 defines 256 total codes). --- HPQ comment number 110 Page=91 Subtype=Highlight Subj=Highlight Author=relliott Comment= addition s/b additional --- HPQ comment number 111 Page=91 Subtype=Highlight Subj=Highlight Author=relliott Comment= busy s/b BUSY --- HPQ comment number 112 Page=91 Subtype=Highlight Subj=Highlight Author=relliott Comment= task set full s/b TASK SET FULL --- HPQ comment number 113 Page=92 Subtype=Highlight Subj=Highlight Author=relliott Comment= Status s/b status --- HPQ comment number 114 Page=92 Subtype=Highlight Subj=Highlight Author=relliott Comment= Status s/b status --- HPQ comment number 115 Page=96 Subtype=Highlight Subj=Highlight Author=relliott Comment= interactions s/b interactions for data transfers --- HPQ comment number 116 Page=97 Subtype=Highlight Subj=Highlight Author=relliott Comment= :A s/b : A --- HPQ comment number 117 Page=97 Subtype=Highlight Subj=Highlight Author=relliott Comment= :T s/b : T --- HPQ comment number 118 Page=98 Subtype=Highlight Subj=Highlight Author=relliott Comment= :T s/b : T --- HPQ comment number 119 Page=98 Subtype=Highlight Subj=Highlight Author=relliott Comment= :A s/b : A --- HPQ comment number 120 Page=98 Subtype=Highlight Subj=Highlight Author=relliott Comment= device server s/b device server or task manager allowing deletion of the first sentence in 5.4.3.4.1. --- HPQ comment number 121 Page=98 Subtype=StrikeOut Subj=Cross-Out Author=relliott Comment= The terminate data transfer request and confirmation may be used by a task manager to terminate partially completed transfers to the Data-In Buffer or from the Data-Out Buffer." Delete that and replace "device server" with "device server or task manager" in the next sentence --- HPQ comment number 122 Page=98 Subtype=Highlight Subj=Highlight Author=relliott Comment= device server s/b device server or task manager --- HPQ comment number 123 Page=98 Subtype=Highlight Subj=Highlight Author=relliott Comment= device server s/b device server or task manager --- HPQ comment number 124 Page=99 Subtype=Highlight Subj=Highlight Author=relliott Comment= application client task to interact with the task s/b application client task to represent the task --- HPQ comment number 125 Page=99 Subtype=Highlight Subj=Highlight Author=relliott Comment= application client task to interact with the task s/b application client task to represent the task --- HPQ comment number 126 Page=99 Subtype=Text Subj=Note Author=relliott Comment= Items b) through g) should be qualified with knowledge that the unit attention condition or service response was reported after the task arrived at the target port. Otherwise, it might still be in flight. This is the subtle ordering assumption in 4.3.3. --- HPQ comment number 127 Page=99 Subtype=Text Subj=Note Author=relliott Comment= This list is incomplete (not that it claims to be complete). Receiving unit attention condition about a reset, etc. - items b) through g) in the previous list - also apply here. This list might have originally been worded as the application client may send these TMFs to actively make the determination, but it's now worded too much like the previous list. --- HPQ comment number 128 Page=100 Subtype=Highlight Subj=Highlight Author=relliott Comment= hard resets, or logical unit resets s/b hard reset, or logical unit reset --- HPQ comment number 129 Page=100 Subtype=Highlight Subj=Highlight Author=relliott Comment= Background operations shall not be aborted by I_T nexus loss. Add "or power loss expected." --- HPQ comment number 130 Page=100 Subtype=Highlight Subj=Highlight Author=relliott Comment= Deferred errors should not be cleared by I_T nexus loss. Add "or power loss expected." --- HPQ comment number 131 Page=103 Subtype=Highlight Subj=Highlight Author=relliott Comment= "that had task(s) aborted" might be incorrect. Is the unit attention condition with COMMANDS CLEARED BY ANOTHER INITIATOR created for all I_T nexuses that were affected by the preempt, regardless of whether or not they actually had tasks aborted? Or is it only created for I_T nexuses that had one or more tasks aborted? --- HPQ comment number 132 Page=105 Subtype=Highlight Subj=Highlight Author=relliott Comment=ACA s/b smallcaps lowercase --- HPQ comment number 133 Page=105 Subtype=Highlight Subj=Highlight Author=relliott Comment=ACA s/b smallcaps lowercase --- HPQ comment number 134 Page=105 Subtype=Text Subj=Note Author=relliott Comment= Consider eliminating the 5.8 Command processing considerations level and upgrading each of the 5.8.x sections to 5.xx. The Unit Attention section, for example, is as important as 5.6 Aborting tasks. --- HPQ comment number 135 Page=106 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA task attribute s/b ACA (smallcaps lowercase) task attribute --- HPQ comment number 136 Page=106 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA task attribute s/b ACA (smallcaps lowercase) task attribute --- HPQ comment number 137 Page=106 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA task attribute s/b ACA (smallcaps lowercase) task attribute --- HPQ comment number 138 Page=108 Subtype=Highlight Subj=Highlight Author=relliott Comment=ACA s/b smallcaps lowercase --- HPQ comment number 139 Page=108 Subtype=Highlight Subj=Highlight Author=relliott Comment=ACA s/b smallcaps lowercase --- HPQ comment number 140 Page=108 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA task attribute s/b ACA (smallcaps lowercase) task attribute --- HPQ comment number 141 Page=109 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA s/b smallcaps lowercase --- HPQ comment number 142 Page=109 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA s/b smallcaps lowercase --- HPQ comment number 143 Page=109 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA s/b smallcaps lowercase --- HPQ comment number 144 Page=109 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA s/b smallcaps lowercase --- HPQ comment number 145 Page=109 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA task attribute s/b ACA (smallcaps lowercase) task attribute --- HPQ comment number 146 Page=109 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA task attribute s/b ACA (small caps lowercase) --- HPQ comment number 147 Page=109 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA task attribute s/b ACA (smallcaps lowercase) task attribute --- HPQ comment number 148 Page=110 Subtype=Highlight Subj=Highlight Author=relliott Comment= Incorrect logical unit selection s/b Incorrect logical unit numbers "select" is an ancient parallel SCSI term, and the logical unit number is what is incorrect, not the logical unit. --- HPQ comment number 149 Page=111 Subtype=Highlight Subj=Highlight Author=relliott Comment= Attention s/b attention --- HPQ comment number 150 Page=111 Subtype=Text Subj=Note Author=relliott Comment= The list of unit attention conditions includes some but not all those defined in SPC-4. What is the basis for including some reasons here? Item j) does serve as a catch-all, but perhaps all the non-SAM related items should be removed. Not covered include: - successful completion of a SET IDENTIFYING INFORMATION command that changes identifying information saved by the logical unit (see SPC-4) - On successful completion of a SET PRIORITY command or change to the mode page - On successful completion of a SET TIMESTAMP command - If the ETC bit