freebsd-skq/share/man/man9/sysctl.9
jhb 9ab3c32d73 - Document the UQUAD sysctl variants.
- Clarify that exactly one of the "access" flags is required and
  list the optional flags in a separate list.  Prefer bundling
  CTLFLAG_TUN into the access flag by not documenting it as an
  optional flag to set.

Approved by:	re (glebius)
MFC after:	1 week
2013-09-12 20:49:20 +00:00

312 lines
11 KiB
Groff

.\"
.\" Copyright (c) 2006 Robert N. M. Watson
.\" All rights reserved.
.\"
.\" 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 September 12, 2013
.Dt SYSCTL 9
.Os
.Sh NAME
.Nm SYSCTL_DECL ,
.Nm SYSCTL_INT ,
.Nm SYSCTL_LONG ,
.Nm SYSCTL_NODE ,
.Nm SYSCTL_OPAQUE ,
.Nm SYSCTL_PROC ,
.Nm SYSCTL_QUAD ,
.Nm SYSCTL_STRING ,
.Nm SYSCTL_STRUCT ,
.Nm SYSCTL_UINT ,
.Nm SYSCTL_ULONG ,
.Nm SYSCTL_UQUAD
.Nd Static sysctl declaration functions
.Sh SYNOPSIS
.In sys/types.h
.In sys/sysctl.h
.Fn SYSCTL_DECL name
.Fn SYSCTL_INT parent nbr name access ptr val descr
.Fn SYSCTL_LONG parent nbr name access ptr val descr
.Fn SYSCTL_NODE parent nbr name access handler descr
.Fn SYSCTL_OPAQUE parent nbr name access ptr len fmt descr
.Fn SYSCTL_PROC parent nbr name access ptr arg handler fmt descr
.Fn SYSCTL_QUAD parent nbr name access ptr val descr
.Fn SYSCTL_STRING parent nbr name access arg len descr
.Fn SYSCTL_STRUCT parent nbr name access ptr type descr
.Fn SYSCTL_UINT parent nbr name access ptr val descr
.Fn SYSCTL_ULONG parent nbr name access ptr val descr
.Fn SYSCTL_UQUAD parent nbr name access ptr val descr
.Sh DESCRIPTION
The
.Nm SYSCTL
kernel interfaces allow code to statically declare
.Xr sysctl 8
MIB entries, which will be initialized when the kernel module containing the
declaration is initialized.
When the module is unloaded, the sysctl will be automatically destroyed.
.Pp
Sysctl nodes are created in a hierarchical tree, with all static nodes being
represented by named C data structures; in order to create a new node under
an existing node in the tree, the structure representing the desired parent
node must be declared in the current context using
.Fn SYSCTL_DECL .
.Pp
New nodes are declared using one of
.Fn SYSCTL_INT ,
.Fn SYSCTL_LONG ,
.Fn SYSCTL_NODE ,
.Fn SYSCTL_OPAQUE ,
.Fn SYSCTL_PROC ,
.Fn SYSCTL_QUAD ,
.Fn SYSCTL_STRING ,
.Fn SYSCTL_STRUCT ,
.Fn SYSCTL_UINT ,
.Fn SYSCTL_ULONG ,
and
.Fn SYSCTL_UQUAD .
Each macro accepts a parent name, as declared using
.Fn SYSCTL_DECL ,
an OID number, typically
.Dv OID_AUTO ,
a node name, a set of control and access flags, and a description.
Depending on the macro, a pointer to a variable supporting the MIB entry, a
size, a value, and a function pointer implementing the MIB entry may also be
present.
.Pp
For most of the above macros, declaring a type as part of the access flags is
not necessary \[em] however, when declaring a sysctl implemented by a function,
including a type in the access mask is required:
.Bl -tag -width ".Dv CTLTYPE_STRING"
.It Dv CTLTYPE_NODE
This is a node intended to be a parent for other nodes.
.It Dv CTLTYPE_INT
This is a signed integer.
.It Dv CTLTYPE_STRING
This is a nul-terminated string stored in a character array.
.It Dv CTLTYPE_S64
This is a 64-bit signed integer.
.It Dv CTLTYPE_OPAQUE
This is an opaque data structure.
.It Dv CTLTYPE_STRUCT
Alias for
.Dv CTLTYPE_OPAQUE .
.It Dv CTLTYPE_UINT
This is an unsigned integer.
.It Dv CTLTYPE_LONG
This is a signed long.
.It Dv CTLTYPE_ULONG
This is an unsigned long.
.It Dv CTLTYPE_U64
This is a 64-bit unsigned integer.
.El
.Pp
All sysctl types except for new node declarations require one of the following
flags to be set indicating the read and write disposition of the sysctl:
.Bl -tag -width ".Dv CTLFLAG_ANYBODY"
.It Dv CTLFLAG_RD
This is a read-only sysctl.
.It Dv CTLFLAG_RDTUN
This is a read-only sysctl which can be set by a system tunable.
.It Dv CTLFLAG_WR
This is a writable sysctl.
.It Dv CTLFLAG_RW
This sysctl is readable and writable.
.It Dv CTLFLAG_RWTUN
This sysctl is readable and writable and can also be set by a system tunable.
.El
.Pp
Additionally, any of the following optional flags may also be specified:
.Bl -tag -width ".Dv CTLFLAG_ANYBODY"
.It Dv CTLFLAG_ANYBODY
Any user or process can write to this sysctl.
.It Dv CTLFLAG_SECURE
This sysctl can be written to only if the effective securelevel of the
process is \[<=] 0.
.It Dv CTLFLAG_PRISON
This sysctl can be written to by processes in
.Xr jail 2 .
.It Dv CTLFLAG_SKIP
When iterating the sysctl name space, do not list this sysctl.
.It Dv CTLFLAG_TUN
Advisory flag that a system tunable also exists for this variable.
.El
.Pp
When creating new sysctls, careful attention should be paid to the security
implications of the monitoring or management interface being created.
Most sysctls present in the kernel are read-only or writable only by the
superuser.
Sysctls exporting extensive information on system data structures and
operation, especially those implemented using procedures, will wish to
