brian@contex.UUCP (Brian Love) (01/22/91)
I am having trouble connecting a CDC Wren IV (344Mb SCSI) hard disk to my A2000HD (see below for full system configuration). The documentation for the Wren indicates very clearly the drive has 1549 cylinders, and (fairly clearly) 9 data heads. The docs are not clear on BlocksPerTrack; by some simple arithmetic, the drive must be able to support at least 50 512-byte blocks per track. Of course, that configuration doesn't work. When connecting a Vertex (80Mb ST-506) drive, I discovered that the prep program does not like initializing disks with more than 7 or 8 surfaces (the Vertex drive has 10). I found I could get around this by telling prep the disk has only 7 surfaces, but telling mount the real number of surfaces. To accomodate for this, I also have to lie to prep and tell it the disk has more cylinders than it really has. For the Wren, 1549cyls/50bpt/9surf would map approximately to 1991cyls/50bpt/7surf. Of course, that solution doesn't completely work either. I've been able to get 4 partitions to work, using approximately 240Mb of the disk. The 5th partition invariably fails. I don't trust all 1549 cylinders after the prep nonsense, so I create 1 partition from cyl 4 thru cyl 1549 (or higher). The format of this partition always fails on cylinder 1402. So I say, fine, I'll run my partitions up to cylinder 1400 and that'll give me 307Mb. And, of course, that doesn't work either. The fifth partition will format all cylinders up to 1400, but I get a requester saying "Not a DOS disk - Unit 6". The problem does not appear to be related to the number of partitions. It will format a single partition from cyl 4 thru 1400, but when it's done (2 and 1/2 hours later) I'm still left with "Not a DOS disk..." I do have a SCSI 1/4" streaming tape drive on the SCSI bus (everything properly terminated, I believe). I have not gotten the tape drive to work at all, yet, however, it does not appear to be a factor in the Wren drive problem. It has it's own SCSI id address and has never come to life during any of my iterations. I heard a rumor that the SCSI driver "prefers" 17 blocks per track. I suppose this can be handled in the same way as the Surfaces discrepancy. I'll try this tonight. I'm exceedingly frustrated. I'm strongly tempted to complain about the lack of robust-ness in Amiga and amiga-related products, but I'll refrain until (hopefully) some kind soul helps me and I know more about the problem. I've reproduced the mountlist file below. I've removed all unchanged keyfields from successive entries (Unit,Flags,Buffers, etc) to save space. mount.wren: /* MountList entries for CDC Wren IV (344 MB SCSI) Hard Disk */ RES5: Device = hddisk.device Unit = 6 Flags = 0 Surfaces = 7 BlocksPerTrack = 50 Reserved = 2 LowCyl = 0 ; HighCyl = 1 Buffers = 50 BufMemType = 5 /* This entry is present only as a reminder ... it is commented out. DH5: LowCyl = 2 ; HighCyl = 2 BufMemType = 0 */ WR1: Surfaces = 9 FileSystem = l:FastFileSystem LowCyl = 3 ; HighCyl = 352 GlobVec = -1 Mount = 1 DosType=0x444F5301 WR2: LowCyl = 353 ; HighCyl = 702 WR3: LowCyl = 703 ; HighCyl = 952 WR4: LowCyl = 953 ; HighCyl = 1202 WR5: LowCyl = 1203 ; HighCyl = 1400 system configuration: A2000HD rev6 motherboard w/1.3 roms, 1M of chip ram, 8M of fast ram (8-up!), 1 internal floppy drive, 1 external floppy drive, A2090A disk controller with Rodime 40M ST-506 (internal), Vertex 80M ST-506 (external), a belated Viper SCSI 1/4" streaming tape drive (internal), and a belated Wren IV 344M SCSI disk (external). For e-mail responses, please respond to an address below. The return-path my mailer uses is broken. Brian C. Love E-mail: contex!brian@uunet.uu.net OR bcl@andrew.cmu.edu Phone : (617) 224-5578 @work OR (617) 395-4476 @home
jesup@cbmvax.commodore.com (Randell Jesup) (01/23/91)
In article <1615@contex.UUCP> brian@contex.UUCP (Brian Love) writes: >I am having trouble connecting a CDC Wren IV (344Mb SCSI) hard disk to my >A2000HD (see below for full system configuration). The documentation for >the Wren indicates very clearly the drive has 1549 cylinders, and (fairly >clearly) 9 data heads. The docs are not clear on BlocksPerTrack; by some >simple arithmetic, the drive must be able to support at least 50 512-byte >blocks per track. Of course, that configuration doesn't work. Hopefully someone has attached one of these before, and will mail you or post. Don't forget that SCSI drives often reserve some number of cylinders and some number of sectors per track or cylinder. BTW, 50*512* 1549*9 is 356Meg, at least by one definition of "meg". Also, the stated size may well include spare cylinders and sectors. You need to find out the number of formatted data blocks (logical blocks) available with 512 byte sectors. This should be available from the manufacturer. More modern controllers have ways of asking the drive for that information directly. >When connecting a Vertex (80Mb ST-506) drive, I discovered that the prep >program does not like initializing disks with more than 7 or 8 surfaces >(the Vertex drive has 10). I found I could get around this by telling prep >the disk has only 7 surfaces, but telling mount the real number of >surfaces. To accomodate for this, I also have to lie to prep and tell it >the disk has more cylinders than it really has. For the Wren, >1549cyls/50bpt/9surf would map approximately to 1991cyls/50bpt/7surf. Of >course, that solution doesn't completely work either. The 2090(a) doesn't support more than 8 heads for st506 drives. This