is set to one and the result of the comparison is true (log parameters) - block descriptor values changed - informational exceptions --- HPQ comment number 151 Page=111 Subtype=Highlight Subj=Highlight Author=relliott Comment= A removable medium may have been changed; s/b A removable medium has possibly been changed --- HPQ comment number 152 Page=111 Subtype=Text Subj=Note Author=relliott Comment=Incorporate 07-459 Unit attention queuing --- HPQ comment number 153 Page=111 Subtype=Highlight Subj=Highlight Author=relliott Comment= perform s/b process --- HPQ comment number 154 Page=111 Subtype=Highlight Subj=Highlight Author=relliott Comment= perform s/b process --- HPQ comment number 155 Page=112 Subtype=Highlight Subj=Highlight Author=relliott Comment= perform s/b process --- HPQ comment number 156 Page=115 Subtype=Text Subj=Note Author=relliott Comment= The row "Power loss expected/COMMANDS CLEARED BY POWER LOSS NOTIFICATION" needs to move down in table 36 to be below (higher specificity) than I_T nexus loss. Section 6.3.5 says it simply aborts tasks; it doesn't wipe out background operations, clear deferred errors, etc. This means it has less impact than a hard reset, logical unit reset, and a set of I_T nexus losses (it has more impact than a single I_T nexus loss...). If the target device experiences hard reset, logical unit reset, or I_T nexus loss, it is not an acceptable substitute to only report COMMANDS CLEARED BY POWER LOSS NOTIFICATION, which its current position in the table endorses. --- HPQ comment number 157 Page=116 Subtype=Highlight Subj=Highlight Author=relliott Comment= hard reset condition s/b hard reset condition (see 6.3.2) --- HPQ comment number 158 Page=116 Subtype=Highlight Subj=Highlight Author=relliott Comment= may use the I_T NEXUS LOSS OCCURRED additional sense code s/b should use ... --- HPQ comment number 159 Page=116 Subtype=Highlight Subj=Highlight Author=relliott Comment= when establishing a unit attention condition s/b when establishing a unit attention condition for an I_T nexus loss --- HPQ comment number 160 Page=117 Subtype=Text Subj=Note Author=relliott Comment= In 6.3,4, add "abort all task management functions received on the I_T nexus;" --- HPQ comment number 161 Page=117 Subtype=Text Subj=Note Author=relliott Comment=In 6.3.3, add "abort all task management functions;" --- HPQ comment number 162 Page=118 Subtype=Text Subj=Note Author=relliott Comment= Does power loss expected also abort task management functions? I think it should do so. If so, then the comparison to "CLEAR TASK SET for all task sets" is incomplete, and the a) b) list needs to be expanded to include "abort all task management functions;" If not, then rules in 7.11 about task management function lifetimes are incorrect. --- HPQ comment number 163 Page=118 Subtype=Highlight Subj=Highlight Author=relliott Comment= protocol standards s/b SCSI transport protocol standards --- HPQ comment number 164 Page=120 Subtype=Highlight Subj=Highlight Author=relliott Comment= the following procedure call s/b a procedure call using the following format: --- HPQ comment number 165 Page=120 Subtype=Highlight Subj=Highlight Author=relliott Comment= Introduction s/b Task management function procedure calls (to parallel the section heading of 5.1 for commands) --- HPQ comment number 166 Page=120 Subtype=Highlight Subj=Highlight Author=relliott Comment= (IN ( nexus ), OUT ( [additional response information] ) s/b (IN ( Nexus ), OUT ( [Additional Response Information] ) --- HPQ comment number 167 Page=120 Subtype=Text Subj=Note Author=relliott Comment= add the following after the Service Response = line: where: Function Name is one of the task management function names listed in table 34 Nexus is either: a) an I_T Nexus argument; b) an I_T_L Nexus Argument; or c) an I_T_L_Q Nexus argument Additional Response Information is the Additional Response Information output argument described below --- HPQ comment number 168 Page=120 Subtype=Highlight Subj=Highlight Author=relliott Comment= Nexus: An I_T nexus, I_T_L nexus, or I_T_L_Q nexus (see 4.7) identifying the task or tasks affected by the task management function. I_T Nexus: A SCSI initiator port and SCSI target port nexus (see 4.7). I_T_L Nexus: A SCSI initiator port, SCSI target port, and logical unit nexus (see 4.7). I_T_L_Q Nexus: A SCSI initiator port, SCSI target port, logical unit, and task identifier nexus (see 4.7). s/b I_T Nexus: The I_T nexus (see 4.7) affected by the task management function. I_T_L Nexus: The I_T_L nexus (see 4.7) affected by the task management function. I_T_L_Q Nexus: The I_T_L_Q nexus (see 4.7) affected by the task management function. --- HPQ comment number 169 Page=121 Subtype=Highlight Subj=Highlight Author=relliott Comment= Request s/b Procedure call: (this is at the same level as Execute Command, not the same level as Send SCSI Command) --- HPQ comment number 170 Page=121 Subtype=Text Subj=Note Author=relliott Comment= At end of SERVICE DELIVERY OR TARGET FAILURE description, add "All output arguments are invalid." That means Additional Response Information is not usable. --- HPQ comment number 171 Page=121 Subtype=Highlight Subj=Highlight Author=relliott Comment= One of the following SCSI transport protocol specific service responses shall be returned s/b Service Response assumes one of the following values to match wording in section 5.1 for Execute Command. Wording could be changed in both places if "assumes" is not agreeable. There is no need for the Service Response values to be determined by the transport protocol here. When included in RESPONSE frames over the wire, they are; when returned from the initiator port to the application client, the values are probably remapped to protocol-independent values (so generic SCSI software isn't affected by the transport protocol choice). --- HPQ comment number 172 Page=122 Subtype=Highlight Subj=Highlight Author=relliott Comment= ACA task attribute s/b ACA (smallcaps lowercase) task attribute --- HPQ comment number 173 Page=122 Subtype=Highlight Subj=Highlight Author=relliott Comment= Request s/b Procedure call: (this is at the same level as Execute Command, not the same level as Send SCSI Command) --- HPQ comment number 174 Page=122 Subtype=Highlight Subj=Highlight Author=relliott Comment= Request s/b Procedure call: (this is at the same level as Execute Command, not the same level as Send SCSI Command) --- HPQ comment number 175 Page=122 Subtype=Highlight Subj=Highlight Author=relliott Comment= Request s/b Procedure call: (this is at the same level as Execute Command, not the same level as Send SCSI Command) --- HPQ comment number 176 Page=123 Subtype=Highlight Subj=Highlight Author=relliott Comment= Request s/b Procedure call: (this is at the same level as Execute Command, not the same level as Send SCSI Command) --- HPQ comment number 177 Page=123 Subtype=Highlight Subj=Highlight Author=relliott Comment= Request s/b Procedure call: (this is at the same level as Execute Command, not the same level as Send SCSI Command) --- HPQ comment number 178 Page=123 Subtype=Highlight Subj=Highlight Author=relliott Comment= Request s/b Procedure call: (this is at the same level as Execute Command, not the same level as Send SCSI Command) --- HPQ comment number 179 Page=124 Subtype=Highlight Subj=Highlight Author=relliott Comment= Request s/b Procedure call: (this