NAME

rump_nfs - mount the nfs file system using a userspace server

SYNOPSIS

file-system PUFFS pseudo-device putter

rump_nfs [options] special node

DESCRIPTION

The rump_nfs utility can be used to mount nfs file systems. It uses rump(3) and p2k(3) to facilitate running the file system as a server in userspace. As opposed to mount_nfs(8), rump_nfs does not use file system code within the kernel and therefore does not require kernel support except puffs(4). Apart from a minor speed penalty (starting from 10% and depending on the workload and file system in question), there is no difference to using in-kernel code.

In case mounting a file system image from a regular file, rump_nfs does not require the use of vnconfig(8) unlike kernel file systems. Instead, the image path can be directly passed as the special file path. The exception is if the image contains a disklabel. In this case vnconfig is required to resolve the start offset for the correct partition within the image.

It is recommended that untrusted file system images be mounted with rump_nfs instead of mount_nfs(8). Corrupt file system images commonly cause the file system to crash the entire kernel, but with rump_nfs only the userspace server process will dump core.

To use rump_nfs via mount(8), the flags -o rump and -t nfs should be given. Similarly, rump_nfs is run instead of mount_nfs(8) if ``rump'' is added to the options field of fstab(5).

Please see mount_nfs(8) for a full description of the available command line options.

SEE ALSO

p2k(3), puffs(3), rump(3), mount_nfs(8)

HISTORY

The rump_nfs utility first appeared in NetBSD5.0. It is currently considered experimental.