[nSLUG] fdisk and cylinder start/end overlap

Donald Teed donald.teed at gmail.com
Tue Feb 1 19:23:55 AST 2005

I've noticed that with some systems, using the n command
in fdisk to generate a series of partitions will behave differently.

Here is a sample to reference:

/dev/hdc5              75         464     3132643+  fd  Linux raid autodetect
/dev/hdc6             465         720     2056288+  fd  Linux raid autodetect

I arrived at this by using the default suggested starting
cylinder when creating /dev/hdc6  This is with Gentoo,
fdisk 2.12i

On other systems, I have seen the next starting cylinder number
suggested as the same value as the end cylinder of
the previous partition.  That is, fdisk would suggest a starting
cylinder of 464 for /dev/hdc6  I know for certain I've seen this
on Gentoo on my Sun, and I've resisted it making that default
and used the next number up.  I also think this has happened
to me with Xandros Linux on x86.

The version of fdisk on the Sun Gentoo is also
2.12i (although using SCSI, if that could matter).
I've also checked Slackware 9 and it uses fdisk and
increments the starting cylinder.

This is a quirky difference, but otherwise I'd like to know if there
is a real danger in overlapping partitions when the start
and end cylinder numbers are shared between two partitions.


More information about the nSLUG mailing list