is at the same level as Execute Command, not the same level as Send SCSI Command) --- HPQ comment number 180 Page=124 Subtype=Highlight Subj=Highlight Author=relliott Comment= Request s/b Procedure call: (this is at the same level as Execute Command, not the same level as Send SCSI Command) --- HPQ comment number 181 Page=125 Subtype=Highlight Subj=Highlight Author=relliott Comment= application client task to interact with the task management function s/b application client task management function to represent the task management function --- HPQ comment number 182 Page=125 Subtype=Text Subj=Note Author=relliott Comment= For commands (5.5), a service response of SERVICE DELIVERY OR TARGET FAILURE leaves the application client task in existence until the initiator receives something else from the target that assures it is gone (a response to a TMF aborting that task). Task management functions should be handled the same way. It is not safe to reuse the task identifier (task tag) if a SERVICE DELIVERY OR TARGET FAILURE is returned. The task management function should be assumed to exist until an I_T NEXUS RESET or LOGICAL UNIT RESET is successfully run (or a unit attention occurs reporting a reset). --- HPQ comment number 183 Page=125 Subtype=Highlight Subj=Highlight Author=relliott Comment= FUNCTION COMPLETE, FUNCTION SUCCEEDED, FUNCTION REJECTED, or SERVICE DELIVERY OR TARGET FAILURE 1. Add INCORRECT LOGICAL UNIT NUMBER 2. If SERVICE DELIVERY OR TARGET FAILURE remains in this list, then there is no reason to list all of them - any service response suffices, so delete the list and just leave "A service response is received". See other comment about excluding SERVICE DELIVERY OR TARGET FAILURE, though. --- HPQ comment number 184 Page=125 Subtype=Text Subj=Note Author=relliott Comment= Add "power loss expected (see 6.3.5)" to the list of things that cause a task management function to no longer exist. Make it item b) ahead of I_T nexus loss. --- HPQ comment number 185 Page=126 Subtype=Highlight Subj=Highlight Author=relliott Comment= (see 7.12.5) confirmation s/b confirmation (see 7.12.5) --- HPQ comment number 186 Page=126 Subtype=Highlight Subj=Highlight Author=relliott Comment= Send Task Management Request s/b Send Task Management Request request --- HPQ comment number 187 Page=126 Subtype=Text Subj=Note Author=relliott Comment= Items b) c) and d) should be qualified with knowledge that the unit attention condition was reported after the task management request arrived at the target port. Otherwise, it might still be in flight. This is the subtle ordering assumption in 4.3.3. --- HPQ comment number 188 Page=127 Subtype=Highlight Subj=Highlight Author=relliott Comment= task manger s/b task manager --- HPQ comment number 189 Page=127 Subtype=Highlight Subj=Highlight Author=relliott Comment= (IN ( Nexus, Service Response )) s/b (IN ( Nexus, Service Response, [Additional Response Information] )) with this added to Input arguments: Additional Response Information: The Additional Response Information output argument for the task management procedure call (see 7.1): --- HPQ comment number 190 Page=127 Subtype=Highlight Subj=Highlight Author=relliott Comment= (IN ( Nexus, Service Response )) s/b (IN ( Nexus, Service Response, [Additional Response Information] )) with this added to Input arguments: Additional Response Information: The Additional Response Information output argument for the task management procedure call (see 7.1): --- HPQ comment number 191 Page=128 Subtype=Highlight Subj=Highlight Author=relliott Comment= Application Client Task s/b Application Client Task Management Function --- HPQ comment number 192 Page=128 Subtype=Highlight Subj=Highlight Author=relliott Comment= application client task s/b application client --- HPQ comment number 193 Page=129 Subtype=Highlight Subj=Highlight Author=relliott Comment= application client task s/b application client --- HPQ comment number 194 Page=129 Subtype=Highlight Subj=Highlight Author=relliott Comment= service response argument s/b Service Response argument or just replace "service response argument is set to a value of" with "service response of" (this is how most of the standard is worded) --- HPQ comment number 195 Page=130 Subtype=StrikeOut Subj=Cross-Out Author=relliott Comment= Delete ", or no task attribute." and move the "or" to earlier in the sentence. SAM-4 requires each task have a task attribute. --- HPQ comment number 196 Page=133 Subtype=Text Subj=Note Author=relliott Comment= When referring to "(simple, ordered, head of queue, and ACA) task", Mixed Case should be used, not lowercase. --- HPQ comment number 197 Page=134 Subtype=Highlight Subj=Highlight Author=relliott Comment= A priority may be assigned s/b A priority is assigned --- HPQ comment number 198 Page=147 Subtype=Highlight Subj=Highlight Author=relliott Comment= Information s/b Additional Response Information --- HPQ comment number 199 Page=148 Subtype=Highlight Subj=Highlight Author=relliott Comment= Change "SAM-3" to "previous versions of this standard" or add SAM-3 as a normative reference in 2.1. --- HPQ comment number 200 Page=148 Subtype=Highlight Subj=Highlight Author=relliott Comment= Change "SAM-4 to SAM-3 terminology mapping" to "Terminology mapping to previous versions of this standard" or add SAM-3 as a normative reference in 2.1. --- HPQ comment number 201 Page=148 Subtype=Highlight Subj=Highlight Author=relliott Comment= Change "SAM-3 term" to "Term used in previous versions of this standard" or add SAM-3 as a normative reference in 2.1. --- HPQ comment number 202 Page=148 Subtype=Highlight Subj=Highlight Author=relliott Comment= SAM-4 equivalent term s/b Term used in this standard --- HPQ comment number 203 Page=148 Subtype=Text Subj=Note Author=relliott Comment= In table C.1, center the left column and left-justify the right column (including the headers) --- HPQ comment number 204 Page=148 Subtype=Highlight Subj=Highlight Author=relliott Comment= Terminology mapping s/b Terminology mapping to previous versions of this standard ************************************************************** Comments attached to Yes ballot from Frederick Knight of Network Appliance: A minor comment relative to Unit Attention precedence on Incorrect Logical Units was posted at http://www.t10.org/ftp/t10/document.07/07-450r0.pdf ************************************************************** Comments attached to Yes ballot from Paul Suhler of Quantum Corp.: Quantum comment number 1 Page=1 Subtype=Text Author=suhlerp Comment= Quantum Corporation Letter Ballot Comments Changes are shown as comments and/or edits. Edits may be easy to miss. suhlerp - Paul Suhler Quantum comment number 2 Page=18 Subtype=StrikeOut Author=suhlerp Comment= Quantum comment number 3 Page=18 Subtype=Caret Author=suhlerp Comment=lists [don't capitalize, for consistency] Quantum comment number 4 Page=22 Subtype=Text Author=suhlerp Comment= ADC-2 will probably have completed INCITS approval by the time SAM-4 finishes LB comment resolution. Quantum comment number 5 Page=26 Subtype=Text Author=suhlerp Comment= Should there be a definition for "logical unit name" (see 4.5.19.3)? A designator can be associated with a logical unit. 3.1.x logical unit name: A name (see 3.1.68) of a logical unit that is world wide unique within the SCSI transport protocol of a SCSI domain in which the SCSI