home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!contex!wildcat.contex.com!jeff
- From: jeff@wildcat.contex.com (Jeff Carey)
- Newsgroups: comp.sys.sgi
- Subject: Re: Using (setting up) a 9 track tape drive (?)
- Message-ID: <3264@contex.contex.com>
- Date: 8 Jan 93 23:20:03 GMT
- References: <1993Jan7.185622.16632@Princeton.EDU>
- Sender: news@contex.contex.com
- Organization: Xyvision Design Systems
- Lines: 88
-
- In article <1993Jan7.185622.16632@Princeton.EDU> brett@surfpix.princeton.edu (Brett Borowski) writes:
- >
- >Howdy!
- >
- > I have a 9-track tape drive that I'd like to get working under 4.0.5.
- >It used to work on the machine that was running 3.3, but that has since
- >been upgraded. Everything works except for the variable block size stuff.
- >There is no device in /dev/mt to accomodate it. I saw Dave Olsen's
- >earlier post, but I need some more specifics. Here are the details
- >about the drive (from mt stat):
- >
- >Controller: SCSI
- >Device: HP: 88780 A657
- >Status: 0x20266
- >Drive type: unknown
- >Media: READY, write protected, at BOT (I have a tape in right now...)
- >
- >Any info would be greatly appreciated!
- >
- >Brett
- >--
- >brett@surfpix.princeton.edu (Finger for more info...)
-
- You need to edit the file /usr/sysgen/master.d/tpsc, adding
- a definition for your tape drive.
- For our Cipher 990 tape drives with NCR scsi piggyback (with prom
- revision 008), we copied the definition for the Kennedy and made small
- changes. Read comments in tpsc file for explanation.
- "man mtio" and/or "man tps" on-line manual pages may be helpful.
- Also, "/usr/include/sys/mtio.h" and/or "/usr/include/sys/tpsc.h" may
- also help. Now that I think about it, "/usr/include/sys/tpsc.h" will
- definitely help.
-
- This is our addition for tpsc file "Revision: 3.19" and "Revision: 3.21",
- which are applicable with IRIX 4.0.1 thru 4.0.5F:
-
- ,
- { KENNEDY96X2, TP9TRACK, 7, 0, "NCR H62", "", 0, 0,
- {3, 6, 2, 1}, MTCAN_BSF|MTCAN_BSR|MTCAN_APPEND|MTCAN_LEOD |
- MTCAN_CHKRDY|MTCAN_VAR| MTCAN_SETSZ|
- MTCAN_SYNC|MTCANT_RET,
- 20, 150, 20*60, 5*60, 512, 60*512 }
-
-
- After editting the file, you must run "/etc/autonconfig" (see
- "man autoconfig"), and reboot.
-
- The "mt -t /dev/mt/tps0d7nrnsv sta" for this drive returns:
-
- Controller: SCSI
- Device: NCR H621: 0-STD1-03-46M990
- Status: 0x20262
- Drive type: 9 track
- Media : READY, writable, at BOT
-
- See "device:", above--it corresponds to the fifth field of the definition.
- The fact that your mt stat returns (among other things):
- Drive type: unknown
- means that the "tp_vendor" field (atleast) of your definition is not
- correct. Where I have `"NCR H62"' you should probably have `"HP: 887"'
- with "tp_vlen" also set to 7. (For these fields, see the definition of
- struct tpsc_types in /usr/include/sys/tpsc.h.) Note the definitons of
- MAX_INQ_VID and MAX_INQ_PID which are 8 and 16, respectively. We have
- not used the "tp_product" field, but have had a problem with the "tp_vendor"
- field when, at one time, we had it 8 characters long...the MAX_INQ_VID
- value must account for the NULL at the end of the string. (I wish this
- had been explained more clearly in the file, or somewhere.) Anyhow,
- (MAX_INQ_?ID - 1) is the longest these fields can be, I believe.
-
- I suspect that you have these correct, but the minor numbers associated
- with the device names are (for scsi device 7, atleast):
-
- 0 crw-rw-rw- 2 root sys 23,224 Jan 5 15:30 /dev/mt/tps0d7
- 0 crw-rw-rw- 2 root sys 23,225 Jan 5 15:30 /dev/mt/tps0d7nr
- 0 crw-rw-rw- 2 root sys 23,227 Jan 5 15:30 /dev/mt/tps0d7nrns
- 0 crw-rw-rw- 2 root sys 23,231 Jan 8 15:34 /dev/mt/tps0d7nrnsv
- 0 crw-rw-rw- 2 root sys 23,229 Jan 8 15:34 /dev/mt/tps0d7nrv
- 0 crw-rw-rw- 2 root sys 23,226 Jan 5 15:30 /dev/mt/tps0d7ns
- 0 crw-rw-rw- 2 root sys 23,230 Jan 8 15:34 /dev/mt/tps0d7nsv
- 0 crw-rw-rw- 2 root sys 23,228 Jan 8 15:34 /dev/mt/tps0d7v
-
- Good luck.
-
- -jeff
-
- --
- Jeff Carey
- jeff@contex.com (or: contex!jeff@uunet.uu.net)
-