> thanks for the review of my patch!
No worries. :)
> Main idea here was to avoid the situation that df errors out if the
> file does not yet exist:
I understand, and this is why I suggested touch before df.
> Wouldn't accessing "/var/." after doing e.g. a stat on the (possibly
> non-existing) file suffice for that? At least my gut feeling says that
> I should avoid write access before actually creating the file.
Yes, it would work most of the time. I believe there are cases where
individual files have different file system stats (if they are bind
mounted from some other partition), but you might not want to worry
about that use case. :)