device containing the logical unit has SCSI ports (see 4.5.4.2). The logical unit name may be made available to other SCSI devices or SCSI ports in SCSI transport protocol specific ways. Quantum comment number 6 Page=26 Subtype=StrikeOut Author=suhlerp Comment= Quantum comment number 7 Page=26 Subtype=Caret Author=suhlerp Comment=Shows [capitalize] Quantum comment number 8 Page=30 Subtype=Text Author=suhlerp Comment= ADT-2 Automation/Drive Interface Transport Protocol - 2 (see 1.3) [used in table A.3] Quantum comment number 9 Page=37 Subtype=StrikeOut Author=suhlerp Comment= Quantum comment number 10 Page=37 Subtype=Caret Author=suhlerp Comment=Attribute Quantum comment number 11 Page=37 Subtype=StrikeOut Author=suhlerp Comment= Quantum comment number 12 Page=37 Subtype=Caret Author=suhlerp Comment=Attribute Quantum comment number 13 Page=37 Subtype=StrikeOut Author=suhlerp Comment= Quantum comment number 14 Page=37 Subtype=Caret Author=suhlerp Comment=Attribute Quantum comment number 15 Page=37 Subtype=StrikeOut Author=suhlerp Comment= Quantum comment number 16 Page=37 Subtype=Caret Author=suhlerp Comment=Attribute Quantum comment number 17 Page=37 Subtype=Text Author=suhlerp Comment= Page break before this paragraph. All the other notational elements start on a new page. Quantum comment number 18 Page=38 Subtype=Text Author=suhlerp Comment= Does this mean that the value of the attribute is the set {x,y,z} ? Maybe an example would help. Quantum comment number 19 Page=46 Subtype=Text Author=suhlerp Comment= [Technical] Both ADC-2 and SSC-2 & -3 include a physical device as part of the SCSI target device. If you wish, I could provide a proposal to add this. Otherwise, it could wait for SAM-5. Quantum comment number 20 Page=51 Subtype=Text Author=suhlerp Comment= Why "logical unit task manager"? Why not "task manager," as in a) above? This is the only use of "logical unit task manager." Quantum comment number 21 Page=61 Subtype=Text Author=suhlerp Comment=SPC-4 ? Quantum comment number 22 Page=63 Subtype=Text Author=suhlerp Comment= Should this say "receives a task or a task management function specifying a W-LUN" ? Quantum comment number 23 Page=64 Subtype=Text Author=suhlerp Comment= Is the use of "SCSI device" rather than "SCSI target device" in the first two sentences intentional? If so, then is the implication that a SCSI initiator device shall be able to issue commands to LUN 0? Or should both sentences be changed to "SCSI target device"? Quantum comment number 24 Page=79 Subtype=Text Author=suhlerp Comment=Too much white space here Quantum comment number 25 Page=88 Subtype=StrikeOut Author=suhlerp Comment= Quantum comment number 26 Page=88 Subtype=Caret Author=suhlerp Comment=determinant Quantum comment number 27 Page=99 Subtype=Text Author=suhlerp Comment= [Technical] How about ... d) Completion of a QUERY TASK task management function specifying the task with a service response of FUNCTION COMPLETE; or e) Completion of a QUERY TASK SET task management function specifying the task set [more words needed?] with a service response of FUNCTION COMPLETE. Quantum comment number 28 Page=99 Subtype=Text Author=suhlerp Comment= [Technical] How about ... h) A service response of FUNCTION COMPLETE following a QUERY TASK task management function directed to the specified task; or i) A service response of FUNCTION COMPLETE following a QUERY TASK SET task management function directed to the specified task set. ************************************************************** Comments attached to No ballot from Mark Evans of Western Digital: Western Digital Corporation Comments on the SCSI Architecture Model - 4 draft standard letter ballot Submitted by Mark Evans of Western Digital Corporation Western Digital Corporation #1 PDF page 20 Since text takes precedence in this standard, and text should introduce figures, tables, etc., I recommend moving the following paragraph above figure 1: "As shown in figure 1, all SCSI implementation standards shall reflect the generic requirements defined herein. In addition, an implementation claiming SCSI compliance shall conform to the applicable implementation requirements defined in this standard and the appropriate SCSI implementation standards. In the event of a conflict between this document and other SCSI standards under the jurisdiction of technical committee T10, the requirements of this standard shall apply." Western Digital Corporation #2 PDF page 20 I recommend that at least one "e.g." be added either in the four rows in figure 2, in the descriptive text that follows, or both. See SBC-3 for an example. Western Digital Corporation #3 PDF page 22 I'm not sure that, "The following standards...", is correct because I thought the OMG document was a specification - not a standard. If this is true, then this should be changed to, "The following standards and specifications...". Western Digital Corporation #4 PDF page 22 "Copies of the following documents..." I think that some of the following documents are not available from ANSI (e.g., the IETF and OMG documents as noted below), therefore, I think this... s/b "Except where noted in the following subclauses, copies of the following documents..." Western Digital Corporation #5 PDF page 23 The definitions (i.e., those with the numbers 3.1.x) are all left and right justified, and several have hyphens added by FrameMaker to split words across lines. All of the other clauses are left justified and right ragged with no hyphens added to split words. At a quick glance, SPC-4 is left and right justified with hyphens added throughout. SBC-3 and SAS- 2 are left justified and right ragged with no hyphens added throughout. Interesting to me, the style guide (i.e., 05-085r7) has the same odd combination as SAM-4. I wonder why that is? One way or the other, I think that all the clauses should have the same format - editor's choice. Western Digital Corporation #6 PDF page 23 "see 3.1.8": This cross reference hot link is broken. When I click on it, nothing happens. This is true for almost all of the "3.1.x" cross references in this subclause and for some of them in other clauses. Western Digital Corporation #7 PDF page 23 Add "(see 8.5.3)" at the end of 3.1.10. Western Digital Corporation #8 PDF page 23 Add "(see 8.9.1)" at the end of 3.1.11. Western Digital Corporation #9 PDF page 23 Add "(see 4.3)" at the end of 3.1.15. Western Digital Corporation #10 PDF page 23 "...possibly a member of a series of defined numeric values," This phrase tells me nothing. Remove the "ly" adverb and add an "e.g.", and then it tells me something. s/b "...sometimes a member of a series of defined numeric values (e.g., an additional sense code)," Western Digital Corporation #11 PDF page 24 "(e.g., block, stream)." s/b "(e.g., a block device or a stream device)." Western Digital Corporation #12 PDF page 24 "...a service delivery subsystem." s/b "...the service delivery subsystem." Western Digital Corporation #13 PDF page 24 Add "(see 8.5.4)" at the end of 3.1.35. Western Digital Corporation #14 PDF page 24 Add "(see 8.5.2)" at the end of 3.1.36. Western Digital Corporation #15 PDF page 26 "...for transmission, but not yet received." s/b "...for transmission, but has not yet arrived at the intended recipient." Western Digital Corporation #16 PDF page 26 "...in an object diagram. Represents an instance..." s/b "...in an object diagram