diff options
author | rodrigc <rodrigc@FreeBSD.org> | 2014-11-20 09:56:34 +0000 |
---|---|---|
committer | rodrigc <rodrigc@FreeBSD.org> | 2014-11-20 09:56:34 +0000 |
commit | cf0bd1042fef79312e6ee5944aab241db7aa1aae (patch) | |
tree | d728486a9161b2e52f09a95110f6ca525149d0c1 /share/man/man9 | |
parent | 3cdc37eb525944302c3b26494c661a5090ad772d (diff) | |
download | FreeBSD-src-cf0bd1042fef79312e6ee5944aab241db7aa1aae.zip FreeBSD-src-cf0bd1042fef79312e6ee5944aab241db7aa1aae.tar.gz |
MFp4: @179066
Add page which describes VNET network stack virtualization infrastructure.
Submitted by: bz
Sponsored by: The FreeBSD Foundation
Sponsored by: CK Software GmbH
Diffstat (limited to 'share/man/man9')
-rw-r--r-- | share/man/man9/Makefile | 1 | ||||
-rw-r--r-- | share/man/man9/vnet.9 | 502 |
2 files changed, 503 insertions, 0 deletions
diff --git a/share/man/man9/Makefile b/share/man/man9/Makefile index 52181a2..c4884eb 100644 --- a/share/man/man9/Makefile +++ b/share/man/man9/Makefile @@ -345,6 +345,7 @@ MAN= accept_filter.9 \ vmem.9 \ vn_fullpath.9 \ vn_isdisk.9 \ + vnet.9 \ vnode.9 \ VOP_ACCESS.9 \ VOP_ACLCHECK.9 \ diff --git a/share/man/man9/vnet.9 b/share/man/man9/vnet.9 new file mode 100644 index 0000000..68689ee --- /dev/null +++ b/share/man/man9/vnet.9 @@ -0,0 +1,502 @@ +.\"- +.\" Copyright (c) 2010 The FreeBSD Foundation +.\" All rights reserved. +.\" +.\" This documentation was written by CK Software GmbH under sponsorship from +.\" the FreeBSD Foundation. +.\" +.\" Redistribution and use in source and binary forms, with or without +.\" modification, are permitted provided that the following conditions +.\" are met: +.\" 1. Redistributions of source code must retain the above copyright +.\" notice, this list of conditions and the following disclaimer. +.\" 2. Redistributions in binary form must reproduce the above copyright +.\" notice, this list of conditions and the following disclaimer in the +.\" documentation and/or other materials provided with the distribution. +.\" +.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND +.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE +.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS +.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) +.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT +.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY +.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF +.\" SUCH DAMAGE. +.\" +.\" $FreeBSD$ +.\" +.Dd November 20, 2014 +.Dt VNET 9 +.Os +.Sh NAME +.Nm VNET +.Nd "network subsystem virtualization infrastructure" +.Sh SYNOPSIS +.Cd "options VIMAGE" +.Cd "options VNET_DEBUG" +.Pp +.In sys/vnet.h +.Pp +.\"------------------------------------------------------------ +.Ss "Constants and Global Variables" +.\" +.Dv VNET_SETNAME +.\" "set_vnet" +.Dv VNET_SYMPREFIX +.\" "vnet_entry_" +.Vt extern struct vnet *vnet0; +.\"------------------------------------------------------------ +.Ss "Variable Declaration" +.Fo VNET +.Fa "name" +.Fc +.\" +.Fo VNET_NAME +.Fa "name" +.Fc +.\" +.Fo VNET_DECLARE +.Fa "type" "name" +.Fc +.\" +.Fo VNET_DEFINE +.Fa "type" "name" +.Fc +.\" +.Bd -literal +#define V_name VNET(name) +.Ed +.\" ------------------------------------------------------------ +.Ss "Virtual Instance Selection" +.\" +.Fo CRED_TO_VNET +.Fa "struct ucred *" +.Fc +.\" +.Fo TD_TO_VNET +.Fa "struct thread *" +.Fc +.\" +.Fo P_TO_VNET +.Fa "struct proc *" +.Fc +.\" +.Fo IS_DEFAULT_VNET +.Fa "struct vnet *" +.Fc +.\" +.Fo VNET_ASSERT +.Fa exp msg +.Fc +.