path: root/doc/operf.1.in

.\" Manpage for operf
.\" Author: Breno Leitao <brenohl@br.ibm.com>
.\" Modifications: Maynard Johnson <maynardj@us.ibm.com>
.TH OPERF 1 "@DATE@" "oprofile @VERSION@"
.SH NAME
operf \- Performance profiler tool for Linux

.SH SYNOPSIS
.B operf
[
.I options
]
[ --system-wide | --pid <pid> | [ command [ args ] ] ]

.SH DESCRIPTION
Operf is the profiler tool provided with OProfile. Operf
uses the Linux Performance Events Subsystem and, thus, does not require the
obsolete oprofile kernel driver.
.P
By default, operf uses <current_dir>/oprofile_data as the session-dir and stores profiling data there.
You can change this by way of the
.I --session-dir
option. The usual post-profiling analysis tools such as
.BI opreport(1)
and
.BI opannotate(1)
can be used to generate profile reports. Unless a
.I session-dir
is specified, the post-processing analysis tools will search for samples in
<current_dir>/oprofile_data first. If that directory does not exist, the
post-processing tools use the standard session-dir of /var/lib/oprofile.
.P
Statistics, such as total samples received
and lost samples, are written to the operf.log file that can be found in the
<session_dir>/samples directory.
.br

