369 lines
14 KiB
Plaintext
369 lines
14 KiB
Plaintext
FIPS Frequently Asked Questions
|
|
-------------------------------
|
|
|
|
Here are some questions that people asked me by email. This file may
|
|
help you with some common problems. Please read it carefully before sending
|
|
mail. If you want to know what FIPS is all about, read the file README.1ST.
|
|
|
|
There are answers to the following questions:
|
|
|
|
|
|
Q01. Since you can split partitions with FIPS, it should also be possible
|
|
to merge partitions, right?
|
|
|
|
Q02. Can I reverse the changes that FIPS made to my harddisk?
|
|
|
|
Q03. What if I did not keep the root and boot sector? Can I still undo the
|
|
partition split?
|
|
|
|
Q04. I heard that the cluster size depends on the partition size and that
|
|
space usage is better with a small cluster size. Can I use FIPS to
|
|
decrease the cluster size of my hard disk?
|
|
|
|
Q05. I want to split a large partition into three or four smaller ones.
|
|
Can I use FIPS multiple times?
|
|
|
|
Q06. FIPS creates a second primary DOS partition by default. Is this allowed?
|
|
|
|
Q07. What does the message "Info: Partition table inconsistency" mean?
|
|
|
|
Q08. FIPS displays an error message and refuses to work. What should I do?
|
|
|
|
Q09. FIPS does not recognize my SCSI disk.
|
|
|
|
Q10. FIPS finds a partition with partition type 56h.
|
|
|
|
Q11. FIPS seems to work fine, and DOS sees the new partition, but Linux
|
|
fdisk sees only one partition.
|
|
|
|
Q12. Does FIPS work with Windows 95?
|
|
|
|
|
|
------------------------------------------------------------------------------
|
|
|
|
Q01. Since you can split partitions with FIPS, it should also be possible
|
|
to merge partitions, right?
|
|
|
|
A01.a
|
|
-----
|
|
NOT IN GENERAL. I don't want to go into technical details, just so much:
|
|
Every DOS formatted partition has a file allocation table (FAT) that holds
|
|
entries for every cluster of the partition. Usually one FAT is not large
|
|
enough to hold the information for both partitions, so that partitions can
|
|
not be easily joined. It is feasible, but I'm probably not going to incor-
|
|
porate this into FIPS.
|
|
|
|
So far I have heard of two software packages that claim to expand partitions.
|
|
One is a free program called "Partition Resizer" (look for a file called
|
|
presz???.zip on you favorite BBS or FTP server), the other is a commercial
|
|
product called PartitionMagic by Powerquest. I did not try these yet.
|
|
|
|
|
|
A01.b
|
|
-----
|
|
YES, if you split the partitions with FIPS before. As you may expect,
|
|
the problem with the FAT does not exist in that case, since it was origi-
|
|
nally formatted large enough. In fact the only thing required is resto-
|
|
ring the original root and boot sector (it does not matter how the data
|
|
on the partition changed in the meantime). This can be done with the
|
|
program 'restorrb' which comes with FIPS. Please consider the following:
|
|
|
|
- You must have the original image of the root and boot sector. For this
|
|
you _must_ save them to a floppy disk when FIPS offers that.
|
|
|
|
- You must not format the first partition in the meantime (because then
|
|
a new FAT is generated which is probably too small).
|
|
|
|
- The data on the second partition (the one that was generated by FIPS)
|
|
is lost, of course. The data on the first partition is preserved.
|
|
|
|
|
|
---------
|
|
|
|
Q02. Can I reverse the changes that FIPS made to my harddisk?
|
|
|
|
A02.
|
|
----
|
|
Yes, see answer A01.b. Make sure that you keep the original root and boot
|
|
sectors that you can save to floppy disk with FIPS.
|
|
|
|
|
|
---------
|
|
|
|
Q03. What if I did not keep the root and boot sector? Can I still undo the
|
|
partition split?
