[comp.unix.i386] ISC won't install on 330meg ESDI

brad@microm.UUCP (Bradley W. Fisher) (08/03/90)

Ok, I give up. I really do *want* to believe that Interacive's UNIX is a
viable solution for an operating system ... but here I am getting bit again :-(

The facts:
	16 mhz Intel 301 386 motherboard
	8 meg of ram
	Phoenix BIOS ver 3.03
	Wangtek 5099EK
	currently running 2 85meg MFM hard disks OK.
	ISC 2.0.2

The problem:
	Cannot get ISC to load onto new Maxtor 4380E hard disk with
	Adaptec 2322B controller.

The procedures and symptoms:
	Boot MSDOS, run debug g=c800:5 to use controller's BIOS for low
	level format routine. Ran 48 hour surface scan with Speed Stor,
	no problem, no new bad sectors discovered.
	Boot ISC, sees drive as 1224 x 15 x 36, ISC says you can't use 
	more than 1024 cylinders, so I tell it there are only 1024. Run
	fdisk, use 100% for UNIX, start cylinder 1, end cylinder 1021.
	Activate UNIX partition. Set up file systems, 60 cyl for swap
	(16 meg), 75 cyl for root (20 meg), 443 for usr (122 meg), 442
	for usr2 (122 meg). ISC creates root filesystem OK, labels it,
	and makes lost+found. Creates usr filesystem OK labels it, BUT
	the following error is displayed ...
	
NOTICE: free block read error on Primary AT Hard Disk DISK unit 0, partition 3

mount failed
mount of usr failed

	... and lost+found is not created. Then ISC merrily goes on trying to
	install files on the hard disk (great error checking in that script!).
	Never does try to create usr2.

Other goodies:
	The same system installs SCO XENIX fine. Installing with a 170 meg
	Micropolis 1355 (1024 x 8 x 35) works just fine too. All filesystems
	are created properly.
	I've tried everything I know of to make this work (like disabling the
	BIOS on the Adaptec,knocking down the number of cylinders a little more)	and keep getting the same results.

I'm not trying to do ISC bashing here, but no we don't want to spend several
hundred dollars to *maybe* get an answer from ISC to *solve* this problem. If
no one on the net has a clue as to what is happening here, we'll have to solve
this the way we've solved other problems ISC couldn't give us a good answer on 
in the past ... switch over to SCO XENIX. A least our money will be going 
towards a product I know will install and run on this configuartion the *first*
time. I would like to see this work though ... really ;-)

--
I'm just a wanna be UNIX guru (IJWBUG)               | Micro Maintenance, Inc.
						     | 2465 W. 12th St. #6
	   -== Brad Fisher ==- 		             | Tempe, Arizona  85281
     ...!asuvax!mcdphx!hrc!microm!brad		     | 602/894-5526

wtm@uhura.neoucom.EDU (Bill Mayhew) (08/03/90)

ISC seems to be a fairly close relative of AT&T 3.2.  I found that
AT&T Unix seems to have a strong prejudice against any non-Western
Digital brand HD controllers.  Unix does not use the BIOS code on
the controller due to the fact that controller BIOSes don't support
reentrancy (as a rule of thumb).  For AT&T (and perhaps ISC if they
inhereted the driver code from AT&T), this means that absolute 100%
register level compatibility is required on the HD interface card.

My system at home is a JDR 386-33 MHz cache board.  I am using a
Priam 330 megabyte drive with 1220 cylinders and a WD-1007V-SE2
floppy/hard ESDI controller.  I strapped the board to provide
sector translation to 63 sectors per track.  The sector translation
is transparent to the O/S, so Unix gets to see the entire drive
minus the 30 meg DOS partition that I set aside.  The only problem
I had is that Unix was not happy unless I allocated the DOS
partition starting at cylinder 0.  I'd rather have the DOS
partition in the high numbered cylinders in case a DOS program bombs
out and tries to write past the end of the partition.  Oh well...

The WD-1007V-SE2 may not be the fastest controller in existance,
but it isn't a slouch by any means either.