representing an instance..." Western Digital Corporation #17 PDF page 26 Delete "e.g.,. the terms name and world wide identifier (WWID) may be interchangeable)" as neither the terms "world wide identifier" or "WWID" are used anywhere else in this document. Alternately, those terms could be defined, but that would be a little odd since they are only used in this definition. Western Digital Corporation #18 PDF page 26 "From the point of view of the application client, the description of command between the time that the application client calls the Send SCSI Command SCSI transport protocol service and the time one of the SCSI target device responses described in 5.5 is received." s/b "From the point of view of the application client, a command from the time that the application client calls the Send SCSI Command SCSI transport protocol service until the application client receives a response for the command from the SCSI target device (see 5.5)." Western Digital Corporation #19 PDF page 27 "...usually according to the temporal order..." s/b "...most often according to the temporal order..." Western Digital Corporation #20 PDF page 29 "The portion (i.e, Q) of an I_T_L nexus (see 3.1.49) in a task set (see 3.1.128). See 4.7.2." s/b "The portion of an I_T_L_Q nexus (i.e., the Q) that is the numerical identifier of the task in the nexus (see 3.1.49) in a task set (see 3.1.128). See 4.7.2." Western Digital Corporation #21 PDF page 29 Add "See clause 7." at the end of 3.1.123. Western Digital Corporation #22 PDF page 29 "...and send command(s) to another SCSI target device." s/b "...and send one or more commands to another SCSI target device." Western Digital Corporation #23 PDF page 30 "...preference (synonymous with may or may not)." s/b "...preference. May is synonymous with the phrase "may or may not"." Western Digital Corporation #24 PDF page 30 "...preference (synonymous with may or may not)." s/b "...preference. May not is synonymous with the phrase "may or may not"." Western Digital Corporation #25 PDF page 30 3.3.7 prohibited: This definition should be deleted as it is not used in this standard. Western Digital Corporation #26 PDF page 31 The paragraph that begins, "Lists sequenced by letters...", s/b ...replaced by something more complete (e.g., how lists are described in the SCSI style guide). Western Digital Corporation #27 PDF page 35 "Solid lines with arrowheads (see figure 4) are the notation..." s/b "...is the notation..." ["notation" meaning, "...any particular system of characters or symbols used to briefly express elements...", and this is my first choice] or "...are the notations..." ["notations" meaning, "...the characters or symbols used in such a system..."] One way or the other, the verb has to agree with the object (i.e., either "is the notation" or "are the notations"). Western Digital Corporation #28 PDF page 36 "Solid lines with diamonds (see figure 5) are the notation..." s/b "...is the notation..." [see my earlier comment] Western Digital Corporation #29 PDF page 37 "Solid lines with triangles (see figure 6) are the notation..." s/b "...is the notation..." [see my earlier comment] Western Digital Corporation #30 PDF page 37 "Dashed lines with arrowheads (see figure 7) are the notation..." s/b "...is the notation..." [see my earlier comment] Western Digital Corporation #31 PDF page 38 "Solid lines (see figure 9) are the notation..." s/b "...is the notation..." [see my earlier comment] Western Digital Corporation #32 PDF page 40 "...transitions, using..." s/b "...transitions using..." Western Digital Corporation #33 PDF page 40 "Using figure 10 as an example, the transition list might read as follows:" s/b "Using figure 10 as an example, the transition list reads as follows:" Western Digital Corporation #34 PDF page 40 "State transitions are logically instantaneous;" s/b "Transitions from one state to another are instantaneous;" Western Digital Corporation #35 PDF page 43 "...the response has been received successfully..." s/b "...the response has been received without error..." Western Digital Corporation #36 PDF page 43 "...until the response has been successfully delivered..." s/b "...until the response has been delivered without error..." Western Digital Corporation #37 PDF page 47 "Each instance of a SCSI Domain class shall contain the following objects: a) one service delivery subsystem; b) one or more SCSI devices; and c) one or more SCSI ports." s/b I think there shall be two or more SCSI devices and two or more SCSI ports, as shown in figure 16. Western Digital Corporation #38 PDF page 48 Delete "persistently". Western Digital Corporation #39 PDF page 50 Delete "persistently". Western Digital Corporation #40 PDF page 51 Delete "persistently". Western Digital Corporation #41 PDF page 52 Delete "originates commands by issuing Send SCSI Command requests (see 5.4.2)." s/b I don't know where this goes, but it doesn't go here. Western Digital Corporation #42 PDF page 54 "The application client task persists until a task complete response is sent..." s/b "The application client task persists until a task complete response is received..." Western Digital Corporation #43 PDF page 60 Delete "persistently". Western Digital Corporation #44 PDF page 61 "c) Data Transfer Terminated operation (see 5.4.3.4.3) to determines..." s/b "c) Data Transfer Terminated operation (see 5.4.3.4.3) to determine..." Western Digital Corporation #45 PDF page 64 "When an application client displays or otherwise makes a 64-bit LUN value visible to a user, it should display it in hexadecimal format..." s/b "When an application client displays or otherwise makes a 64-bit LUN value visible, the application client should display the value in hexadecimal format..." Western Digital Corporation #46 PDF page 64 "...an application client may display it as a single..." s/b "...an application client may display the value as a single..." Western Digital Corporation #47 PDF page 64 "...an application client should display it as a single..." s/b "...an application client should display the value as a single..." Western Digital Corporation #48 PDF page 65 "All logical unit number structure fields beyond byte 1 shall be zero (see table 3)." s/b "Byte 2 through byte 7 in an 8-byte single level logical unit number structure using the peripheral device addressing method shall contain 00h (see table 3)." Western Digital Corporation #49 PDF page 65 "The 00b in the ADDRESS METHOD field specifies peripheral device addressing (see 4.6.6) and the 00h in the BUS IDENTIFIER field specifies the current level (see 4.6.7)." s/b "A value of 00b in the ADDRESS METHOD field specifies peripheral device addressing (see 4.6.6). A value of 00h in the in the BUS IDENTIFIER field specifies the current level (see 4.6.7)." Western Digital Corporation #50 PDF page 65 "All logical unit number structure fields beyond byte 1 shall be zero (see table 4)." s/b "Byte 2 through byte 7 in an 8-byte single level logical unit number structure using the flat space addressing method shall contain 00h (see table 4)." Western Digital Corporation #51 PDF page 65 "The 01b in the ADDRESS METHOD field specifies flat space addressing (see 4.6.8) at the current level." s/b "A value of 01b in the ADDRESS METHOD field specifies flat space addressing (see 4.6.8) at the current level." Western Digital Corporation #52 PDF page 66 "All logical