ktr.9 revision 143270
Copyright (c) 2001 John H. Baldwin <jhb@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: head/share/man/man9/ktr.9 143270 2005-03-08 01:37:36Z hmp $

.Dd February 15, 2001 .Dt KTR 9 .Os .Sh NAME .Nm CTR0 , CTR1 , CTR2 , CTR3 , CTR4 , CTR5 .Nd kernel tracing facility .Sh SYNOPSIS n sys/param.h n sys/ktr.h .Vt "extern int ktr_cpumask" ; .Vt "extern int ktr_entries" ; .Vt "extern int ktr_extend" ; .Vt "extern int ktr_mask" ; .Vt "extern int ktr_verbose" ; .Vt "extern struct ktr_entry ktr_buf[]" ; .Ft void .Fn CTR0 "u_int mask" "char *format" .Ft void .Fn CTR1 "u_int mask" "char *format" "arg1" .Ft void .Fn CTR2 "u_int mask" "char *format" "arg1" "arg2" .Ft void .Fn CTR3 "u_int mask" "char *format" "arg1" "arg2" "arg3" .Ft void .Fn CTR4 "u_int mask" "char *format" "arg1" "arg2" "arg3" "arg4" .Ft void .Fn CTR5 "u_int mask" "char *format" "arg1" "arg2" "arg3" "arg4" "arg5" .Sh DESCRIPTION KTR provides a circular buffer of events that can be logged in a .Xr printf 9 style fashion. These events can then be dumped with .Xr ddb 4 , .Xr gdb 1 or .Xr ktrdump 8 .

p Events are created and logged in the kernel via the .Dv CTR Ns Ar x macros. The first parameter is a mask of event types

q Dv KTR_* defined in n sys/ktr.h . The event will be logged only if any of the event types specified in .Fa mask are enabled in the global event mask stored in .Va ktr_mask . The .Fa format argument is a .Xr printf 9 style format string used to build the text of the event log message. Following the .Fa format string are zero to five arguments referenced by .Fa format . Note that the different macros differ only in the number of arguments each one takes, as indicated by its name. Each event is logged with a timestamp in addition to the log message.

p The .Va ktr_entries variable contains the number of entries in the .Va ktr_buf array. These variables are mostly useful for post-mortem crash dump tools to locate the base of the circular trace buffer and its length.

p The .Va ktr_mask variable contains the run time mask of events to log.

p The CPU event mask is stored in the .Va ktr_cpumask variable.

p The .Va ktr_verbose variable stores the verbose flag that controls whether events are logged to the console in addition to the event buffer. .Sh EXAMPLES This example demonstrates the use of tracepoints at the .Dv KTR_PROC logging level. d -literal -compact void mi_switch() { ... /* * Pick a new current process and record its start time. */ ... CTR3(KTR_PROC, "mi_switch: old proc %p (pid %d, %s)", p, p->p_pid, p->p_comm); ... cpu_switch(); ... CTR3(KTR_PROC, "mi_switch: new proc %p (pid %d, %s)", p, p->p_pid, p->p_comm); ... } .Ed .Sh SEE ALSO .Xr ktr 4 , .Xr ktrdump 8 .Sh HISTORY The KTR kernel tracing facility first appeared in sx 3.0 and was imported into .Fx 5.0 . .Sh BUGS Currently there is one global buffer shared among all CPUs. It might be profitable at some point in time to use per-CPU buffers instead so that if one CPU halts or starts spinning, then the log messages it emitted just prior to halting or spinning will not be drowned out by events from the other CPUs.