|
|
|
|
A03.
|
|
----
|
|
Difficult. For the moment, your only option is to read the file TECHINFO.TXT
|
|
to learn about the hard disk structures, recalculate the old partition
|
|
and format info by hand and use a disk editor the change them. You must
|
|
roughly proceed as follows (I will call the partition that was split off
|
|
of the original partition the 'second' partition, the remaining part of
|
|
the original partition the 'first' one):
|
|
|
|
|
|
1. Make sure there is no important data left on the second partition,
|
|
since it will be deleted. Make copies of all root and boot sectors
|
|
(on floppy disk!) in case you make a mistake. Also make sure that
|
|
you have a bootable DOS floppy with the disk editor on it. Backup
|
|
your data!
|
|
|
|
2. Choose 'edit physical drive' and 'edit partition table' in the disk
|
|
editor menu (assuming that your disk editor supports this).
|
|
|
|
3. Examine the table to make sure which two partitions you want to merge.
|
|
|
|
4. Take end head, cylinder, sector from the second partition and enter the
|
|
values in the corresponding fields of the first partition.
|
|
|
|
5. Add the number of sectors of the first partition to the number of sectors
|
|
of the second and enter the new value in the number of sectors field
|
|
of the first partition.
|
|
|
|
6. Delete the entry for the second partition completely (overwrite with
|
|
zeroes).
|
|
|
|
7. Look for the boot sector of the first partition and enter the new no.
|
|
of sectors in the no. of sectors field in the boot sector.
|
|
|
|
|
|
Please make sure you know what you are doing. I can't take any responsibi-
|
|
lity if you mess up your hard disk. Read the TECHINFO.TXT file carefully.
|
|
|
|
|
|
---------
|
|
|
|
Q04. I heard that the cluster size depends on the partition size and that
|
|
space usage is better with a small cluster size. Can I use FIPS to
|
|
decrease the cluster size of my hard disk?
|
|
|
|
A04.
|
|
----
|
|
In the partition that you split off of the original partition, the cluster
|
|
size is automatically adapted to the new size when you format it. In the
|
|
original partition, you can not change the cluster size without reformat-
|
|
ting and thereby deleting all data.
|
|
|
|
If your new partition is big enough, you might copy the data to the new
|
|
partition and format the old one, but be aware of two things:
|
|
|
|
1. You will not be able to reverse the partition split afterwards (see A1.a).
|
|
|
|
2. Newer format programs check to see if the partition has already been
|
|
formatted and in this case will preserve the format (to make data re-
|
|
covery easier in case the formatting was a mistake). You might have
|
|
to trick the format program into thinking that the partition is new.
|
|
Possibly the /u switch will do this (I have no possibility to check
|
|
this, please let me know if it works), but if not, you might have to use
|
|
a disk editor and overwrite the boot sector of the partition with zeroes
|
|
to invalidate it.
|
|
|
|
Note: I received a report that even overwriting the boot sector was not
|
|
enough. I can only imagine that there was still information in the
|
|
BIOS tables about the old format - I suggest to reboot after overwriting
|
|
the boot sector to clear all tables. Sometimes the DOS tools are just
|
|
too intelligent :-(
|
|
|
|
If you experience problems here, drop me a line.
|
|
|
|
|
|
The usual cluster size of a partition is roughly as follows:
|
|
|
|
0 - 32MB 512 Bytes
|
|
32 - 64MB 1024 Bytes
|
|
64 - 128MB 2048 Bytes
|
|
128 - 256MB 4096 Bytes
|
|
256 - 512MB 8192 Bytes
|
|
512 -1024MB 16384 Bytes
|
|
|
|
|
|
---------
|
|
|
|
Q05. I want to split a large partition into three or four smaller ones.
|
|
Can I use FIPS multiple times?
|
|
|
|
A05.