unit number structure fields beyond byte 3 shall be zero (see table 5)." s/b "Byte 4 through byte 11 in an 12-byte single level logical unit number structure using the extended flat space addressing method shall contain 00h (see table 3)." Western Digital Corporation #53 PDF page 66 "The 11b in the ADDRESS METHOD field with a 2h in the EXTENDED ADDRESS METHOD field specifies extended flat space addressing (see 4.6.12) at the current level. The 01b in the LENGTH field specifies that the LUN specified in the EXTENDED FLAT SPACE ADDRESS field is three bytes in length." s/b "A value of 11b in the ADDRESS METHOD field with a value of 2h in the EXTENDED ADDRESS METHOD field specifies extended flat space addressing (see 4.6.12) at the current level. A value of 01b in the LENGTH field specifies that the LUN specified in the EXTENDED FLAT SPACE ADDRESS field is three bytes in length." Western Digital Corporation #54 PDF page 67 "N/A" s/b "n/a" Western Digital Corporation #55 PDF page 69 "If the SCSI device does not relay any commands or task management functions to the addressed dependent logical unit, it shall follow the rules for selection of incorrect logical units described in 5.8.4 and 7.12." s/b "If the SCSI device does not relay any commands or task management functions to the addressed dependent logical unit, then the SCSI device shall follow the rules for selection of incorrect logical units described in 5.8.4 and 7.12." Western Digital Corporation #56 PDF page 69 "If the SCSI device does relay some commands and task management functions to the addressed dependent logical unit, it shall:" s/b "If the SCSI device does relay some commands and task management functions to the addressed dependent logical unit, then the SCSI device shall:" Western Digital Corporation #57 PDF page 69 "However, bus identifier zero shall specify that the command or task management function is to be relayed to a logical unit within the SCSI device at the current level." s/b "However, if the BUS IDENTIFIER field is set to 00h, then the command or task management function shall be relayed to a logical unit within the SCSI device at the current level." Western Digital Corporation #58 PDF page 70 "If the SCSI device does not relay any commands or task management functions to the addressed dependent logical unit, it shall follow the rules for selection of incorrect logical units described in 5.8.4 and 7.12." s/b "If the SCSI device does not relay any commands or task management functions to the addressed dependent logical unit, then the SCSI device shall follow the rules for selection of incorrect logical units described in 5.8.4 and 7.12." Western Digital Corporation #59 PDF page 70 "If the SCSI device does relay some commands and task management functions to the addressed dependent logical unit, it shall:" s/b "If the SCSI device does relay some commands and task management functions to the addressed dependent logical unit, then the SCSI device shall:" Western Digital Corporation #60 PDF page 77 "Two-way communications shall be possible between all logical units and all SCSI target ports, however, communications between any logical unit and any SCSI target port may be inactive. Two-way communications shall be available between each task manager and all task routers. Each SCSI target port shall accept commands sent to LUN 0 or the REPORT LUNS well-known logical unit and the task router shall route them to a device server for processing. REPORT LUNS commands (see SPC-4) shall be accepted by the logical unit with the logical unit number zero or the REPORT LUNS well-known logical unit from any SCSI target port and shall return the logical unit inventory available via that SCSI target port. The availability of the same logical unit through multiple SCSI target ports is discovered by matching logical unit name values in the INQUIRY command Device Identification VPD page (see SPC-4)." s/b "Two-way communications shall be possible between all logical units and all SCSI target ports in a SCSI target device. However, communications between any logical unit and any SCSI target port in a SCSI target device may be inactive. Two-way communications shall be available between each task manager and all task routers in the SCSI target ports in the SCSI target device. Each SCSI target port in a SCSI target device shall accept commands sent to LUN 0 or the REPORT LUNS well-known logical unit, and the task router in that SCSI target port shall route the commands to a device server in a logical unit in the SCSI target device for processing. REPORT LUNS commands (see SPC-4) shall be accepted by the logical unit with the logical unit number zero or the REPORT LUNS well-known logical unit from any SCSI target port in the SCSI target device, and the logical unit shall return the logical unit inventory available via that SCSI target port. An application client determines the availability of the same logical unit through multiple SCSI target ports in a SCSI target device by matching logical unit name values in the Device Identification VPD page (see SPC-4)." Western Digital Corporation #61 PDF page 79 "Two-way communications shall be possible between all logical units and all SCSI target ports, however, communications between any logical unit and any SCSI target port may be inactive. Two-way communications shall be available between each task manager and all task routers. Each SCSI target port shall accept commands sent to LUN 0 or the REPORT LUNS well-known logical unit and the task router shall route them to a device server for processing. REPORT LUNS commands (see SPC-4) shall be accepted by the logical unit with the logical unit number zero or the REPORT LUNS well-known logical unit from any SCSI target port and shall return the logical unit inventory available via that SCSI target port. The availability of the same logical unit through multiple SCSI target ports is discovered by matching logical unit name values in the INQUIRY command Device Identification VPD page (see SPC-4)." s/b "Two-way communications shall be possible between all logical units and all SCSI target ports in a SCSI device. However, communications between any logical unit and any SCSI target port in a SCSI device may be inactive. Two-way communications shall be available between each task manager and all task routers in the SCSI target ports in the SCSI device. Each SCSI target port in a SCSI device shall accept commands sent to LUN 0 or the REPORT LUNS well-known logical unit, and the task router in that SCSI target port shall route the commands to a device server in a logical unit in the SCSI device for processing. REPORT LUNS commands (see SPC-4) shall be accepted by the logical unit with the logical unit number zero or the REPORT LUNS well-known logical unit from any SCSI target port in the SCSI device, and the logical unit shall return the logical unit inventory available via that SCSI target port. An application client determines the availability of the same logical unit through multiple SCSI target ports in a SCSI device by matching logical unit name values in the Device Identification VPD page (see SPC-4)." Western Digital Corporation #62 PDF page 79 "A SCSI target device may be connected to multiple SCSI domains such that a SCSI initiator port is only able to communicate with its logical units using a single SCSI target port." s/b "A SCSI