From owner-freebsd-scsi Tue Mar 19 05:55:31 1996 Return-Path: owner-freebsd-scsi Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id FAA29551 for freebsd-scsi-outgoing; Tue, 19 Mar 1996 05:55:31 -0800 (PST) Received: from Sisyphos (Sisyphos.MI.Uni-Koeln.DE [134.95.212.10]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id FAA29524 for ; Tue, 19 Mar 1996 05:55:22 -0800 (PST) Received: by Sisyphos id AA19879 (5.67b/IDA-1.5 for freebsd-scsi@freebsd.org); Tue, 19 Mar 1996 14:54:21 +0100 Message-Id: <199603191354.AA19879@Sisyphos> From: se@zpr.uni-koeln.de (Stefan Esser) Date: Tue, 19 Mar 1996 14:54:21 +0100 X-Mailer: Mail User's Shell (7.2.6 alpha(2) 7/9/95) To: freebsd-scsi@freebsd.org Subject: HP DAT switches ... Sender: owner-freebsd-scsi@freebsd.org X-Loop: FreeBSD.org Precedence: bulk The following information regarding HP DAT drives has been posted to "Comp.periphs.scsi" on Nov. 21st, 1994. It should still apply, AFAIK ... The signature of the original poster is: ----------------------------------------------------------------- Kevin Jones. | Hewlett Packard Ltd, | Computer Peripherals Bristol, kev%hpcpbla@hplb.hpl.hp.com | Filton Road, | Stoke Gifford, | Bristol. BS12 6QZ. | ENGLAND. ----------------------------------------------------------------- This response does not represent the official position of, or statement by, the Hewlett-Packard Company. The above data is provided for informational purposes only. It is supplied without warranty of any kind. ################################################################### Here is a summary of how HP DAT drive configuration switches work. For those not familiar with SCSI, some of the descriptions will make little sense. ==================================================================== HP35470A, HP35480A, C1533A, C1553A DDS DRIVE CONFIGURATION SWITCHES ------------------------------------------------------- Note: The data supplied applies to drives containing standard distributor product code. A drive with this code will report "09" in bytes 34 and 35 of its standard SCSI Inquiry data. Even if it does not report "09", it is possible (though not certain) that its switch configurations are very similar to the description below. There are 8 configuration switches on the underside of the drive. These are numbered 1 thru 8. SWITCHES 1+2 ----------------- These control data compression on the HP35480. (The 35470 does not have data compression circuitry. These switches should be left set to "1" in the 35470). When data compression is ENABLED the drive (HP35480) will write compressed data in DDS-DC format. When compression is disabled, the drive will write DDS format data. Note that the drive will read and de-compress DDS-DC written tapes irrespective of whether compression is ENABLED or DISABLED, ie. switching compression on/off only applies to WRITES. SWITCH1 = 1 means the drive defaults to writing DDS-DC format data. = 0 means the drive defaults to writing DDS format data. HP35470A drives cannot de-compress data from DDS-DC format tapes. The 35470A can return the compressed data via SCSI for a host to de-compress. The level of support for decompressing DAT data in host drivers is currently NIL, so for practical purposes, it is reasonable to say that you can't read a DDS-DC tape with a non compressing drive. (Note: Default compression may also be controlled via a pin on the rear of the drive. This enables some boxed drives to have a compression on/off selector switch). SWITCH2 = 1 means the host has the ability to turn compression on/off via the SCSI Mode Select command (data compression mode page). Thus the drive powers up with compression determined by switch1, but the host may subsequently use a Mode Select command to ENABLE/DISABLE compression. (Note: some UNIX systems, eg HP-UX have drivers which select compression using Mode Select depending on minor numbers in the device file.) = 0 means that the drive ignores host requests to change compression. Furthermore, a 35480 drive with this switch "0" will report that it is a 35470 drive in its Inquiry data. It is reccomended that both switches 1 and 2 be set "ON". SWITCHES 3 TO 8 --------------------- These switches configure various SCSI "features". Note that there is no 1-1 mapping of features to switches. A specific pattern of switches will enable a SET OF FEATURES. There