\" +.Fo CURVNET_SET +.Fa "struct vnet *" +.Fc +.\" +.Fo CURVNET_SET_QUIET +.Fa "struct vnet *" +.Fc +.\" +.Fo CURVNET_RESTORE +.Fc +.\" +.Fo VNET_ITERATOR_DECL +.Fa "struct vnet *" +.Fc +.\" +.Fo VNET_FOREACH +.Fa "struct vnet *" +.Fc +.\" ------------------------------------------------------------ +.Ss "Locking" +.\" +.Fo VNET_LIST_RLOCK +.Fc +.Fo VNET_LIST_RUNLOCK +.Fc +.Fo VNET_LIST_RLOCK_NOSLEEP +.Fc +.Fo VNET_LIST_RUNLOCK_NOSLEEP +.Fc +.\" ------------------------------------------------------------ +.Ss "Startup and Teardown Functions" +.\" +.Ft "struct vnet *" +.Fo vnet_alloc +.Fa void +.Fc +.\" +.Ft void +.Fo vnet_destroy +.Fa "struct vnet *" +.Fc +.\" +.Fo VNET_SYSINIT +.Fa ident +.Fa "enum sysinit_sub_id subsystem" +.Fa "enum sysinit_elem_order order" +.Fa "sysinit_cfunc_t func" +.Fa "const void *arg" +.Fc +.\" +.Fo VNET_SYSUNINIT +.Fa ident +.Fa "enum sysinit_sub_id subsystem" +.Fa "enum sysinit_elem_order order" +.Fa "sysinit_cfunc_t func" +.Fa "const void *arg" +.Fc +.\" ------------------------------------------------------------ +.Ss "Eventhandlers" +.\" +.Fo VNET_GLOBAL_EVENTHANDLER_REGISTER +.Fa "const char *name" +.Fa "void *func" +.Fa "void *arg" +.Fa "int priority" +.Fc +.\" +.Fo VNET_GLOBAL_EVENTHANDLER_REGISTER_TAG +.Fa "eventhandler_tag tag" +.Fa "const char *name" +.Fa "void *func" +.Fa "void *arg" +.Fa "int priority" +.Fc +.\" ------------------------------------------------------------ +.Ss "Sysctl Handling" +.Fo SYSCTL_VNET_INT +.Fa parent nbr name access ptr val descr +.Fc +.Fo SYSCTL_VNET_PROC +.Fa parent nbr name access ptr arg handler fmt descr +.Fc +.Fo SYSCTL_VNET_STRING +.Fa parent nbr name access arg len descr +.Fc +.Fo SYSCTL_VNET_STRUCT +.Fa parent nbr name access ptr type descr +.Fc +.Fo SYSCTL_VNET_UINT +.Fa parent nbr name access ptr val descr +.Fc +.Fo VNET_SYSCTL_ARG +.Fa req arg1 +.Fc +.\" ------------------------------------------------------------ +.Sh DESCRIPTION +.Nm +is the name of a technique to virtualize the network stack. +The basic idea is to change global resources most notably variables into +per network stack resources and have functions, sysctls, eventhandlers, +etc. access and handle them in the context of the correct instance. +Each (virtual) network stack is attached to a +.Em prison , +with +.Vt vnet0 +being the unrestricted default network stack of the base system. +.Pp +The global defines for +.Dv VNET_SETNAME +and +.Dv VNET_SYMPREFIX +are shared with +.Xr kvm 3 +to access internals for debugging reasons. +.\" ------------------------------------------------------------ +.Ss "Variable Declaration" +.\" +Variables are virtualized by using the +.Fn VNET_DEFINE +macro rather than writing them out as +.Em type name . +One can still use static initialization or storage class specifiers, e.g., +.Pp +.Dl Li static VNET_DEFINE(int, foo) = 1; +or +.Dl Li static VNET_DEFINE(SLIST_HEAD(, bar), bars); +.Pp +Static initialization is not possible when the virtualized variable +would need to be referenced, e.g., with +.Dq TAILQ_HEAD_INITIALIZER() . +In that case a +.Fn VNET_SYSINIT +based initialization function must be used. +.Pp +External variables have to be declared using the +.Fn VNET_DECLARE +macro. +In either case the convention is to define another macro, +that is then used throughout the implementation to access that variable. +The variable name is usually prefixed by +.Em V_ +to express that it is virtualized. +The +.Fn VNET +macro will then translate accesses to that variable to the copy of the +currently selected instance (see the +.Sx "Virtual instance selection" +section): +.Pp +.Dl Li #define V_name VNET(name) +.Pp +.Em NOTE: +Do not confuse this with the convention used by +.Xr VFS 9 . +.Pp +The +.Fn VNET_NAME +macro returns the offset within the memory region of the virtual network +stack instance. +It is usually only used with +.Fn SYSCTL_VNET_* +macros. +.