>  > Won't matter much. Right now to mount an mfs file system is a fixed size.
>  > Memory is allocated at fs creation, not when files are created. So
>  > deleting a file won't change its memory footprint.
>
> Well, sort of.  The memory is "allocated", but it's lazily allocated,
> i.e. physical pages aren't actually there until they are first used.
>
> The pages for an MFS can also be paged out to swap space.
could it be changed that when file is deleted all pages belonging to it
will be unmapped (remapped to /dev/zero back)?

now when i put 50MB file to mfs and then delete it, mfs process has 50MB
RSS and have to swap it out