are 2 tables below which map switch settings to feature sets. The Left hand column of these tables specifies "host", and a 3rd table shows additional features that are configured depending on the host. eg. If you want "EWEOM on read error", "Switch Buffered" and "Caution Hard Error" features only, then switch setting 111100 (for both tables) will do this. (111100 enables hosts of type "Other" which results in the "EWEOM on read error" feature being enabled). ------------------------------------------------------------------------------ | Host | Switch | Perfor | Caution | Default | Parity | No Data || Switches| | | Buffered | mance | hard | Fixed | off | phase || | | | | | error | Mode | | discon || 345678 | |-----------------------------------------------------------------------------| | HP | 1 | | 1 | | | || 111111 | | HP | 1 | | | | | || 110111 | | HP | 1 | 1 | 1 | | | || 111011 | | DEC | 1 | | 1 | | | || 111110 | | DEC | 1 | 1 | 1 | | | || 111010 | | DEC | 1 | 1 | | | | || 110010 | | Other | 1 | | 1 | | | || 111100 | | Other | 1 | | 1 | | | 1 || 011100 | | Other | 1 | | 1 | | 1 | || 101100 | | Other | 1 | | 1 | | 1 | 1 || 001100 | | Other | 1 | | 1 | 1 | | || 110100 | | Other | 1 | | 1 | 1 | | 1 || 010100 | | Other | 1 | | 1 | 1 | 1 | || 100100 | | Other | 1 | | 1 | 1 | 1 | 1 || 000100 | | Other | 1 | 1 | 1 | | | || 111000 | | Other | 1 | 1 | | | | || 110001 | | Other | 1 | 1 | 1 | | | 1 || 011000 | | Other | 1 | 1 | | | | 1 || 010001 | | Other | 1 | 1 | 1 | | 1 | || 101000 | | Other | 1 | 1 | | | 1 | || 100001 | | Other | 1 | 1 | 1 | | 1 | 1 || 001000 | | Other | 1 | 1 | 1 | 1 | | || 110000 | | Other | 1 | 1 | 1 | 1 | | 1 || 010000 | | Other | 1 | 1 | 1 | 1 | 1 | || 100000 | | Other | 1 | 1 | 1 | 1 | 1 | 1 || 000000 | | Other | | | 1 | | | || 111101 | | Other | | 1 | 1 | | | || 111001 | ------------------------------------------------------------------------------ Distributor Switch Selections -- until 35470A version 9.47 until 35480A version 9.49 (inclusive) ------------------------------------------------------------------------------------||--------- | Host | MRS | Perfor | Caution | Def | Parity | No Data | Inq | Fix | DC_ctrl || Switches| | | ON | mance | hard | Fixed | off | phase | 35480 | 512 | _by_ || | | | | | error | Mode | | discon | | | density || 345678 | |-----------------------------------------------------------------------------------||---------| | HP | | | 1 | | | | | | || 111111 | | HP | 1 | | 1 | | | | | | || 011111 | | HP | | | | | | | | | || 110111 | | HP | 1 | | | | | | | | || 010111 | | HP | | 1 | 1 | | | | | | || 111011 | | HP | 1 | 1 | 1 | | | | | | || 011011 | | HP | | | 1 | | | | 1 | | || 100011 | | DEC | | | 1 | | | | | | || 111110 | | DEC | 1 | | 1 | | | | | | || 011110 | | DEC | | 1 | 1 | | | | | | || 111010 | | DEC | 1 | 1 | 1 | | | | | | || 011010 | | DEC | | 1 | 1 | | | 1 | | | || 101010 | | DEC | 1 | 1 | 1 | | | 1 | | | || 001010 | | Other | 1 | | 1 | 1 | | | | | || 110101 | | Other | 1 | 1 | 1 | | | | | | || 111001 | | Other | 1 | 1 | 1 | | | 1 | | | || 011001 | | Other | 1 | 1 | 1 | | 1 | | | | || 101001 | | Other | 1 | 1 | 1 | | 1 | 1 | | | || 001001 | | Other | 1 | 1 | 1 | 1 | | 1 | | | || 010001 | | Other | 1 | 1 | 1 | 1 | 1 | 1 | | | || 000001 | | Other | | | 1 | | | | | | || 111100 | | Other | | 1 | 1 | | | | | | 1 || 001100 | | Other | | | 1 | 1 | | | | | || 110100 | | Other | | 1 | 1 | | | | | | || 111000 | | Other | | 1 | 1 | | | 1 | | | || 011000 | | Other | | 1 | 1 | | 1 | | | | || 101000 | | Other | | 1 | 1 | | 1 | 1 | | | || 001000 | | Other | | 1 | 1 | | | | | 1 | || 110000 | | Other | | 1 | 1 | 1 | | 1 | | | || 010000 | | Other | | 1 | 1 | 1 | 1 | 1 | | | || 000000 | ------------------------------------------------------------------------------------||---------| Distributor Switch Selections -- after 35470A version 9.47 after 35480A version 9.49 (inclusive) The reason for the 2 tables is that the mappings changed for newer firmware revs. This was not a good idea (IMHO), but life's like that. You can find out "partly" what rev of firmware you have by looking at bytes 32 and 33 of the standard inquiry data. This will tell you the major number of the revision. If its rev8 or less then use the "old" table, if its rev10 or higher, use the newer table. If its rev9 then you would have to recover the 0xC0 Inquiry vital product