Date: Thu, 22 Dec 2005 19:23:24 GMT From: "gs_stoller@juno.com" <gs_stoller@juno.com> To: freebsd-questions@freebsd.org Cc: GS_Stoller@juno.com Subject: slice-table / partition-table Message-ID: <20051222.112422.14922.174120@webmail49.nyc.untd.com>
next in thread | raw e-mail | index | archive | help
I have 2 hard drives on my system. One (my main drive) has an
MS-DOS slice (naturally the first slice) and 3 FreeBSD slices.
The other drive had 1 OS2 slice (just because an OS2 CDROM was used
to slice it up) and 1 MS-DOS slice. At one time, I was thinking of
putting the boot program on the second hard drive, and may have done
it with the CDROM of FreeBSD 4.7 . Recently, while I was running
MS-DOS and processing e-mail (trying to remove an attachment from
a letter in order to save it in the MS-DOS slice on the second hard
drive) through juno.com , I discovered that the MS-DOS slice on this
second hard drive was hosed (naturally, that discovery could only be
made while running MS-DOS ), and verified that fact by trying to go to =
a directory on that second drive. MS-DOS didn't go there but complain=
ed, and following that I couldn't see anything on that second drive.
To investigate this, I rebooted into my FreeBSD 4.3 slice and
found that the slice-table (often called "partition-table") on the sec=
ond drive was identical to that of the first drive. In fact, when I tri=
ed fdisk on each drive, I found the results identical except for the s=
ize (# of cylinders) of the drives. Originally, the second drive's fdi=
sk results stated 3 heads and some sectors per cylinder (but I don't re=
call how many). I conclude that there is a strong possibility that only=
the slice-table sector has been copied (from the first drive), though=
I have no idea how this could have happened (though I consider the poss=
ible putting of the boot program on the second hard drive as possibly co=
ntributory). Does anyone have any idea of how this can happen? Has any=
one seen this or the like before?
I also conclude that there is a strong possibility that the rest of the=
second disk (i.e., my data on it) is okay, and if I can rebuild the sl=
ice-table then I can recover the rest of the disk and all my data. I t=
hink I can do this if I can get the following things/information:
1) is there a duplicate of the slice-table =
somewhere on the disk, and if so, where is it, and how can I exam=
ine it rather than the primary slice-table ,
and is there a way to have its contents replace
the contents of the primary slice-table ,
2) is there a program that can look at a(n entire) disk and break i=
t up into contiguous sectors that belong to particular systems (in m=
y case OS2 or MS-DOS ), {I would be willing to write such a pr=
ogram if someone would tell me how to read sectors without a slice-=
table [say with just an LBA number] and how to recognize whet=
her a sector belongs to OS2 or MS-DOS , and any timing constraint=
s for reading/writing cylinders [e.g., usual timing for getting a =
read/write head to cylinder 0 , the time
to move between cylinders, etc.], if such constraints exist [mayb=
e we have fixed heads and so no timing constraints]},
3) FreeBSD programs (or Linux programs or MS-DOS programs i=
f FreeBSD programs aren't available) that can read any sector give=
n its LBA number or the like [somewhat like fdisk reads/writ=
es of sector 0 ], what could be really useful [if I have to rebuild=
files] is a program that reads a range of sectors and greps it=
s contents, reporting the LBA =
number [and some contents] of those sectors that pass the grep =
test,
4) the structure of the slice-table's sector and =
of the slice-table and FreeBSD programs that can write specif=
ic values (given as arguments [say in structures]) into the sl=
ice-table without affecting the contents of any other sector on that=
disk,
5) how to mark a slice "<UNUSED>",
6) FreeBSD programs (or Linux programs or MS-DOS programs if =
FreeBSD programs aren't available) other than fdisk that can wri=
te the slice-table's sector (including its magic number and boo=
t code) without affecting the contents of any other sectoron that di=
sk,
7) how can I recognize sectors that were in a slice assigned for =
MS-DOS usage, and/or is there a way of finding the "first" (i.e., o=
ne with lowest LBA ) and "last" sector assigned for MS-DOS us=
age or used by MS-DOS ,
8) FreeBSD programs (or Linux programs or MS-DOS programs if =
FreeBSD programs aren't available) software to rebuild files and =
[Windows-98] directories from fragments (with human direction) [t=
his is just in case there is more damage than I suspect at present,
I'm hoping not to need this one],
9) any caveats or information that would prove useful in this ende=
avor.
If I have to rebuild the slice-table from scratch, I presume that I c=
an make any entries I wish in the slice-table and then look at the con=
tents of the slices to see if they are what I would like to find there=
and appropriately change the slice-table until I find it set up as it=
was before. Anything wrong with my presumption?
If I succeed in this, I'm thinking of publishing an article on how to r=
ecover from this sort of problem.
uname -a for my system gives:
FreeBSD 4.3-RELEASE FreeBSD 4.3-RELEASE #0: Sat Apr 21 10:54:49 GMT 200=
1 jkh@narf.osd.bsdi.com:/usr/src/sys/compile/GENERIC i386
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20051222.112422.14922.174120>