Manual browser: rump(3)
|RUMP(3)||Library Functions Manual||RUMP(3)|
NAMErump — Rump Kernel
LIBRARYrump Library (librump, -lrump)
DESCRIPTIONA rump kernel is a virtualized kernel instance which retains only part of the NetBSD kernel. Rump kernels are aimed at virtualizing kernel drivers and do not support for example creating processes, managing virtual memory address spaces or scheduling threads. These features are unnecessary overhead when virtualizing drivers. Rump kernels are created by linking a desired set of components together. On NetBSD, these components are available as userspace libraries with the prefix librump. The high-level rumpuser(3) hypercall interface is used by a rump kernel to request resources and services from the host it runs on. Like any virtualized kernel, a rump kernel also provides its own set of namespaces, such as a file system hierarchy and TCP ports, that are independent of the ones on the host and of any other rump kernel instances. It should be noted that the presence of the provided namespaces depends on the components that the rump kernel was constructed with.
Since a rump kernel does not provide support for applications processes, existing entities are used as rump kernel clients. The relationship between the client and the rump kernel defines the execution model of the rump kernel. A local client will reside in the same address space and manipulate the rump kernel with function calls and direct memory references. Remote and microkernel clients are disjoint from the rump kernel and make requests though various protocols, see for example p2k(3), rump_sp(7), and rumphijack(3). Remote clients will also work over a TCP/IP network, or other similar communication medium.
A rump kernel is bootstrapped by calling rump_init(). On a POSIX Host, environment variables can be used to adjust some operating parameters:
- If set, the number indicates the number of virtual CPUs configured into a rump kernel. The special value "host" can be used to specify the number of of host CPUs available. If the value is unset, two CPUs will be configured.
- If set to non-zero, activates bootverbose.
- If set to 0, prevents the rump kernel from creating any kernel threads. This is possible usually only for file systems, as other subsystems depend on threads to work.
- If set, indicates the maximum amount of memory that a rump kernel will request from the hypervisor via rumpuser_malloc(). When the rump kernel is close to the allocation limit, it will attempt to make more memory available by flushing its caches. The default is as much as the host allows.
- Sets the value of the kern.maxvnodes sysctl node to the indicated amount. Adjusting this may be useful for example when testing vnode reclaim code paths. While the same value can be set by means of sysctl, the env variable is often more convenient for quick testing. As expected, this option has effect only in rump kernels which support VFS. The current default is 1024 vnodes.
The standardized way for a client to make requests into a rump kernel is to use rump kernel system calls, which have equivalent syntax and semantics with regular NetBSD system calls. The parameters are expected to be in the NetBSD type system unless a system call translation component, for example librumpkern_sys_linux, is linked into the rump kernel, in which case system call parameters will be automatically translated from the client type system into the NetBSD type system and back. The rump kernel system calls are made available to a client by including <rump/rump_syscalls.h>. It is also possible to configure unmodified binaries to make syscalls into a rump kernel with rumphijack(3).
SEE ALSOrump_server(1), p2k(3), rump_etfs(3), rump_lwproc(3), rumpclient(3), rumphijack(3), rumpuser(3), ukfs(3), rump_sp(7)
Antti Kantee, Flexible Operating System Internals: The Design and Implementation of the Anykernel and Rump Kernerls, Aalto University Doctoral Dissertations, 2012.
Antti Kantee, Rump File Systems: Kernel Code Reborn, Proceedings of the 2009 USENIX Annual Technical Conference, pp. 201-214, June 2009.
Arnaud Ysmal and Antti Kantee, Fs-utils: File Systems Access Tools for Userland, EuroBSDCon 2009, September 2009.
Antti Kantee, Environmental Independence: BSD Kernel TCP/IP in Userspace, Proceedings of AsiaBSDCon 2009, pp. 71-80, March 2009.
Antti Kantee, Rump Device Drivers: Shine On You Kernel Diamond, Proceedings of AsiaBSDCon 2010, pp. 75-84, March 2010.
Antti Kantee, Kernel Development in Userspace - The Rump Approach, BSDCan 2009, May 2009.
Web page with more information, including links to the publications: http://www.NetBSD.org/docs/rump/
HISTORYRump kernels appeared as an experimental concept in NetBSD 5.0. The first stable version was released in NetBSD 6.0.
|July 14, 2013||NetBSD 7.0|