.\" Automatically generated by Pandoc 2.9.2.1
.\"
.TH "UFTRACE-REPLAY" "1" "Sep, 2018" "Uftrace User Manuals" ""
.hy
.SH NAME
.PP
uftrace-replay - Print recorded function trace
.SH SYNOPSIS
.PP
uftrace replay [\f[I]options\f[R]]
.SH DESCRIPTION
.PP
This command prints trace data recorded using the
\f[C]uftrace-record\f[R](1) command.
The traced functions are printed like a C program in time order.
.SH REPLAY OPTIONS
.TP
-f \f[I]FIELD\f[R], --output-fields=\f[I]FIELD\f[R]
Customize field in the output.
Possible values are: duration, tid, addr, time, delta, elapsed, task and
module.
Multiple fields can be set by using comma.
Special field of `none' can be used (solely) to hide all fields.
Default is `duration,tid'.
See \f[I]FIELDS\f[R].
.TP
--flat
Print flat format rather than C-like format.
This is usually for debugging and testing purpose.
.TP
--column-view
Show each task in separate column.
This makes easy to distinguish functions in different tasks.
.TP
--column-offset=\f[I]DEPTH\f[R]
When \f[C]--column-view\f[R] option is used, this option specifies the
amount of offset between each task.
Default is 8.
.TP
--task-newline
Interleave a new line when task is changed.
This makes easy to distinguish functions in different tasks.
.TP
--no-comment
Do not show comments of returned functions.
.TP
--libname
Show libname name along with function name.
.TP
--format=\f[I]TYPE\f[R]
Show format style output.
Currently, normal and html styles are supported.
.SH COMMON OPTIONS
.TP
-F \f[I]FUNC\f[R], --filter=\f[I]FUNC\f[R]
Set filter to trace selected functions and their children functions.
This option can be used more than once.
See \f[I]FILTERS\f[R].
.TP
-N \f[I]FUNC\f[R], --notrace=\f[I]FUNC\f[R]
Set filter not to trace selected functions and their children functions.
This option can be used more than once.
See \f[I]FILTERS\f[R].
.TP
-C \f[I]FUNC\f[R], --caller-filter=\f[I]FUNC\f[R]
Set filter to trace callers of selected functions only.
This option can be used more than once.
See \f[I]FILTERS\f[R].
.TP
-T \f[I]TRG\f[R], --trigger=\f[I]TRG\f[R]
Set trigger on selected functions.
This option can be used more than once.
See \f[I]TRIGGERS\f[R].
.TP
-D \f[I]DEPTH\f[R], --depth \f[I]DEPTH\f[R]
Set trace limit in nesting level.
See \f[I]FILTERS\f[R].
.TP
-t \f[I]TIME\f[R], --time-filter=\f[I]TIME\f[R]
Do not show functions which run under the time threshold.
If some functions explicitly have the `trace' trigger applied, those are
always traced regardless of execution time.
See \f[I]FILTERS\f[R].
.TP
--no-libcall
Do not show library calls.
.TP
--no-event
Do not show any events.
Implies \f[C]--no-sched\f[R].
.TP
--no-sched
Do not show schedule events.
.TP
--match=\f[I]TYPE\f[R]
Use pattern match using TYPE.
Possible types are \f[C]regex\f[R] and \f[C]glob\f[R].
Default is \f[C]regex\f[R].
.TP
--disable
Start replay with tracing disabled.
This is only meaningful when used with a \f[C]trace_on\f[R] trigger.
.SH COMMON ANALYSIS OPTIONS
.TP
-H \f[I]FUNC\f[R], --hide=\f[I]FUNC\f[R]
Set filter not to trace selected functions.
It doesn\[cq]t affect their subtrees, but hides only the given
functions.
This option can be used more than once.
See \f[I]FILTERS\f[R].
.TP
--kernel-full
Show all kernel functions and events occurred outside of user functions.
.TP
--kernel-only
Show kernel functions only without user functions.
.TP
--event-full
Show all (user) events outside of user functions.
.TP
--tid=\f[I]TID\f[R][,\f[I]TID\f[R],\&...]
Only print functions called by the given tasks.
To see the list of tasks in the data file, you can use
\f[C]uftrace report --task\f[R] or \f[C]uftrace info\f[R].
