rump_sp —
rump remote system call support
The rump_sp facility allows clients to attach to a rump
kernel server over a socket and perform system calls. While making a local
rump system call is faster than calling the host kernel, a remote system call
over a socket is slower. This facility is therefore meant mostly for
operations which are not performance critical, such as configuration of a rump
kernel server.
The NetBSD base system comes with multiple preinstalled
clients which can be used to configure a rump kernel and request diagnostic
information. These clients run as hybrids partially in the host system and
partially against the rump kernel. For example, network-related clients will
typically avoid making any file system related system calls against the rump
kernel, since it is not guaranteed that a rump network server has file system
support. Another example is DNS: since a rump server very rarely has a DNS
service configured, host networking is used to do DNS lookups.
Some examples of clients include
rump.ifconfig which configures interfaces,
rump.sysctl which is used to access the
sysctl(7) namespace and
rump.traceroute which is used to display a network
trace starting from the rump kernel.
Also, almost any unmodified dynamically linked application (for
example telnet(1) or
ls(1)) can be used as a rump
kernel client with the help of system call hijacking. See
rumphijack(3) for more
information.
A remote rump server is specified using an URL. Currently two types of URLs are
supported: TCP and local domain sockets. The TCP URL is of the format
tcp://ip.address:port/ and the local domain URL is unix://path. The latter can
accept relative or absolute paths. Note that absolute paths require three
leading slashes.
To preserve the standard usage of the rump clients' counterparts
the environment variable RUMP_SERVER is used to
specify the server URL. To keep track of which rump kernel the current shell
is using, modifying the shell prompt is recommended -- this is analogous to
the visual clue you have when you login from one machine to another.
The current scheme gives all connecting clients root credentials. It is
recommended to take precautions which prevent unauthorized access. For a unix
domain socket it is enough to prevent access to the socket using file system
permissions. For TCP/IP sockets the only available means is to prevent network
access to the socket with the use of firewalls. More fine-grained access
control based on cryptographic credentials may be implemented at a future
date.
Get a list of file systems supported by a rump kernel server (in case that
particular server does not support file systems, an error will be returned):
$ env RUMP_SERVER=unix://sock rump.sysctl vfs.generic.fstypes
rump_sp first appeared in NetBSD
6.0.