\" ------------------------------------------------------------ +.Ss "Virtual Instance Selection" +.\" +There are three different places where the current virtual +network stack pointer is stored and can be taken from: +.Bl -enum -offset indent +.It +a +.Em prison : +.Dl "(struct prison *)->pr_vnet" +.Pp +For convenience the following macros are provided: +.Bd -literal -compact -offset indent +.Fn CRED_TO_VNET "struct ucred *" +.Fn TD_TO_VNET "struct thread *" +.Fn P_TO_VNET "struct proc *" +.Ed +.It +a +.Em socket : +.Dl "(struct socket *)->so_vnet" +.It +an +.Em interface : +.Dl "(struct ifnet *)->if_vnet" +.El +.Pp +.\" +In addition the currently active instance is cached in +.Dq "curthread->td_vnet" +which is usually only accessed through the +.Dv curvnet +macro. +.Pp +.\" +To set the correct context of the current virtual network instance, use the +.Fn CURVNET_SET +or +.Fn CURVNET_SET_QUIET +macros. +The +.Fn CURVNET_SET_QUIET +version will not record vnet recursions in case the kernel was compiled +with +.Cd "options VNET_DEBUG" +and should thus only be used in well known cases, where recursion is +unavoidable. +Both macros will save the previous state on the stack and it must be restored +with the +.Fn CURVNET_RESTORE +macro. +.Pp +.Em NOTE: +As the previous state is saved on the stack, you cannot have multiple +.Fn CURVNET_SET +calls in the same block. +.Pp +.Em NOTE: +As the previous state is saved on the stack, a +.Fn CURVNET_RESTORE +call has to be in the same block as the +.Fn CURVNET_SET +call or in a subblock with the same idea of the saved instances as the +outer block. +.Pp +.Em NOTE: +As each macro is a set of operations and, as previously explained, cannot +be put into its own block when defined, one cannot conditionally set +the current vnet context. +The following will +.Em not +work: +.Bd -literal -offset indent +if (condition) + CURVNET_SET(vnet); +.Ed +.Pp +nor would this work: +.Bd -literal -offset indent +if (condition) { + CURVNET_SET(vnet); +} +CURVNET_RESTORE(); +.Ed +.Pp +.\" +Sometimes one needs to loop over all virtual instances, for example to update +virtual from global state, to run a function from a +.Xr callout 9 +for each instance, etc. +For those cases the +.Fn VNET_ITERATOR_DECL +and +.Fn VNET_FOREACH +macros are provided. +The former macro defines the variable that iterates over the loop, +and the latter loops over all of the virtual network stack instances. +See +.Sx "Locking" +for how to savely traverse the list of all virtual instances. +.Pp +.\" +The +.Fn IS_DEFAULT_VNET +macro provides a safe way to check whether the currently active instance is the +unrestricted default network stack of the base system +.Pq Vt vnet0 . +.Pp +.\" +The +.Fn VNET_ASSERT +macro provides a way to conditionally add assertions that are only active with +.Cd "options VIMAGE" +compiled in and either +.Cd "options VNET_DEBUG" +or +.Cd "options INVARIANTS" +enabled as well. +It uses the same semantics as +.Xr KASSERT 9 . +.\" ------------------------------------------------------------ +.Ss "Locking" +.