data page which contains (in ASCII) full details of the firmware rev. If you can't figure out what rev of firmware you have then you'll just have to "experiment" with settings from both tables. Rev's 9.47/9.49 shipped until approx Feb 1993. The following table describes "extra" features enabled for hosts of type "HP" "DEC" and "Other".... ------------------------------------------------------------------------------ | Host | Async | EWEOM | Write | No | Attn | Full | Signed | Trunc | | | Sense | on read | Zero | EWEOM | After | Reset | Residue | Inquiry | | | error | FM | residue | Load | | 76543 | by 3 | |----------------------------------------------------------------------------| | HP | 1 | 1 | 1 | 1 | 1 | | | | | DEC | | | | | | 1 | 1 | 1 | | Other| | 1 | | | | | | | ------------------------------------------------------------------------------ Asyncronous sense ---------------------- TRUE: The mark bit may be set on asynchronous request sense. FALSE: The mark bit will never be set on asynchronous request sense. Comments: This feature is used on some HP-UX systems whose drivers "forget" if they're at a filemark and send a SCSI "request sense" command (in the absence of a pre-ceding check condition) to find out where they are. The effect of this feature being enabled for systems which do not want asynchronous sense should be benign. Attn after load -------------------- TRUE: "Unit Attention" and "Not ready to transition" posted for command subsequent to any load, including those that just rewind or do nothing. Comments: Another HP driver'ism. SOme HP-UX (or it might have been MPE) driver's expect to see SCSI Unit Attention sense after a tape is loaded despite the fact that it was the host that issued the load command. Caution for Hard Error ------------------------ TRUE: If a hard read or write error occurs, resulting in additional sense being set to 1100h (unrecovered read error) or 0C00h (Write error) then display caution on front panel. This indicates that the drive should be cleaned. FALSE: No special LED display for hard read or write errors. DC ctrl by density (Nov 17 1994 and later) --------------------------------------------- TRUE: Enables DC on writes to be enabled by mode select density code 3 and disabled by mode select density code 0. FALSE: Mode select density code has no effect on DC. Mode sense density code indicates drive/tape density. Comments: SUN drivers, Exabyte drives and Exabyte drivers control data compression via the density code. eg. based on device file name or device file minor numbers the driver sends a density code of 0 or 3 on device open. If you want device-file based control of DC on your SUN/whatever system then this is the option for you! EW-EOM on read error ----------------------------- TRUE: EOM bit set for read errors on hitting mark. FALSE: EOM bit only set if at Physical end of partition or media. fixed mode ------------------- TRUE: Default block size is set to 1024 bytes. FALSE: Default block size is set to 0 bytes. Comments: Sme applications (esp. PC based) which are used to talking to QIC drives expect a SCSI fixed block length of 1K. SCSI states that they should configure this with a mode select command, but some packages just assume the drive is pre-configured for 1K fixed size blocks. fix 512 (November 1994 and later) -------------------------------------------- TRUE: Default block size is set to 512 bytes. FALSE: Default block size depends on "fixed mode" Comments: Sme applications (esp. PC based) which are used to talking to QIC drives expect a SCSI fixed block length of 512 bytes. SCSI states that they should configure this with a mode select command, but some packages just assume the drive is pre-configured for 512 byte fixed size blocks. Full reset --------------- TRUE: ``DEC'' defined reset. Tape is rewound. FALSE: Soft reset, drive takes very little action for SCSI reset, logical position unchanged but current command aborted. Comments: You have to be brain dead to want a "full reset". This means you loose logical tape position after a bus reset. immediate ----------------- TRUE: The immediate bit is overridden and assumed to be ``1'' on every command that supports it. ``switch buffered''= FALSE mode is also enforced (see definition below). FALSE: The Immed flags are used. ``switch buffered'' mode is left as-is. Comments: