The Fatted Calf/The Fatted Calf.iso/Applications/Utilities/Formatter/ReadMe

Internet Message Format | 1992-12-26 | 8KB

Date: Sat, 16 Feb 91 13:06:30 PST
Subject: Version 1.2



Feb 16, 1991

    Here is the promised version 1.2 of my disk formatting utility. Please
destroy any earlier versions you see floating around, there was a MAJOR bug that
affects people formatting drives other than the Fujitsu M226XS family.

Bug fixes:
    1) the automated disktab entry algorithm was improved yet again. My entries
now exactly match those generated by NeXT for all of the NeXT formatted drives
I could gain access to (about 3 different models). This last improvement was
VERY minor, it will increase the size of your disk by 1 block. I really do not
think it is worth the effort to rebuild your disk if you used an earlier
version.
    2) Only the logical sector size and not the physical sector size was being
set to 1024 bytes if your drive was NOT a Fujitsu. This bug appeared in every
version since 1.0, where it was introduced. My regression testing did not pick
this bug up since I had already set the physical block size of my test disks
to 1024 bytes. The utility would happily set the logical size to 1024 bytes and
leave the physical sector size at whatever it was previously! I was too lazy
to reformat my drives back to 512 byte sectors every time I regression tested.
Sorry if this caused you undo stress! Once again, this bug DID NOT affect
Fujitsu drives!

New features:
    1) Better (hopefully) help information again. Please read this! The most
major change has to do with the synchronous transfer jumper settings of the
Fujitsu M2263XS family on 030 cubes. It seems that the boot roms in the old
cubes can not negotiate or reject synchronous transfer bids during the boot
sequence. In my system (which is an 030 until my much awaited upgrade comes in)
I did not see this problem for the following reason: I still had a valid kernel
on the disk with the next higher SCSI ID. The NeXT would try to boot off of my
Fujitsu, give up because of the synchronous transfer bidding, go on to the next
SCSI device, load the kernel, (which appears to be able to handle the bidding),
and SWITCH BACK to the Fujitsu! So I now have answers to all of the outstanding
"It-works-fine-for-me-but-you-have-a-problem" issues.

Outstanding issues:
    1) The history info for version 1.2 is incorrect, however, I did not want
to turn another revision just for that.

                    Rory.

P.S.  extraction instructions: save the stuff between the two lines labeled
"CUT HERE" to a file called foob. Next type "uudecode foob", which will create
the file "formatter_1.2.tar.Z". Next type "uncompress formatter_1.2.tar.Z".
Next type "tar xvf formatter_1.2.tar". This should produce an executable file.
Any of the intermmediate files created (foob, formatter_1.2.tar) may now be
deleted.



Date: Tue, 19 Feb 91 13:12:50 PST
Return-Path: <[e-mail]>
Subject: README file
Status: R

Feb. 19,1991
Hello!
    This is a form letter response to general questions about Formatter. There
are currently about 60 people across the country who are in contact with me and
who are using this program. The current version is 1.2. Please make an effort
to remove previous versions you may find. If formatter does not appear at your
local FTP sites, please feel free to upload it AND THIS README FILE. Please
make sure to tar and compress the files before uploading. It would also be a
good idea to use the following naming convention:

    formatter_1.2.tar.Z     ( the actual program )
    formatter.readme.tar.Z    ( this file )

Advantages to reformatting:
1) expect about 5% greater capacity from your drive. This may not be worth the
effort for small drives as illustrated below:
    40MB -> 41MB
    80MB -> 82MB
    100MB -> 105MB
    300MB -> 315MB
    600MB -> 630MB
    1200MB -> 1260MB
    These are just examples of about what to expect. The actual capacity of your
drive will be affected by things such as the amount of space reserved for flaw
mapping. Also, marketing hype usually calls a Mbyte 1000000 bytes instead of
1048576 bytes, which confuses many people.
    Speed improvements have been reported by some, however they are VERY SLIGHT.
If you own a Fujitsu M226XS drive, the caching parameters will be modified to
give approximately a 15% increase in the maximum transfer rate. Transfer rates
are dependent on the length of your SCSI bus and the number of SCSI devices
attached to your bus. In my system I observed the transfer rate of my M2263S
to go from 700Kbytes/second to 800Kbytes/second.
    There is also some confusion as to disk transfer rates. Most, if not all,
disk manufacturers quote maximum BURST transfer rates, or the rate at which
data can be transfered to/from the drive's ram buffer. These figures are often
in the 4 to 5 Megabyte/second range. The numbers I refer to are the maximum
SUSTAINABLE transfer rates, which include command processing times in your
computer, the disk drive, and the actual reading/writing of the data from/to
the disks physical media.

Disadvantages to reformatting:
    This destroys all of the data on the disk being formatted. BE CAREFUL! Also,
you will need to be familiar with fstabs, disktabs, builddisk, etc. Many people
do not want to hassle with system administration such as this. All of the info
needed is in NeXT's manuals, however, this is not something a novice would want
to do.
    As a side note, it should be noted that hooking up SCSI devices is not
simply "plug and play". There is a risk of damaging all of the disks on your
computer and the computer itself if you do not know what you are doing. SCSI
bus termination is yet another thing that confuses many. If you are not sure,
ASK SOMEONE WHO IS! Also, do not take everthing you read on UUNET NEWS as
the truth!


Please read all of the manuals (online and printed) regarding the following:

builddisk
fstab
disktab
mount

Also, PLEASE BE SURE TO READ THE HELP INFORMATION PROVIDED BY THE PROGRAM

Notes on Quantum drives:
    I have talked to Quantum technical support and this is what they said:
    The newer Quantums are factory formatted to a physical sector size of
512 bytes/sector. Although you may change the logical sector size to 1024 bytes
per sector, you can not change the physical sector size, thus there are no
capacity gains by setting 1024 bytes/sector. The newer Quantums will report
an error if the mode select command attempts to alter the physical sector size.
    I have tried to program my utility defensively in situations such as
this. When this happens, my utility will not issue the format command and your
drive should remain unchanged with all data intact.

Theory of operation:
    1) At program startup, my program opens the generic SCSI device, "/dev/sg".

    2) When you select as SCSI ID, the program issues an ioctl to the generic
    SCSI device selecting that SCSI ID. The logical unit number is set to zero.
    The program then issues both INQUIRY and READ CAPACITY commands to the
    device. The program then parses the data returned by the device and displays
    the device information.

    3) When you select the FORMAT button, the program issues MODE SENSE commands
    to get the format parameters for your drive. The program then attempts to
    set both the logical and physical sector size to 1024 bytes/sector by
    issueing a MODE SELECT command. All of the other format parameters are
    left unchanged. If the MODE SELECT command did not encounter errors, the
    program will issue a FORMAT command. IF THE MODE SELECT COMMAND REPORTS
    ERRORS, THIS UTILITY CAN NOT FORMAT YOUR DRIVE.

    4) When you select the DISKTAB button, the program issues MODE SENSE,
    INQUIRY, and READ CAPACITY commands to gather the information necessary
    to build a disktab entry.


                        Rory Bolt
                        [e-mail]