|
|
----
|
|
Yes. You must format the newly created partitions between successive uses
|
|
of FIPS. Regarding the cluster size, consider the following example:
|
|
|
|
- Suppose you want to split a 1GB hard disk into four partitions of
|
|
256K each.
|
|
|
|
- The original partition is formatted with a cluster size of 16KB.
|
|
|
|
- The first split is into 256KB / 768KB. The cluster size of the first
|
|
partition remains 16KB, although 4KB would be enough. It can only be
|
|
changed by reformatting, see A04.
|
|
|
|
- The new partition of 768KB is formatted, still with a cluster size of 16KB.
|
|
|
|
- The second split is made into 256KB / 256KB / 512KB.
|
|
|
|
- The third partition is formatted with a cluster size of 8KB.
|
|
|
|
- The third split is made into 256KB / 256KB / 256KB / 256KB.
|
|
|
|
- The fourth partition is formatted with a cluster size of 4KB.
|
|
|
|
- So now you have the cluster sizes 16K - 16K - 8K - 4K.
|
|
|
|
- Since the second and third partition are still empty, it is possible
|
|
to reformat them to a cluster size of 4K. See A04. for details on
|
|
reformatting.
|
|
|
|
|
|
---------
|
|
|
|
Q06. FIPS creates a second primary DOS partition by default. Is this allowed?
|
|
|
|
|
|
A06.
|
|
----
|
|
See the section 'What FIPS does' in FIPS.DOC for a discussion of this
|
|
issue.
|
|
|
|
|
|
---------
|
|
|
|
Q07. What does the message "Info: Partition table inconsistency" mean?
|
|
|
|
A07.
|
|
----
|
|
|
|
The partition table in the master boot record (root sector) consists of
|
|
four entries with several fields each. The strange thing about this table
|
|
is that some of the fields are redundant. Look here:
|
|
|
|
| | Start | | End | Start |Number of|
|
|
Part.|bootable|Head Cyl. Sector|System|Head Cyl. Sector| Sector |Sectors | MB
|
|
-----+--------+----------------+------+----------------+--------+---------+----
|
|
1 | yes | 0 148 1| 83h| 15 295 63| 149184| 149184| 72
|
|
2 | no | 1 0 1| 06h| 15 139 63| 63| 141057| 68
|
|
3 | no | 0 140 1| 06h| 15 147 63| 141120| 8064| 3
|
|
4 | no | 0 0 0| 00h| 0 0 0| 0| 0| 0
|
|
|
|
|
|
The start and end (head/cylinder/sector) values can be calculated from the
|
|
start sector and number of sectors (after inquiring the BIOS about the number
|
|
of heads and number of sectors per track), and in fact that's exactly what
|
|
DOS does. These field are completely unused by DOS (and every other OS that
|
|
I know of), so they could as well be set to all zeroes. It does however not
|
|
hurt to keep them in a consistent state.
|
|
|
|
When fdisk creates a partition table entry, it should enter the correct
|
|
values in these fields. Now how can there be an inconsistency? There are
|
|
at least two possibilities:
|
|
|
|
1. Some fdisk programs seem to write incorrect values, especially end
|
|
cylinders that are off by one or two.
|
|
|
|
2. You have an EIDE drive that uses address translation. Modern hard disks
|
|
usually have more than 1024 cylinders, but DOS does not allow for cylinder
|
|
numbers greater than 1024. This caused the hard disk controller manufacturers
|
|
to implement a trick: they decrease the number of cylinders and increase
|
|
the number of heads that DOS is told when asking for the drive geometry.
|
|
So DOS thinks it has a drive with e.g. 63 sectors, 32 heads and 1000
|
|
cylinders, whereas the correct values are 63 sectors, 16 heads, 2000
|
|
cylinders. Now if DOS asks for the first sector on cylinder 500, it
|
|
really gets the first sector on cylinder 1000. This trick is called
|
|
address translation.
