General
cancel
Showing results for 
Search instead for 
Did you mean: 

strange error reported on fdisk's p option :: boundary limit

Ragu_1
Regular Advisor

strange error reported on fdisk's p option :: boundary limit

I am working on a Debian `sid' system at home. I am getting this echo for the `p' option in `fdisk /dev/hda', "Partition 1 does not end on cylinder boundary". Have got `sarge' on another partition of the same HDD, the fdisk there seems normal with no error being reported on the p option. Passing lba32 to the kernel via boot-loader, LBA on in the BIOS setting.
Share and share alike
2 REPLIES
Stuart Browne
Honored Contributor

Re: strange error reported on fdisk's p option :: boundary limit

That's not out of the ordinary.

It should work in that manner, but it will result in some loss of available space with regards to partitioning (from memory).

If you just juggle the numbers a bit, you can get it to finish on a nice cylinder boundary quite easily (i.e. use the +M instead of block counts).
One long-haired git at your service...
Jerome Henry
Honored Contributor

Re: strange error reported on fdisk's p option :: boundary limit

Hi,

I feel Stuart could have been rewarded more on this... he gave you the solution. To re-explain his tip :
this message means that your partition does't end at the end of a cylinder.
Going the easy way, disks are made of plates, read by heads. Each side of the plate is composed of cylinders, on which sectors are drawn.
Partitionning a disk means deciding that 'those cylinders read by those heads will be a whole', then 'those other cylinders read by those heads another one', and so on.
On a normal setting , a partition is a certain number of cylinders and associated sectors. You get this message if the number of sectors read as being the partition doesn't represent a full number of cylinder (say 52,21 cylinders instead of 52,00). It's abnormal, but most linux won't complain more.
The cause can be :
- you installed 2 OSes, that don't use the same counting system (say CHS and LBA, or LBA and LBA32). It seems to be the case for you. The most recent install re-write the MBR and writes its count. The older system complains as the counting method is not adapted to what it seens. It's still a false positive, unless the other OS is NTFS-type, for which partition management if quite odd ! Most of the time, it complains but still starts.
- It's a false positive, as your BIOS is older type and multimode (CHS and LBA for example), preventing your OS to read precisely BIOS values and type. This false positive doesn't exist anymore in most of recent versions.
From the older version, you can change the xay fdisk counts, using the arguments Stuart suggested, you'll find more details on man fdisk.

hth
J
You can lean only on what resists you...