==Bill==
-- 
Bill Mayhew  Northeastern Ohio Universities College of Medicine
Rootstown, OH  44272-9995  USA    phone: 216-325-2511
wtm@uhura.neoucom.edu   ....!uunet!aablue!neoucom!wtm
via internet: (140.220.001.001)

marz@cbnewsm.att.com (martin.zam) (08/03/90)

Article: 9512 of comp.unix.i386
From: brad@microm.UUCP (Bradley W. Fisher)
Newsgroups: comp.unix.i386
Subject: ISC won't install on 330meg ESDI
Keywords: ISC ESDI woes
> 
> 
> The facts:
> 	16 mhz Intel 301 386 motherboard
> 	8 meg of ram
> 	Phoenix BIOS ver 3.03
> 	Wangtek 5099EK
> 	currently running 2 85meg MFM hard disks OK.
> 	ISC 2.0.2
> 
> The problem:
> 	Cannot get ISC to load onto new Maxtor 4380E hard disk with
> 	Adaptec 2322B controller.
> 
> The procedures and symptoms:
> 	Boot MSDOS, run debug g=c800:5 to use controller's BIOS for low
> 	level format routine. Ran 48 hour surface scan with Speed Stor,
> 	no problem, no new bad sectors discovered.

Brad,
  This whole problem exists because ISC has a bug in their low-level format
  program.  It cannot successfully format ESDI drives!  This is, of course,
  an undocumented "feature" of ISC.  There are other "features" in ISC.  I
  have solved some other problems, but the answer you need follows:

  You need to Boot MSDOS, run debug g=c800:5 and using the controller's BIOS
  routines "Turn on 63 Sector translation".  This is one of the options in
  a submenu.  This fakes the system into seeing more (logical) sectors per
  track so that the number of cylinders is (logically) less than 1024.

  This step is usually done right after the low-level format, so if you've
  done anything that affects the low-level format, it may have to be re-run.

  This is how i got my M9380 drive to work with ISC.  Also, if you get
  messages about "Too many bad blocks" from ISC, I know how to fix this
  so it will still install.  Post or Email if you see this one.  It's
  a LU-LU.  Hope this helps.

> 	Boot ISC, sees drive as 1224 x 15 x 36, ISC says you can't use 
> 	more than 1024 cylinders, so I tell it there are only 1024. Run
> 	fdisk, use 100% for UNIX, start cylinder 1, end cylinder 1021.
> 	Activate UNIX partition. Set up file systems, 60 cyl for swap
> 	(16 meg), 75 cyl for root (20 meg), 443 for usr (122 meg), 442
> 	for usr2 (122 meg). ISC creates root filesystem OK, labels it,
> 	and makes lost+found. Creates usr filesystem OK labels it, BUT
> 	the following error is displayed ...
> 	
> NOTICE: free block read error on Primary AT Hard Disk DISK unit 0, partition 3
> 
> mount failed
> mount of usr failed
> 
> 	... and lost+found is not created. Then ISC merrily goes on trying to
> 	install files on the hard disk (great error checking in that script!).
> 	Never does try to create usr2.
> 
> Other goodies:
> 	The same system installs SCO XENIX fine. Installing with a 170 meg
> 	Micropolis 1355 (1024 x 8 x 35) works just fine too. All filesystems
> 	are created properly.
> 	I've tried everything I know of to make this work (like disabling the
> 	BIOS on the Adaptec,knocking down the number of cylinders a little more)	and keep getting the same results.
> 
> I'm not trying to do ISC bashing here, but no we don't want to spend several
> hundred dollars to *maybe* get an answer from ISC to *solve* this problem. If
> no one on the net has a clue as to what is happening here, we'll have to solve
> this the way we've solved other problems ISC couldn't give us a good answer on 
  ISC didn't have a clue.  Adaptec didn't have a clue(I have the same
  controller).  I will also warn you to turn off the read-ahead cache on this
  controller card.  I have heard that this can cause UNIX systems to get flakey.

> in the past ... switch over to SCO XENIX. A least our money will be going 
> towards a product I know will install and run on this configuartion the *first*
> time. I would like to see this work though ... really ;-)
> 
> --
> I'm just a wanna be UNIX guru (IJWBUG)               | Micro Maintenance, Inc.
> 						     | 2465 W. 12th St. #6
> 	   -== Brad Fisher ==- 		             | Tempe, Arizona  85281
>      ...!asuvax!mcdphx!hrc!microm!brad		     | 602/894-5526
> 
> 

					Martin Zam
					(201)564-2554