> hi. i've got a lot of memory on one of my systems, so i want to put up a 1GB
> memory file system.
> 
> if i put this line in /etc/fstab
> 
> 	swap /big mfs rw,-s=1g 0 0
> 
> then when i mount it, i get:
>     
>     # mount /big
>     Warning: inode blocks/cyl group (141) >= data blocks (29) in last
>         cylinder group. This implies 470 sector(s) cannot be allocated.
> 
> and i get a file system with about 124MB, e.g.,
>     
>     # df -k /big
>     Filesystem  1K-blocks     Used     Avail Capacity  Mounted on
>     mfs:13118      127495        1    121119     0%    /big
> 
> so, what am i doing wrong?

well, after consulting the source tree, here's the story: mount_mfs is a link to
newfs. when invoked as mount_mfs, the program looks at the RLIMIT_DATA to
determine how large it can make a memory-based filesystem, and, if it doesn't
have enough memory, then it silently reduces the request to something it can
manage.

on my 1.6 system, "ulimit -d" reports 131204KB.

this is perfectly reasonable except for the "silently" part, mount_mfs should
either warn or error.

    
so, here's my follow-up question: my gut reaction to fixing this is to:

1. rename mount_mfs to mount_mfs2 (you need "mfs" in the filename for
newfs to go into mount_mfs mode)
    
2. create a three line shell script as mount_mfs:
    
    #!/bin/sh
    
    ulimit -d unlimited
    mount_mfs2 "$@"
    

can someone suggest a better fix.
    
/mtr