|
|
|
|
Some newer EIDE drives allow the address translation to be switched
|
|
on and off in the BIOS. In some cases this is changed after the disk
|
|
is formatted. This means that the drive geometry that DOS gets when
|
|
querying the disk controller differs from the geometry that the drive
|
|
was formatted with and which is reflected in the partition table. This
|
|
is no problem for DOS, but it was a problem for FIPS until release 1.4.
|
|
|
|
In the current release, if FIPS detects this kind of problem, it will
|
|
adapt the partition table to the changed disk geometry.
|
|
|
|
The bottomline is that you need not worry about this message, it is
|
|
perfectly normal.
|
|
|
|
|
|
---------
|
|
|
|
Q08. FIPS displays an error message and refuses to work. What should I do?
|
|
|
|
A08.
|
|
----
|
|
Although this is already addressed in FIPS.DOC, I can not emphasize it
|
|
enough: If you send me email, please include the _DEBUG SCRIPT_ that you can
|
|
produce with the -d switch. If you don't, I will have to ask you for it,
|
|
and it will take longer to solve your problem.
|
|
|
|
|
|
---------
|
|
|
|
Q09. FIPS does not recognize my SCSI disk.
|
|
|
|
A09.
|
|
----
|
|
There exist older SCSI adapters (an Adaptec 1524 was reported to have that
|
|
'feature') that need a device driver that is loaded from the config.sys
|
|
before the disk can be accessed (i.e. the system must be booted from a
|
|
second hard disk or floppy). This device driver does not provide a 'BIOS
|
|
level' interface but a 'DOS level' interface to the hard disk (for the
|
|
technicians: it hooks into the DOS interrupt 21h instead of the BIOS inter-
|
|
rupt 13h). This means that the partition table can only be accessed via a
|
|
special fdisk program that knows about the adapters' internals. FIPS will
|
|
not work on these drives (and in fact even DOS' fdisk won't either).
|
|
Possibly there exists a newer driver for that adapter that will provide a
|
|
BIOS level interface - ask the manufacturer.
|
|
|
|
|
|
---------
|
|
|
|
Q10. FIPS only finds a partition with partition type 56h, no DOS partition.
|
|
|
|
A10.
|
|
----
|
|
|
|
You have OnTrack Disk Manager installed. Read the relevant section in
|
|
SPECIAL.DOC.
|
|
|
|
|
|
---------
|
|
|
|
Q11. FIPS seems to work fine, and DOS sees the new partition, but Linux
|
|
fdisk sees only one partition.
|
|
|
|
A11.
|
|
----
|
|
|
|
Somehow DOS loads a different partition table than the one in the master
|
|
boot record.
|
|
|
|
There are two possible causes:
|
|
|
|
1. You use a device driver like OnTrack Disk Manager. See the file
|
|
SPECIAL.DOC
|
|
|
|
2. You have a virus in the master boot record. This does not happen often,
|
|
but it is quite possible. Some viruses install themselves in the MBR and
|
|
copy the original MBR to some other place. When DOS tries to access the
|
|
partition table, the virus intercepts the BIOS call and returns the backup
|
|
copy, in order to hide from possible detection. Check this with a virus
|
|
scanner after booting from a clean DOS boot disk. You can remove a virus
|
|
from the MBR by using DOS fdisk with the /mbr option, but be aware that
|
|
in this case the 'backup' partition table is not restored. Thus the boot
|
|
sector will contain the new info as changed by FIPS, but the partition table
|
|
will contain the original (single partition) setup. This inconsistency
|
|
must be corrected (e.g. with a disk editor).
|
|
|
|
|
|
---------
|
|
|
|
Q12. Does FIPS work with Windows 95?
|
|
|
|
A12.
|
|
----
|
|
|
|
Yes. The file system of Windows 95 is the same as that of DOS. The only
|
|
difference are the long file names, but FIPS works on a level below the
|
|
directory level, so this is not a problem. Several people reported that
|
|
FIPS worked flawlessly on a Win 95 partition.
|