This option can also be used more than once.
.TP
--demangle=\f[I]TYPE\f[R]
Use demangled C++ symbol names for filters, triggers, arguments and/or
return values.
Possible values are \[lq]full\[rq], \[lq]simple\[rq] and \[lq]no\[rq].
Default is \[lq]simple\[rq] which ignores function arguments and
template parameters.
.TP
-r \f[I]RANGE\f[R], --time-range=\f[I]RANGE\f[R]
Only show functions executed within the time RANGE.
The RANGE can be <start>\[ti]<stop> (separated by \[lq]\[ti]\[rq]) and
one of <start> and <stop> can be omitted.
The <start> and <stop> are timestamp or elapsed time if they have
<time_unit> postfix, for example `100us'.
The timestamp or elapsed time can be shown with \f[C]-f time\f[R] or
\f[C]-f elapsed\f[R] option respectively.
See \f[I]FILTERS\f[R].
.SH FILTERS
.PP
The uftrace tool supports filtering out uninteresting functions.
When uftrace is called it receives two types of function filter; an
opt-in filter with \f[C]-F\f[R]/\f[C]--filter\f[R] and an opt-out filter
with \f[C]-N\f[R]/\f[C]--notrace\f[R].
These filters can be applied either at record time or replay time.
.PP
The first one is an opt-in filter.
By default, it doesn\[cq]t show anything.
But when one of the specified functions is met, printing is started.
When the function returns, printing is stopped again.
.PP
For example, consider a simple program which calls \f[C]a()\f[R],
\f[C]b()\f[R] and \f[C]c()\f[R] in turn.
.IP
.nf
\f[C]
$ cat abc.c
void c(void) {
    /* do nothing */
}

void b(void) {
    c();
}

void a(void) {
    b();
}

int main(void) {
    a();
    return 0;
}

$ gcc -pg -o abc abc.c
\f[R]
.fi
.PP
Normally uftrace replay will show all the functions from
\f[C]main()\f[R] to \f[C]c()\f[R].
.IP
.nf
\f[C]
$ uftrace ./abc
# DURATION    TID     FUNCTION
 138.494 us [ 1234] | __cxa_atexit();
            [ 1234] | main() {
            [ 1234] |   a() {
            [ 1234] |     b() {
   3.880 us [ 1234] |       c();
   5.475 us [ 1234] |     } /* b */
   6.448 us [ 1234] |   } /* a */
   8.631 us [ 1234] | } /* main */
\f[R]
.fi
.PP
But when the \f[C]-F b\f[R] filter option is used, it will not show
\f[C]main()\f[R] or \f[C]a()\f[R] but only \f[C]b()\f[R] and
\f[C]c()\f[R].
Note that the filter was set on \f[C]uftrace replay\f[R], not at record
time.
.IP
.nf
\f[C]
$ uftrace record ./abc
$ uftrace replay -F b
# DURATION    TID     FUNCTION
            [ 1234] | b() {
   3.880 us [ 1234] |   c();
   5.475 us [ 1234] | } /* b */
\f[R]
.fi
.PP
The second type of filter is opt-out.
When used, everything is shown by default, but printing stops once one
of the specified functions is met.
When the given function returns, printing is started again.
.PP
In the above example, you can omit the function \f[C]b()\f[R] and all
calls it makes with the \f[C]-N\f[R] option.
.IP
.nf
\f[C]
$ uftrace record ./abc
$ uftrace replay -N b
# DURATION    TID     FUNCTION
 138.494 us [ 1234] | __cxa_atexit();
            [ 1234] | main() {
   6.448 us [ 1234] |   a();
   8.631 us [ 1234] | } /* main */
\f[R]
.fi
.PP
You can hide the function \f[C]b()\f[R] only without affecting the calls
it makes in its subtree functions with \f[C]-H\f[R] option.
.IP
.nf
\f[C]
$ uftrace record ./abc
$ uftrace replay -H b
# DURATION    TID     FUNCTION
 138.494 us [ 1234] | __cxa_atexit();
            [ 1234] | main() {
            [ 1234] |   a() {
   3.880 us [ 1234] |     c();
   6.448 us [ 1234] |   } /* a */
   8.631 us [ 1234] | } /* main */
\f[R]
.fi
.PP
The above \f[C]-H\f[R] option is especially useful when hiding std
namespace functions in C++ programs by using \f[C]-H \[ha]std::\f[R]
option setting.