target device may have SCSI target ports connected to different SCSI domains such that a SCSI initiator port is only able to communicate with the logical units in the SCSI target device using the SCSI target ports in a single SCSI domain." Western Digital Corporation #63 PDF page 82 "This standard does not require a SCSI target device to have the ability to detect the presence of a SCSI initiator device with multiple SCSI initiator ports. Therefore, a SCSI target device handles a SCSI initiator device with multiple SCSI initiator ports exactly as it would handle multiple separate SCSI initiator devices (e.g., a SCSI target device handles the configurations shown in figure 29 and figure 30 in exactly the same way it handles the configuration shown in figure 28)." s/b "This standard does not require a SCSI target device to be able to detect that a SCSI initiator device contains more than one SCSI initiator port. In the cases where a SCSI target device does not detect that a SCSI initiator device contains more than one SCSI initiator port, the SCSI target device interacts with the SCSI initiator device as if each SCSI initiator port was contained in a separate SCSI initiator device (e.g., a SCSI target device operates in the configurations shown in figure 29 and figure 30 in the same way it operates in the configuration shown in figure 28)." Western Digital Corporation #64 PDF page 89 "This status indicates that the device server has successfully completed the task." s/b "This status indicates that the device server has completed the task without error." Western Digital Corporation #65 PDF page 90 "If the UA_INTLCK_CTRL field in the Control mode page contains 11b (see SPC-4), termination of a command with BUSY status shall cause a unit attention condition to be established for the SCSI initiator port that sent the command with an additional sense code set to PREVIOUS BUSY STATUS unless a PREVIOUS BUSY STATUS unit attention condition already exists." s/b "If the UA_INTLCK_CTRL field in the Control mode page contains 11b (see SPC-4), then completion of a command with BUSY status shall cause a unit attention condition to be established for the I_T nexus on which the command was received with an additional sense code set to PREVIOUS BUSY STATUS unless a PREVIOUS BUSY STATUS unit attention condition already exists." Western Digital Corporation #66 PDF page 90 "Retry delay timer," s/b "The retry delay timer," Western Digital Corporation #67 PDF page 90 "This status shall be returned whenever a command attempts to access a logical unit in a way that conflicts with an existing reservation." s/b "This status shall be returned whenever a command is directed by an application client to access a logical unit in a way that conflicts with an existing reservation." Western Digital Corporation #68 PDF page 90 "If the UA_INTLCK_CTRL field in the Control mode page contains 11b (see SPC-4), termination of a command with RESERVATION CONFLICT status shall cause a unit attention condition to be established for the SCSI initiator port that sent the command with an additional sense code set to PREVIOUS RESERVATION CONFLICT STATUS unless a PREVIOUS RESERVATION CONFLICT STATUS unit attention condition already exists." s/b "If the UA_INTLCK_CTRL field in the Control mode page contains 11b (see SPC-4), then completion of a command with RESERVATION CONFLICT status shall cause a unit attention condition to be established for the I_T nexus on which the command was received with an additional sense code set to PREVIOUS RESERVATION CONFLICT STATUS unless a PREVIOUS RESERVATION CONFLICT STATUS unit attention condition already exists." Western Digital Corporation #69 PDF page 90 "...prevents accepting a received task from that I_T nexus into the task set," s/b "...prevents the logical unit from accepting an additional task received from that I_T nexus into the task set," Western Digital Corporation #70 PDF page 90 "(i.e., for each SCSI target port, allow at least one command from each SCSI initiator port that has identified itself to the SCSI target port in a SCSI transport protocol specific manner (e.g., login), or by the successful transmission of a c command)." s/b "(i.e., a logical unit should allow at least one command into the task set for any I_T nexus that has been identified in a SCSI transport protocol specific manner (e.g., a login), or by the successful reception of a command)." Western Digital Corporation #71 PDF page 90 "Retry delay timer," s/b "The retry delay timer," Western Digital Corporation #72 PDF page 90 "If the UA_INTLCK_CTRL field in the Control mode page contains 11b (see SPC-4), termination of a command with TASK SET FULL status shall cause a unit attention condition to be established for the SCSI initiator port that sent the command with an additional sense code set to PREVIOUS TASK SET FULL STATUS unless a PREVIOUS TASK SET FULL STATUS unit attention condition already exists." s/b "If the UA_INTLCK_CTRL field in the Control mode page contains 11b (see SPC-4), then completion of a command with TASK SET FULL status shall cause a unit attention condition to be established for the I_T nexus on which the command was received with an additional sense code set to PREVIOUS TASK SET FULL STATUS unless a PREVIOUS TASK SET FULL STATUS unit attention condition already exists." Western Digital Corporation #73 PDF page 95 "This standard assumes that the buffering resources available to the logical unit are limited and may be less than the amount of data that is capable of being transferred in one command." s/b "This standard assumes that the buffering resources available to a logical unit are limited, and the buffer in the logical unit may not be capable of containing all of the data required to be transferred for one command." Western Digital Corporation #74 PDF page 95 "...media..." s/b "...logical unit..." Western Digital Corporation #75 PDF page 96 "Random buffer access occurs when the device server requests data transfers to or from segments of the application client's buffer that have an arbitrary offset and byte count. Buffer access is sequential when successive transfers access a series of increasing, adjoining buffer segments. Support for random buffer access by a SCSI transport protocol standard is optional. A device server implementation designed for any SCSI transport protocol implementation should be prepared to use sequential buffer access when necessary." s/b Move this paragraph above the one that begins, "If a SCSI transport protocol supports random buffer access,". Western Digital Corporation #76 PDF page 97 "DELIVERY SUCCESSFUL:The data was delivered successfully." s/b a space after the colon. Western Digital Corporation #77 PDF page 97 "DELIVERY FAILURE:A service delivery subsystem error occurred while attempting to deliver the data." s/b a space after the colon. Western Digital Corporation #78 PDF page 98 "DELIVERY SUCCESSFUL:The data was delivered successfully." s/b a space after the colon. Western Digital Corporation #79 PDF page 98 "DELIVERY FAILURE:A service delivery subsystem error occurred while attempting to receive the data." s/b a space after the colon. Western Digital Corporation #80 PDF page 99 "The application client maintains an application client task to interact with the task from the time the Send SCSI Command SCSI transport protocol service request is invoked