has no affect on SCSI. >I've been able to get 4 partitions to work, using approximately 240Mb of >the disk. The 5th partition invariably fails. I don't trust all 1549 >cylinders after the prep nonsense, so I create 1 partition from cyl 4 thru >cyl 1549 (or higher). The format of this partition always fails on >cylinder 1402. So I say, fine, I'll run my partitions up to cylinder 1400 >and that'll give me 307Mb. And, of course, that doesn't work either. The >fifth partition will format all cylinders up to 1400, but I get a requester >saying "Not a DOS disk - Unit 6". I seem to remember some problem with 2090's and very large SCSI drives (very large for the time it was designed, back around 86 or so). There may be some size limit at which a bug is hit. I'll ask Steve Beats about it, he knows more about it than I do. Also, at least some versions of FFS had bugs with partitions of 300+meg. I think these are fixed in 1.3.2 or maybe 1.3.3 (they certainly are in 2.0, though in 2.0 to 2.02 it had a 500Meg partition size limit, that's been fixed since then). -- Randell Jesup, Keeper of AmigaDos, Commodore Engineering. {uunet|rutgers}!cbmvax!jesup, jesup@cbmvax.commodore.com BIX: rjesup The compiler runs Like a swift-flowing river I wait in silence. (From "The Zen of Programming") ;-)
amiga-relay@ee.udel.edu (01/25/91)
From: Tyler Sarna <tsarna@polar.bowdoin.edu> In comp.sys.amiga.hardware, Randell Jesup <jesup@cbmvax.commodore.com> wrote: > In article <1615@contex.UUCP> brian@contex.UUCP (Brian Love) writes: > >I am having trouble connecting a CDC Wren IV (344Mb SCSI) hard disk to my > >A2000HD (see below for full system configuration). The documentation for Hey, this wouldn't happen to be a CDC 94171-344 drive, would it? If so, here's the setup i'm using with mine (and a GVP Series II): LowCyl: 1 HiCyl: 1545 Heads: 9 Sectors: 46 Blocks: 640299 (Btw, this drive, though it claims to be a 344 meg drive in the various places I've found info about it, formats out at 312 megs. Not too disappointing, though, since it was sold to me as a 268 meg drive :-) Now, assuming that the above information is useful to you, would you do me a favor? I got no docs with the drive (I picked it up NEW for $875 from a local Navy guy I know who builds and sells PC's on the side. He was heading over to Saudi Arabia and wanted to clear out his stock before he left). Would you be willing to photocopy the docs and mail them to me? I'd send a few bucks to cover snail postage and your time and trouble. If it's not a problem, e-mail be and I'll send you my address. Thanks, I hope this works for you! ------///------------------------------------------------------------ /// Tyler "Ty" Sarna E-Mail: tsarna@polar.bowdoin.edu \\\/// "I think cement is more interesting than people think" --\XX/---------------------------------------------------------------
brian@contex.UUCP (Brian Love) (01/29/91)
I wrote about a week ago: >I am having trouble connecting a CDC Wren IV (344Mb SCSI) hard disk to my >A2000HD (see below for full system configuration). I lied in the original post. I was able to get a single partition. But I don't want a single 300Mb partition (I know, bitch, bitch, bitch). I had settled on 5 partitions; the fifth partition going from cyl 1203 thru cyl 1400. It would format all of partition 5, but it would fail to initialize the disk. It wasn't the size; previous partitions worked. It wasn't where the partition ended; a single partition of the whole disk would work. Therefore, I rationalized, it must be where the partition started. I moved the starting cylinder of the fifth partition towards zero, and sure enough, it started working. My suspicion is the number of heads, blockspertrack, and cylinder are being combined to produce an absolute block number (for the partition's start) that is overflowing some field the driver stores in the root blocks of the disk. Anyway it works now. Randell Jesup confirmed that the A2090A driver can only do 8 heads for st506 drives. He says this has no affect on SCSI drives, but I still found it necessary to lie to prep about the number of heads (and therefore cylinders) on the disk. The last partition is currently starting at cyl 953. I'm getting about 287M out of the disk which is good enough for me. Again, here's my current working mountlist file (abridged): /* MountList entries for CDC Wren IV (344 MB SCSI) Hard Disk */ RES5: Device = hddisk.device Unit = 6 Flags = 0 Surfaces = 7 /* 8 would probably work here... */ BlocksPerTrack = 46 /* NOTE: More correct value for device */ Reserved = 2 LowCyl = 0 ; HighCyl = 1 Buffers = 50 BufMemType = 5 /* This entry is present only as a reminder ... it is commented out. DH5: LowCyl = 2 ; HighCyl = 2 BufMemType = 0 */ WR1: Surfaces = 9 FileSystem = l:FastFileSystem LowCyl = 6 ; HighCyl = 205 GlobVec = -1 Mount = 1 DosType=0x444F5301 WR2: LowCyl = 206 ; HighCyl = 355 WR3: LowCyl = 356 ; HighCyl = 505 WR4: LowCyl = 506 ; HighCyl = 952 WR5: LowCyl = 953 ; HighCyl = 1400 Thanks to everyone who replied. It is interesting the number of replies I got telling me the best solution was to throw the A2090A in the trash compactor and get a "real" controller. While I admit, I'm pretty disgusted with the quality of the A2090A, when it works it works. I should also point out that it is the only controller I know of which can control both st506 and SCSI drives, which has been a real bonus for me. Brian C. Love E-mail: contex!brian@uunet.uu.net OR bcl@andrew.cmu.edu Phone : (617) 224-5578 @work OR (617) 395-4476 @home