implement access control to limit the undesired exposure of information about
other processes, network connections, etc.
.Pp
The following top level sysctl name spaces are commonly used:
.Bl -tag -width ".Va regression"
.It Va compat
Compatibility layer information.
.It Va debug
Debugging information.
Various name spaces exist under
.Va debug .
.It Va hw
Hardware and device driver information.
.It Va kern
Kernel behavior tuning; generally deprecated in favor of more specific
name spaces.
.It Va machdep
Machine-dependent configuration parameters.
.It Va net
Network subsystem.
Various protocols have name spaces under
.Va net .
.It Va regression
Regression test configuration and information.
.It Va security
Security and security-policy configuration and information.
.It Va sysctl
Reserved name space for the implementation of sysctl.
.It Va user
Configuration settings relating to user application behavior.
Generally, configuring applications using kernel sysctls is discouraged.
.It Va vfs
Virtual file system configuration and information.
.It Va vm
Virtual memory subsystem configuration and information.
.El
.Sh EXAMPLES
Sample use of
.Fn SYSCTL_DECL
to declare the
.Va security
sysctl tree for use by new nodes:
.Bd -literal -offset indent
SYSCTL_DECL(_security);
.Ed
.Pp
Examples of integer, opaque, string, and procedure sysctls follow:
.Bd -literal -offset indent
/*
* Example of a constant integer value. Notice that the control
* flags are CTLFLAG_RD, the variable pointer is NULL, and the
* value is declared.
*/
SYSCTL_INT(_debug_sizeof, OID_AUTO, bio, CTLFLAG_RD, NULL,
sizeof(struct bio), "sizeof(struct bio)");
/*
* Example of a variable integer value. Notice that the control
* flags are CTLFLAG_RW, the variable pointer is set, and the
* value is 0.
*/
static int doingcache = 1; /* 1 => enable the cache */
SYSCTL_INT(_debug, OID_AUTO, vfscache, CTLFLAG_RW, &doingcache, 0,
"Enable name cache");
/*
* Example of a variable string value. Notice that the control
* flags are CTLFLAG_RW, that the variable pointer and string
* size are set. Unlike newer sysctls, this older sysctl uses a
* static oid number.
*/
char kernelname[MAXPATHLEN] = "/kernel"; /* XXX bloat */
SYSCTL_STRING(_kern, KERN_BOOTFILE, bootfile, CTLFLAG_RW,
kernelname, sizeof(kernelname), "Name of kernel file booted");
/*
* Example of an opaque data type exported by sysctl. Notice that
* the variable pointer and size are provided, as well as a format
* string for sysctl(8).
*/
static l_fp pps_freq; /* scaled frequence offset (ns/s) */
SYSCTL_OPAQUE(_kern_ntp_pll, OID_AUTO, pps_freq, CTLFLAG_RD,
&pps_freq, sizeof(pps_freq), "I", "");
/*
* Example of a procedure based sysctl exporting string
* information. Notice that the data type is declared, the NULL
* variable pointer and 0 size, the function pointer, and the
* format string for sysctl(8).
*/
SYSCTL_PROC(_kern_timecounter, OID_AUTO, hardware, CTLTYPE_STRING |
CTLFLAG_RW, NULL, 0, sysctl_kern_timecounter_hardware, "A",
"");
.Ed
.Sh SYSCTL NAMING
When adding, modifying, or removing sysctl names, it is important to be
aware that these interfaces may be used by users, libraries, applications,
or documentation (such as published books), and are implicitly published application interfaces.
As with other application interfaces, caution must be taken not to break
existing applications, and to think about future use of new name spaces so as
to avoid the need to rename or remove interfaces that might be depended on in
the future.
.Pp
The semantics chosen for a new sysctl should be as clear as possible,
and the name of the sysctl must closely reflect its semantics.
Therefore the sysctl name deserves a fair amount of consideration.
It should be short but yet representative of the sysctl meaning.
If the name consists of several words, they should be separated by
underscore characters, as in
.Va compute_summary_at_mount .
Underscore characters may be omitted only if the name consists of not more
than two words, each being not longer than four characters, as in
.Va bootfile .
For boolean sysctls, negative logic should be totally avoided.
That is, do not use names like
.Va no_foobar
or
.Va foobar_disable .
They are confusing and lead to configuration errors.
Use positive logic instead:
.Va foobar ,
.Va foobar_enable .
.Pp
A temporary sysctl node that should not be relied upon must be designated
as such by a leading underscore character in its name. For example:
.Va _dirty_hack .
.Sh SEE ALSO
.Xr sysctl 3 ,
.Xr sysctl 8 ,
.Xr sysctl_add_oid 9 ,
.Xr sysctl_ctx_free 9 ,
.Xr sysctl_ctx_init 9 ,
.Xr sysctl_remove_oid 9
.Sh HISTORY
The
.Xr sysctl 8
utility first appeared in
.Bx 4.4 .
.Sh AUTHORS
.An -nosplit
The
.Nm sysctl
implementation originally found in
.Bx
has been extensively rewritten by
.An Poul-Henning Kamp
in order to add support for name lookups, name space iteration, and dynamic
addition of MIB nodes.
.Pp
This man page was written by
.An Robert N. M. Watson .