.SH RUN MODES
One (and only one) of the following
.SB run modes
must be specified:
.TP
.BI command [args]
The command or application to be profiled.
.I args
are the input arguments that the command or application requires.
.br
.TP
.BI "--pid / -p " PID
.RS
This option enables operf to profile a running application.
.I PID
should be the process ID of the process you wish to profile.  When
finished profiling (e.g., when the profiled process ends), press
Ctrl-c to stop operf. If you run
.BI operf
.BI --pid
as a background job (i.e., with the &), you
.B must
stop it in a controlled manner in order for it to process the profile
data it has collected.  Use
.BI kill
.BI -SIGINT
.BI <operf-PID>
for this purpose.
.P
.B Limitation:
When using this option to profile a multi-threaded application that also forks
new processes, be aware that samples for processes that are forked before profiling
is started may not be recorded (depending on timing of thread creation and when
operf is started).
.RE
.TP
.BI "--system-wide / -s"
This option is for performing a system-wide profile.  You must
have root authority to run operf in this mode.  When finished profiling,
Ctrl-c to stop operf. If you run
.BI operf
.BI --system-wide
as a background job (i.e., with the &), you
.B must
stop it in a controlled manner in order for it to process the profile
data it has collected.  Use
.BI kill
.BI -SIGINT
.BI <operf-PID>
for this purpose.
It is recommended
that when running operf with this option, the user's current working
directory should be /root or a subdirectory of /root to avoid
storing sample data files in locations accessible by regular users.
.br
.SH OTHER OPTIONS
.TP
.BI "--vmlinux / -k " vmlinux_path
.RS
A vmlinux file that matches the running kernel that has symbol and/or debuginfo.
Kernel samples will be attributed to this binary, allowing post-processing tools
(like opreport) to attribute samples to the appropriate kernel symbols.
.P
The kernel symbol information may be obtained from /proc/kallsyms if
the user does not specify a vmlinux file.  The symbol addresses are given
in /proc/kallsyms if permitted by the setting of /proc/sys/kernel/kptr_restrict.
.P
If the
.I --vmlinux
option is not used and kernel symbols cannot be obtained from /proc/kallsyms,
then all kernel samples are attributed to "no-vmlinux", which is simply
a bucket to hold the samples and not an actual file.
.RE
.TP
.BI "--events / -e " event1[,event2[,...]]
This option is for passing a comma-separated list of event specifications
for profiling. Each event spec is of the form:
.br
.I "   name:count[:unitmask[:kernel[:user]]]"
.P
.RS
The
.I count
value is used to control the sampling rate for profiling; it is the number
of events to occur between samples. The rate is lowered by specifying a higher
.I count
value \(em i.e., a higher number of events to occur between samples.
.P
You can specify
.I unitmask
values using either a numerical value (hex values
.I must
begin with "0x") or a symbolic name (if the
.I name=<um_name>
field is shown in the
.B ophelp
output). For some named unit masks, the hex value is not unique; thus, OProfile
tools enforce specifying such unit masks value by name.
If no unit mask is specified, the default unit mask value for the event is used.
.P
The
.I kernel
and
.I user
parts of the event specification are binary values ('1' or '0') indicating
whether or not to collect samples for kernel space and user space.
.br
.B Note:
In order to specify the
.I kernel/user
bits, you must also specify a
.I unitmask
value, even if the processor type (or the specified event) does not use unit masks \(em
in which case, use the value '0' to signify a null unit mask; for example:
.br
   -e INST_RETIRED_ANY_P:100000:0:1:0
.br
                         ^      ^ ^ ^
                         |      | | |--- '0': do not record user space samples
                         |      | |-- '1': record kernel space samples
                         |      |-- '0': the null unit mask
                         |--count value
.P
Event names for some IBM PowerPC systems include a
.I _GRP<n>
(group number) suffix. You can pass either the full event name or the base event name
(i.e., without the suffix) to
.B operf.
If the base event name is passed,
.B operf
will automatically choose an appropriate group number suffix
for the event; thus, OProfile post-processing tools will always show real event
names that include the group number suffix.
.P
When no event specification is given, the default event for the running
processor type will be used for profiling.
Use
.BI ophelp
to list the available events for your processor type.
.RE
.br
.TP
.BI "--callgraph / -g"
This option enables the callgraph to be saved during profiling. NOTE: The
full callchain is recorded, so there is no depth limit.
.br
.TP
.BI "--separate-thread / -t"
This option categorizes samples by thread group ID (tgid) and thread ID (tid).
The '--separate-thread' option is useful for seeing per-thread samples in
multi-threaded applications.  When used in conjunction with the '--system-wide'
option, the '--separate-thread' option is also useful for seeing per-process
(i.e., per-thread group) samples for the case where multiple processes are
executing the same program during a profiling run.
.br
.TP
.BI "--separate-cpu / -c"
This option categorizes samples by cpu.
.br
.TP
.BI "--session-dir / -d " path
This option specifies the session path to hold the sample data. If not specified,
the data is saved in the
.I oprofile_data
directory on the current path.
.br
.TP
.BI "--lazy-conversion / -l"
.RS
Use this option to reduce the overhead of
.BI operf
during profiling. Normally, profile data received from the kernel is converted
to OProfile format during profiling time. This is typically not an issue when
profiling a single application. But when using the
.I --system-wide
option, this on-the-fly conversion process can cause noticeable overhead,
particularly on busy multi-processor systems. The
.I --lazy-conversion
option directs
.BI operf
to wait until profiling is completed to do the conversion of profile data.
.P
.B Note:
This option is
.B not
recommended to be used in conjunction with the
.I --pid
option for profiling multi-threaded processes. Depending
on the order of thread creation (or forking of new processes),
you may not get any samples for the new threads/processes.
.RE
.TP
.BI "--append / -a"
By default,
.I operf
moves old profile data from <session_dir>/samples/current to
<session_dir>/samples/previous.  If a 'previous' profile already existed,
it will be replaced.  If the
.I --append
option is passed, old profile data is left in place and new profile data will
be added to it, and the 'previous' profile (if one existed) will remain untouched.
To access the 'previous' profile, simply add a session specification to the normal
invocation of oprofile post-processing tools.  For example:
.br
.BI "   opreport"
.BI session:previous
.br
.TP
.BI "--verbose / -V " level
A comma-separated list of debugging control values, used to increase the verbosity of the output.
Valid values are:  debug, record, convert, misc, sfile, arcs, or the special value, 'all'.
.br
.TP
.BI "--version / -v"
Show operf version.
.br
.TP
.BI "--help / -h"
Display brief usage message.
.br
.TP
.BI "--usage / -u"
Display brief usage message.
.br

.SH EXAMPLE
$ operf make

.SH VERSION
This man page is current for @PACKAGE@-@VERSION@.

.SH SEE ALSO
opreport(1), opannotate(1).