SCSI commands affected include: Load, Rewind, Erase, Write Filemark. Assuming immediate mode for filemarks can drastically improove the performance of systems which fail to set this bit when sending write filemark commands. The downside is that you don't flush data to tape in response to a filemark command, and for some applications this compromises data-integrity checkpointing. infinite flush ------------------ TRUE: Infinite buffer flush timeout (ie write delay time=0). For mode select, write delay time is silently ignored. FALSE: Default write delay time is used. Mode select may be used to re-specify timeout. Comments: Infinite flush means data is left lying around in the buffer for long periods of time (and is subject to LOSS if power fails). The plus side of this is that VERY SLOW applications don't continually stop/start the mechanism (it will flush every 5 seconds by default). ANother plus point of this feature is that you won't loose capacity if the drive flushes partially written groups. Inquiry 35480 ---------------- TRUE: The product ID in the Inquiry data will be HP35480 (even if the drive is a bit more up to date (eg. C1533A or C1553A). This enables some drivers with hard coded product ID's to work with newer drivers (HP-UX S800 9.0 for C1553A, BackupExec on Windows NT for C1533A and C1553A). MRS on ------------ TRUE: MRS (Media recognition system) is implemented. Tapes have to be of DDS grade, otherwise they are treated as write protected and write commands will be rejected with sense key=7, additional sense =3000h. FALSE: MRS is not implemented. You can write any tape you like. no data disconnect ----------------------- TRUE: No disconnect during data phase. NB: save data pointers may still be sent. FALSE: Disconnects may occur during data phase. Comments: Useful for hosts which cannot disconnect on ODD BYTE boundaries. Some SGI hosts can't disconnect on ODD WORD (32 bit) boundaries so this feature should be enabled for SGI connect. Downside = DOn't do SCSI operations exceeding 64Kbytes otherwise the bus could hang-up waiting for device-side I/O. The drive can normally guarantee to have at least 64K of data/space before it starts data phase. no EW-EOM residue ----------------------- TRUE: if check condition reported for EW-EOM then sense data will not have valid flag set. FALSE: check condition reported for EW-EOM then sense data will have valid flag set. no parity --------------- TRUE: The drive is configured not to check for parity on the SCSI bus. FALSE: The drive is configured to check for parity on the SCSI bus. COmments: The drive always generates SCSI parity. Not checking SCSI parity, especially when operating a synchronous bus, is a bad idea. performance ----------------- TRUE: ``immediate'' and ``inifinite flush'' are selected (TRUE) (see definitions) FALSE: ``immediate'' and ``inifinite flush'' are not selected (FALSE) (see definitions) Comments: This can dramatically improove performance when applications "do silly things", like try and disable buffering or write lots of non-immediate filemarks. If they do silly things and this feature is not enabled then the drive will spend most of its life stream-failing. signed residue ----------------- TRUE: Residue is negative on back space. FALSE: Residue is absolute, positive, for all spacing. Comments: This is a DEC'ism. switch buffered ------------------ TRUE: Default buffered mode is used. Buffered mode may be switched using mode select. FALSE: Buffered mode = 1 is enforced. Mode select buffered mode field is silently ignored. This mode is enforced if ``immediate'' is TRUE. This is ``TRUE'' by default. trunc inquiry by 3 ----------------------- TRUE: Standard inquiry page is truncated by 3 bytes to 40h bytes. FALSE: Inquiry page remains 43h bytes long. Comments: This is a DEC'ism. I think DECstation 4000/5000's will hang during BOOT if they see "too much" inquiry command data. write zero fm --------------- TRUE: Allow write zero filemarks on write protected tape, and also if no tape is loaded. FALSE: Write zero filemarks is treated as any other write command with respect to write protected tapes and no tape present. Comments: AN HP'ism. Writing zero filemarks is the SCSI reccomended method of flushing the data buffer to tape. Some HP drivers do this before closing, even if the tape is write protected!