freebsd-dev/share/man/man9/SDT.9
Mark Johnston 4b4ad3a219 Rename the dtrace-* man pages to dtrace_* for consistency with other
subsection man pages (e.g. geom_*, mac_*, snd_*).
2015-07-05 23:23:12 +00:00

347 lines
11 KiB
Groff

.\" Copyright (c) 2013-2015 Mark Johnston <markj@freebsd.org>
.\" 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 April 18, 2015
.Dt SDT 9
.Os
.Sh NAME
.Nm SDT
.Nd a DTrace framework for adding statically-defined tracing probes
.Sh SYNOPSIS
.In sys/param.h
.In sys/queue.h
.In sys/sdt.h
.Fn SDT_PROVIDER_DECLARE prov
.Fn SDT_PROVIDER_DEFINE prov
.Fn SDT_PROBE_DECLARE prov mod func name
.Fn SDT_PROBE_DEFINE prov mod func name
.Fn SDT_PROBE_DEFINE0 prov mod func name
.Fn SDT_PROBE_DEFINE1 prov mod func name arg0
.Fn SDT_PROBE_DEFINE2 prov mod func name arg0 arg1
.Fn SDT_PROBE_DEFINE3 prov mod func name arg0 arg1 arg2
.Fn SDT_PROBE_DEFINE4 prov mod func name arg0 arg1 arg2 arg3
.Fn SDT_PROBE_DEFINE5 prov mod func name arg0 arg1 arg2 arg3 arg4
.Fn SDT_PROBE_DEFINE6 prov mod func name arg0 arg1 arg2 arg3 arg4 arg5
.Fn SDT_PROBE_DEFINE7 prov mod func name arg0 arg1 arg2 arg3 arg4 arg5 \
arg6
.Fn SDT_PROBE_DEFINE0_XLATE prov mod func name
.Fn SDT_PROBE_DEFINE1_XLATE prov mod func name arg0 xarg0
.Fn SDT_PROBE_DEFINE2_XLATE prov mod func name arg0 xarg0 arg1 xarg1
.Fn SDT_PROBE_DEFINE3_XLATE prov mod func name arg0 xarg0 arg1 xarg1 \
arg2 xarg2
.Fn SDT_PROBE_DEFINE4_XLATE prov mod func name arg0 xarg0 arg1 xarg1 \
arg2 xarg2 arg3 xarg3
.Fn SDT_PROBE_DEFINE5_XLATE prov mod func name arg0 xarg0 arg1 xarg1 \
arg2 xarg2 arg3 xarg3 arg4 xarg4
.Fn SDT_PROBE_DEFINE6_XLATE prov mod func name arg0 xarg0 arg1 xarg1 \
arg2 xarg2 arg3 xarg3 arg4 xarg4 arg5 xarg5
.Fn SDT_PROBE_DEFINE7_XLATE prov mod func name arg0 xarg0 arg1 xarg1 \
arg2 xarg2 arg3 xarg3 arg4 xarg4 arg5 xarg5 arg6 xarg6
.Fn SDT_PROBE0 prov mod func name
.Fn SDT_PROBE1 prov mod func name arg0
.Fn SDT_PROBE2 prov mod func name arg0 arg1
.Fn SDT_PROBE3 prov mod func name arg0 arg1 arg2
.Fn SDT_PROBE4 prov mod func name arg0 arg1 arg2 arg3
.Fn SDT_PROBE5 prov mod func name arg0 arg1 arg2 arg3 arg4
.Fn SDT_PROBE6 prov mod func name arg0 arg1 arg2 arg3 arg4 arg5
.Fn SDT_PROBE7 prov mod func name arg0 arg1 arg2 arg3 arg4 arg5 arg6
.Sh DESCRIPTION
The
.Nm
macros allow programmers to define static trace points in kernel code.
These trace points are used by the
.Nm
framework to create DTrace probes, allowing the code to be instrumented
using
.Xr dtrace 1 .
By default,
.Nm
trace points are disabled and have no effect on the surrounding code.
