If a user tries to use a swap file instead of a swap partition (which is a perfectly valid choice and supported by mkswap, swapon etc.), then during the init scripts, the swap file will fail to be activated. The problem is due to the fact that in /etc/rc.sysinit, swap is activated before the other filesystems in /etc/fstab have been mounted completely. The filesystem containing the swapfile ought to be mounted as RW when swapon -a is invoked, otherwise swapon will find that the swap space cannot be written to (which is obviously useless) and fail.

The solution is simple, the line used to activate swap can be moved down in /etc/rc.sysinit to a point just after when the local filesystems are mounted RW.

Cheers, Jack H