\" +For public access to the list of virtual network stack instances +e.g., by the +.Fn VNET_FOREACH +macro, read locks are provided. +Macros are used to abstract from the actual type of the locks. +If a caller may sleep while traversing the list, it must use the +.Fn VNET_LIST_RLOCK +and +.Fn VNET_LIST_RUNLOCK +macros. +Otherwise, the caller can use +.Fn VNET_LIST_RLOCK_NOSLEEP +and +.Fn VNET_LIST_RUNLOCK_NOSLEEP . +.\" ------------------------------------------------------------ +.Ss "Startup and Teardown Functions" +.\" +To start or tear down a virtual network stack instance the internal +functions +.Fn vnet_alloc +and +.Fn vnet_destroy +are provided and called from the jail framework. +They run the publicly provided methods to handle network stack +startup and teardown. +.Pp +For public control, the system startup interface has been enhanced +to not only handle a system boot but to also handle a virtual +network stack startup and teardown. +To the base system the +.Fn VNET_SYSINIT +and +.Fn VNET_SYSUNINIT +macros look exactly as if there were no virtual network stack. +In fact, if +.Cd "options VIMAGE" +is not compiled in they are compiled to the standard +.Fn SYSINIT +macros. +In addition to that they are run for each virtual network stack +when starting or, in reverse order, when shutting down. +.\" ------------------------------------------------------------ +.Ss "Eventhandlers" +.\" +Eventhandlers can be handled in two ways: +.Pp +.Bl -enum -offset indent -compact +.It +save the +.Em tags +returned in each virtual instance and properly free the eventhandlers +on teardown using those, or +.It +use one eventhandler that will iterate over all virtual network +stack instances. +.El +.Pp +For the first case one can just use the normal +.Xr EVENTHANDLER 9 +functions, while for the second case the +.Fn VNET_GLOBAL_EVENTHANDLER_REGISTER +and +.Fn VNET_GLOBAL_EVENTHANDLER_REGISTER_TAG +macros are provided. +These differ in that +.Fn VNET_GLOBAL_EVENTHANDLER_REGISTER_TAG +takes an extra first argument that will carry the +.Fa "tag" +upon return. +Eventhandlers registered with either of these will not run +.Fa func +directly but +.Fa func +will be called from an internal iterator function for each vnet. +Both macros can only be used for eventhandlers that do not take +additional arguments, as the variadic arguments from an +.Xr EVENTHANDLER_INVOKE 9 +call will be ignored. +.\" ------------------------------------------------------------ +.Ss "Sysctl Handling" +.\" +A +.Xr sysctl 9 +can be virtualized by using one of the +.Fn SYSCTL_VNET_* +macros. +.Pp +They take the same arguments as the standard +.Xr sysctl 9 +functions, with the only difference, that the +.Fa ptr +argument has to be passed as +.Ql &VNET_NAME(foo) +instead of +.Ql &foo +so that the variable can be selected from the correct memory +region of the virtual network stack instance of the caller. +.Pp +For the very rare case a sysctl handler function would want to +handle +.Fa arg1 +itself the +.Fn VNET_SYSCTL_ARG req arg1 +is provided that will translate the +.Fa arg1 +argument to the correct memory address in the virtual network stack +context of the caller. +.\" ------------------------------------------------------------ +.Sh SEE ALSO +.Xr jail 2 , +.Xr kvm 3 , +.Xr EVENTHANDLER 9 , +.\" .Xr pcpu 9 , +.Xr KASSERT 9 , +.Xr sysctl 9 +.\" .Xr SYSINIT 9 +.Sh HISTORY +The virtual network stack implementation first appeared in +.Fx 8.0 . +.Sh AUTHORS +This manual page was written by +.An Bjoern A. Zeeb, CK Software GmbH, +under sponsorship from the FreeBSD Foundation. |