.PP
If users only care about specific functions and want to know how they
are called, one can use the caller filter.
It makes the function as leaf and prints the parent functions to the
function.
.IP
.nf
\f[C]
$ uftrace record ./abc
$ uftrace replay -C b
# DURATION    TID     FUNCTION
            [ 1234] | main() {
            [ 1234] |   a() {
   5.475 us [ 1234] |     b();
   6.448 us [ 1234] |   } /* a */
   8.631 us [ 1234] | } /* main */
\f[R]
.fi
.PP
In the above example, functions not in the calling path were not shown.
Also the function `c' - which is a child of the function `b' - is also
hidden.
.PP
In addition, you can limit the print nesting level with -D option.
.IP
.nf
\f[C]
$ uftrace record ./abc
$ uftrace replay -D 3
# DURATION    TID     FUNCTION
 138.494 us [ 1234] | __cxa_atexit();
            [ 1234] | main() {
            [ 1234] |   a() {
   5.475 us [ 1234] |     b();
   6.448 us [ 1234] |   } /* a */
   8.631 us [ 1234] | } /* main */
\f[R]
.fi
.PP
In the above example, uftrace only prints functions up to a depth of 3,
so leaf function \f[C]c()\f[R] was omitted.
Note that the \f[C]-D\f[R] option also works with \f[C]-F\f[R].
.PP
Sometimes it\[cq]s useful to see long-running functions only.
This is good because there are usually many tiny functions that are not
interesting.
The \f[C]-t\f[R]/\f[C]--time-filter\f[R] option implements the
time-based filter that only records functions which run longer than the
given threshold.
In the above example, the user might want to see functions running more
than 5 microseconds like below:
.IP
.nf
\f[C]
$ uftrace record ./abc
$ uftrace replay -t 5us
# DURATION    TID     FUNCTION
 138.494 us [ 1234] | __cxa_atexit();
            [ 1234] | main() {
            [ 1234] |   a() {
   5.475 us [ 1234] |     b();
   6.448 us [ 1234] |   } /* a */
   8.631 us [ 1234] | } /* main */
\f[R]
.fi
.PP
You can also see replay output with different time threshold for the
same recorded data.
.IP
.nf
\f[C]
$ uftrace replay -t 6us
# DURATION    TID     FUNCTION
 138.494 us [ 1234] | __cxa_atexit();
            [ 1234] | main() {
   6.448 us [ 1234] |   a();
   8.631 us [ 1234] | } /* main */
\f[R]
.fi
.PP
In addition, The \f[C]-r\f[R] option can show functions executed within
the given time range.
When using this option, you can see TIMESTAMP or ELAPSED fields as well
as DURATION and TID.
.IP
.nf
\f[C]
$ uftrace replay -r 502716.387320101\[ti]502716.387322389
#     TIMESTAMP      DURATION    TID     FUNCTION
502716.387320101   0.289 us [ 6126] |   fgets();
502716.387320584            [ 6126] |   get_values_from() {
502716.387320709   0.245 us [ 6126] |     strdup();
502716.387321172   0.144 us [ 6126] |     strsep();
502716.387321542   0.223 us [ 6126] |     atoi();
502716.387321983   0.239 us [ 6126] |     atoi();
502716.387322389   1.805 us [ 6126] |   } /* get_values_from */

$ uftrace replay -r 40us\[ti] | head -10
#  ELAPSED   DURATION    TID     FUNCTION
  40.141 us            [ 6126] |   get_values_from() {
  40.269 us   0.249 us [ 6126] |     strdup();
  40.756 us   0.149 us [ 6126] |     strsep();
  41.119 us   0.235 us [ 6126] |     atoi();
  41.578 us   0.211 us [ 6126] |     atoi();
  41.957 us   1.816 us [ 6126] |   } /* get_values_from */
  42.124 us   0.220 us [ 6126] |   fgets();
  42.529 us            [ 6126] |   get_values_from() {
  42.645 us   0.236 us [ 6126] |     strdup();
\f[R]
.fi
.PP
You can also set triggers on filtered functions.
