Questions about partitioning and ext3

[Daniel Pittman]

JEMF <jemf at gabcmt.eb.mil.br> writes:
> Daniel Pittman escreveu:
>> JEMF <jemf at gabcmt.eb.mil.br> writes:
>>>Why? Geometry?
>> Yup; when the flash device pretends to have geometry so that it doesn't
>> confuse software that still lives in DOS land, it caused that.
>
> How the system calculate this geometry?

Basically, magic.  Seriously, there are a bunch of heuristics, or it can
come from the DOS partition table, or from the BIOS, but it really is
pretty much just invented in (hopefully) the same way that DOS-ish
operating systems and the BIOS will do, so they also work.

>>>I think 8239 KB (8.05 MB) was used by journal. But the amount of  blocks
>>>decreased after formatted (471665 to 456730). Why?
>> The difference, of around 23,000 blocks, is five percent of the
>> available space on the filesystem.
>
> No! The difference is 14935 blocks! I mentioned in the previous message
> the difference between sizes of the unformatted partition and formated
> partition.

You are right -- my mental math is broken this morning.  I approximated
five percent in my head, then managed to miss-subtract the two numbers
to make them match.  How embarrassing. :/

My expectation would be that the difference is caused by filesystem
meta-data, such as inode allocation tables, which consume some storage
space on the raw device, but are not available for file storage.

The ext3 file system uses fixed size and location tables, so that space
is consumed irregardless of the space used for files.

   Daniel