On Wed, Jan 07, 2026 at 08:00:17AM -0700, Theo de Raadt wrote:
> > With fmt_scaled(1) the df utility reports 37 million fewer inodes
> > than actually are available, and worse, it reports 400K more inodes
> > than actually are in use. In larger file system the deviation
> > matters.
> 
> It is obviously better if it rounds DOWN, than if it rounds UP.
> 
> If it rounds down, your eyeball measure will be short at runtime.

My apologies, I typed 'more' where I should've typed 'less'.

Perhaps underreporting the number of free inodes is tolerable, but it
seems an unnecessary complication when the utility underreports the
number of USED inodes.

I looked into improving at least the column alignment (without applying
any scaling to the numbers), but gave up on that direction because
widening the 'iused' and 'ifree' columns even further seems just a
stopgap measure as disks are becoming larger and larger (and perhaps
worse, widening those columns takes away from space on the same line
that probably better is used for printing the mountpoint).

Having considered various alternatives, I really think it is best to
scale a decimal number as a decimal number and use this diff
https://marc.info/?l=openbsd-tech&m=176778893806944&w=2

Kind regards,

Job