until it receives one of the following SCSI target device responses:" s/b "An application client maintains an application client task to interact with the task from the time the Send SCSI Command SCSI transport protocol service request is invoked until the application client receives one of the following SCSI target device responses:" Western Digital Corporation #81 PDF page 99 "Some commands (e.g., commands with immediate bits like SEND DIAGNOSTIC, or write commands when a write cache is enabled) start background operations that operate after the task containing the command is no longer in the task set." s/b "Some commands initiate background operations that are processed after the task containing the command is no longer in the task set (i.e., status has been returned for the command). For examples, see the SEND DIAGNOSTIC command when used to initiate a background self-test (see SPC-4) or write commands when write cache is enabled (see SBC-3)." Western Digital Corporation #82 PDF page 100 This sentence part is included in the preceding comment at the bottom of the previous page. Western Digital Corporation #83 PDF page 100 Delete "completed", as the sense data is not reported for a "completed" command, but as part of the command completion process for the subsequent command. Western Digital Corporation #84 PDF page 100 "Unless a command completes with a GOOD or CONDITION MET status the degree to which the required command processing has been completed is vendor specific." s/b "Unless a command completes with GOOD status or CONDITION MET status, the degree to which the required command processing has been completed is vendor specific." Western Digital Corporation #85 PDF page 104 "When a logical unit is aborting one or more tasks received on an I_T nexus using the TASK ABORTED status it should complete all of those tasks before entering additional tasks received on that I_T nexus into the task set." s/b "When a logical unit completes one or more tasks received on an I_T nexus with a status of TASK ABORTED, the logical unit should terminate all of the affected tasks before entering any other tasks received on that I_T nexus into the task set." Western Digital Corporation #86 PDF page 105 "When a command completes with a CHECK CONDITION status, the application client may request that the device server alter command processing by establishing an ACA condition, using the NACA bit in the CONTROL byte of the CDB as follows:" s/b "An application client uses the NACA bit in the CONTROL byte of the CDB (see 5.2) to specify whether or not the device server establishes an ACA condition when a command completes with CHECK CONDITION status. The meaning of the value in the NACA bit is as follows:" Western Digital Corporation #87 PDF page 105 "Which I_T nexuses are associated with the task set is influenced by the TST field in the Control mode page (see SPC-4)." s/b "Which I_T nexuses are associated with a task set is specified by the value in the TST field in the Control mode page (see SPC-4)." Western Digital Corporation #88 PDF page 106 "When a command completes with a CHECK CONDITION status, the application client may request that the device server alter command processing by establishing an ACA condition, using the NACA bit in the CONTROL byte of the CDB as follows: a) If the NACA bit is set to zero, an ACA condition shall not be established (see 5.8.1.1); or b) If the NACA bit is set to one, an ACA condition shall be established." s/b "An application client specifies if an ACA condition is established when a command completes with CHECK CONDITION status (see 5.8.1.1)." [All of the words that are recommended for replacement are in 5.8.1.1 and don't need to be repeated here.] Western Digital Corporation #89 PDF page 110 "...the command shall be terminated with CHECK CONDITION status, sense key set to ILLEGAL REQUEST, and additional sense code set to INVALID MESSAGE ERROR." s/b "...the command shall be terminated with CHECK CONDITION status with the sense key set to ILLEGAL REQUEST and the additional sense code set to INVALID MESSAGE ERROR." Western Digital Corporation #90 PDF page 111 "...(e.g., a unit attention condition with an additional sense code set to POWER ON OCCURRED may be followed by one with an additional sense code set to MICROCODE HAS BEEN CHANGED)." s/b "...(e.g., a unit attention condition with an additional sense code set to COMMANDS CLEARED BY ANOTHER INITIATOR may be followed by a unit attention condition with an additional sense code set to MODE PARAMETERS CHANGED)." [I think the example to be replaced is a poor example (i.e., if both a POWER ON OCCURRED and a MICROCODE HAS BEEN CHANGED occurred, most SCSI target devices would only report the POWER ON OCCURRED), and the suggested replacement is a much more likely scenario.] Western Digital Corporation #91 PDF page 113 "Events that occur in the SCSI device..." s/b "Events that occur in a SCSI device..." Western Digital Corporation #92 PDF page 125 "The application client maintains an application client task to interact with the task management function from the time the Send Task Management Request SCSI transport protocol service request is invoked until it receives one of the following SCSI target device responses:" s/b "An application client maintains an application client task to interact with the task management function from the time the Send Task Management Request SCSI transport protocol service request is invoked until the application client receives one of the following SCSI target device responses:" Western Digital Corporation #93 PDF page 134 "If the task has a task attribute other than SIMPLE, the task priority is not used." s/b "If a task has a task attribute other than SIMPLE, then task priority is not used." Western Digital Corporation #94 PDF page 134 "If the Task Priority argument is set to zero or is not contained within the Send SCSI Received SCSI transport protocol service indication (see 5.4.2) and a priority has been assigned to the I_T_L nexus, the device server shall use that priority as the task priority." s/b "If the Task Priority argument is set to zero or is not contained within the Send SCSI Received SCSI transport protocol service indication (see 5.4.2), and a priority has been assigned to the I_T_L nexus, then the device server shall use the priority specified for the I_T_L nexus as the task priority." Western Digital Corporation #95 PDF page 134 "If no priority has been assigned to the I_T_L nexus using the SET PRIORITY command and the logical unit does not support the INITIAL PRIORITY field in the Control Extension mode page the device server shall set the task priority to 0h (i.e., vendor specific) or the task shall have no task priority." s/b "If no priority has been assigned to the I_T_L nexus using the SET PRIORITY command, and the logical unit does not support the INITIAL PRIORITY field in the Control Extension mode page, then the device server shall set the task priority to 0h (i.e., vendor specific), or the task shall have no task priority." Western Digital Corporation #96 PDF page 140 "This standard defines the identifier attributes and name attributes for the attribute listed in A.1." s/b "This standard defines the identifier attributes and name attributes for the SCSI architecture model objects listed in A.1." ******************** End of Ballot Report ********************