When a DTrace probe corresponding to a given trace point is enabled, threads
that execute the trace point will call a handler and cause the probe to fire.
Moreover, trace points can take arguments, making it possible to pass data
to the DTrace framework when an enabled probe fires.
.Pp
Multiple trace points may correspond to a single DTrace probe, allowing
programmers to create DTrace probes that correspond to logical system events
rather than tying probes to specific code execution paths.
For instance, a DTrace probe corresponding to the arrival of an IP packet into
the network stack may be defined using two
.Nm
trace points: one for IPv4 packets and one for IPv6 packets.
.Pp
In addition to defining DTrace probes, the
.Nm
macros allow programmers to define new DTrace providers, making it possible to
namespace logically-related probes.
An example is FreeBSD's sctp provider, which contains
.Nm
probes for FreeBSD's
.Xr sctp 4
implementation.
.Pp
The
.Fn SDT_PROVIDER_DECLARE
and
.Fn SDT_PROVIDER_DEFINE
macros are used respectively to declare and define a DTrace provider named
.Ar prov
with the
.Nm
framework.
A provider need only be defined once; however, the provider must be declared
before defining any
.Nm
probes belonging to that provider.
.Pp
Similarly, the
.Fn SDT_PROBE_DECLARE
and
.Fn SDT_PROBE_DEFINE*
macros are used to declare and define DTrace probes using the
.Nm
framework.
Once a probe has been defined, trace points for that probe may be added to
kernel code.
DTrace probe identifiers consist of a provider, module, function and name, all
of which may be specified in the
.Nm
probe definition.
Note that probes should not specify a module name: the module name of a probe is
used to determine whether or not it should be destroyed when a kernel module is
unloaded.
See the
.Sx BUGS
section.
Note in particular that probes must not be defined across multiple kernel
modules.
.Pp
If
.Ql -
character (dash) is wanted in a probe name,
then it should be represented as
.Ql __
(double underscore) in the probe
.Ar name
parameter passed to various
.Fn SDT_*
macros,
because of technical reasons
(a dash is not valid in C identifiers).
.Pp
The
.Fn SDT_PROBE_DEFINE*
macros also allow programmers to declare the types of the arguments that are
passed to probes.
This is optional; if the argument types are omitted (through use of the
.Fn SDT_PROBE_DEFINE
macro), users wishing to make use of the arguments will have to manually cast
them to the correct types in their D scripts.
It is strongly recommended that probe definitions include a declaration of their
argument types.
.Pp
The
.Fn SDT_PROBE_DEFINE*_XLATE
macros are used for probes whose argument types are to be dynamically translated
to the types specified by the corresponding
.Ar xarg
arguments.
This is mainly useful when porting probe definitions from other operating
systems.
As seen by
.Xr dtrace 1 ,
the arguments of a probe defined using these macros will have types which match
the
.Ar xarg
types in the probe definition.
However, the arguments passed in at the trace point will have types matching the
native argument types in the probe definition, and thus the native type is
dynamically translated to the translated type.
So long as an appropriate translator is defined in
.Pa /usr/lib/dtrace ,
scripts making use of the probe need not concern themselves with the underlying
type of a given
.Nm
probe argument.
.Pp
The
.Fn SDT_PROBE*
macros are used to create
.Nm
trace points.
They are meant to be added to executable code and can be used to instrument the
code in which they are called.
.Sh PROVIDERS
A number of kernel DTrace providers are available.
In general, these providers define stable interfaces and should be treated as
such: existing D scripts may be broken if a probe is renamed or its arguments
are modified.
However, it is often useful to define ad-hoc
.Nm
probes for debugging a subsystem or driver.
Similarly, a developer may wish to provide a group of
.Nm
probes without committing to their future stability.
Such probes should be added to the
.Ql sdt
provider instead of defining a new provider.
.Sh EXAMPLES
The DTrace providers available on the current system can be listed with
.Bd -literal -offset indent
dtrace -l | sed 1d | awk '{print $2}' | sort -u
.Ed
.Pp
A detailed list of the probes offered by a given provider can be obtained by
specifying the provider using the
.Fl P
flag.
For example, to view the probes and argument types for the
.Ql sched
provider, run
.Bd -literal -offset indent
dtrace -lv -P sched
.Ed
.Pp
The following probe definition will create a DTrace probe called
.Ql icmp:::receive-unreachable ,
which would hypothetically be triggered when the kernel receives an ICMP packet
of type Destination Unreachable:
.Bd -literal -offset indent
SDT_PROVIDER_DECLARE(icmp);
SDT_PROBE_DEFINE1(icmp, , , receive__unreachable,
"struct icmp *");
.Ed
This particular probe would take a single argument: a pointer to the struct
containing the ICMP header for the packet.
Note that the module name of this probe is not specified.
.Pp
Consider a DTrace probe which fires when the network stack receives an IP
packet.
Such a probe would be defined by multiple tracepoints:
.Bd -literal -offset indent
SDT_PROBE_DEFINE3(ip, , , receive, "struct ifnet *",
"struct ip *", "struct ip6_hdr *");
int
ip_input(struct mbuf *m)
{
struct ip *ip;
...
ip = mtod(m, struct ip *);
SDT_PROBE3(ip, , , receive, m->m_pkthdr.rcvif, ip, NULL);
...
}
int
ip6_input(struct mbuf *m)
{
struct ip6_hdr *ip6;
...
ip6 = mtod(m, struct ip6_hdr *);
SDT_PROBE3(ip, , , receive, m->m_pkthdr.rcvif, NULL, ip6);
...
}
.Ed
In particular, the probe should fire when the kernel receives either an IPv4
packet or an IPv6 packet.
.Pp
Consider the ICMP probe discussed above.
We note that its second argument is of type
.Ar struct icmp ,
which is a type defined in the FreeBSD kernel to represent the ICMP header of
an ICMP packet, defined in RFC 792.
Linux has a corresponding type,
.Ar struct icmphdr ,
for the same purpose, but its field names differ from FreeBSD's
.Ar struct icmp .
Similarly, illumos defines the
.Ar icmph_t
type, again with different field names.
Even with the
.Ql icmp:::pkt-receive
probes defined in all three operating systems,
one would still have to write OS-specific scripts to extract a given field out
of the ICMP header argument.
Dynamically-translated types solve this problem: one can define an
OS-independent
.Xr c 7
struct to represent an ICMP header, say
.Ar struct icmp_hdr_dt ,
and define translators from each of the three OS-specific types to
.Ar struct icmp_hdr_dt ,
all in the
.Xr dtrace 1
library path.
Then the FreeBSD probe above can be defined with:
.Bd -literal -offset indent
SDT_PROBE_DEFINE1_XLATE(ip, , , receive, "struct icmp *",
"struct icmp_hdr_dt *");
.Ed
.Sh SEE ALSO
.Xr dtrace 1 ,
.Xr dtrace_io 4 ,
.Xr dtrace_ip 4 ,
.Xr dtrace_proc 4 ,
.Xr dtrace_sched 4 ,
.Xr dtrace_tcp 4 ,
.Xr dtrace_udp 4
.Sh AUTHORS
.An -nosplit
DTrace and the
.Nm
framework were originally ported to FreeBSD from Solaris by
.An John Birrell Aq Mt jb@FreeBSD.org .
This manual page was written by
.An Mark Johnston Aq Mt markj@FreeBSD.org .
.Sh BUGS
The
.Nm
macros allow the module and function names of a probe to be specified as part of
a probe definition.
The DTrace framework uses the module name of probes to determine which probes
should be destroyed when a kernel module is unloaded, so the module
name of a probe should match the name of the module in which its defined.
.Nm
will set the module name properly if it is left unspecified in the probe
definition; see the
.Sx EXAMPLES
section.
.Pp
One of the goals of the original
.Nm
implementation (and by extension, of FreeBSD's port) is that inactive
.Nm
probes should have no performance impact.
This is unfortunately not the case;
.Nm
trace points will add a small but non-zero amount of latency to the code
in which they are defined.
A more sophisticated implementation of the probes will help alleviate this
problem.