See \f[I]TRIGGERS\f[R] section below for details.
.SH TRIGGERS
.PP
The uftrace tool supports triggering actions on selected function calls
with or without filters.
Currently supported triggers are \f[C]depth\f[R], \f[C]backtrace\f[R],
\f[C]trace_on\f[R] and \f[C]trace_off\f[R].
The BNF for trigger specifications is like below:
.IP
.nf
\f[C]
<trigger>    :=  <symbol> \[dq]\[at]\[dq] <actions>
<actions>    :=  <action>  | <action> \[dq],\[dq] <actions>
<action>     :=  \[dq]depth=\[dq]<num> | \[dq]backtrace\[dq] | \[dq]trace_on\[dq] | \[dq]trace_off\[dq] |
                 \[dq]color=\[dq]<color> | \[dq]time=\[dq]<time_spec> | \[dq]filter\[dq] | \[dq]notrace\[dq] | \[dq]hide\[dq]
<time_spec>  :=  <num> [ <time_unit> ]
<time_unit>  :=  \[dq]ns\[dq] | \[dq]nsec\[dq] | \[dq]us\[dq] | \[dq]usec\[dq] | \[dq]ms\[dq] | \[dq]msec\[dq] | \[dq]s\[dq] | \[dq]sec\[dq] | \[dq]m\[dq] | \[dq]min\[dq]
\f[R]
.fi
.PP
The \f[C]depth\f[R] trigger is to change filter depth during execution
of the function.
It can be used to apply different filter depths for different functions.
And the \f[C]backtrace\f[R] trigger is used to print a stack backtrace
at replay time.
.PP
The color trigger is to change the color of the function in replay
output.
The supported colors are \f[C]red\f[R], \f[C]green\f[R], \f[C]blue\f[R],
\f[C]yellow\f[R], \f[C]magenta\f[R], \f[C]cyan\f[R], \f[C]bold\f[R], and
\f[C]gray\f[R].
.PP
The following example shows how triggers work.
We set a filter on function \f[C]b()\f[R] with the \f[C]backtrace\f[R]
action and change the maximum filter depth under \f[C]b()\f[R] to 2.
.IP
.nf
\f[C]
$ uftrace record ./abc
$ uftrace replay -T \[aq]b\[at]filter,backtrace,depth=2\[aq]
# DURATION    TID     FUNCTION
  backtrace [ 1234] | /* [ 0] main */
  backtrace [ 1234] | /* [ 1] a */
            [ 1234] | b() {
   3.880 us [ 1234] |   c();
   5.475 us [ 1234] | } /* b */
\f[R]
.fi
.PP
The \f[C]trace_on\f[R] and \f[C]trace_off\f[R] actions (the \f[C]_\f[R]
can be omitted as \f[C]traceon\f[R] and \f[C]traceoff\f[R]) control
whether uftrace shows functions or not.
The trigger runs at replay time, not run time, so it can handle kernel
functions as well.
Contrast this with triggers used under \f[C]uftrace record\f[R].
.PP
The \f[C]time\f[R] trigger is to change time filter setting during
execution of the function.
It can be used to apply different time filter for different functions.
.PP
The \f[C]filter\f[R] and \f[C]notrace\f[R] triggers have same effect as
\f[C]-F\f[R]/\f[C]--filter\f[R] and \f[C]-N\f[R]/\f[C]--notrace\f[R]
options respectively.
.PP
The \f[C]hide\f[R] trigger has the same effect as
\f[C]-H\f[R]/\f[C]--hide\f[R] option that hides the given functions, but
do not affect to the functions in their subtree unlike the
\f[C]notrace\f[R] trigger.
.SH FIELDS
.PP
The uftrace allows for user to customize the replay output with a couple
of fields.
Here the field means info on the left side of the pipe (|) character.
By default it uses duration and tid fields, but you can use other fields
in any order like:
.IP
.nf
\f[C]
$ uftrace replay -f time,delta,duration,addr
#     TIMESTAMP      TIMEDELTA  DURATION     ADDRESS     FUNCTION
    74469.340757350              1.583 us       4004d0 | __monstartup();
    74469.340762221   4.871 us   0.766 us       4004f0 | __cxa_atexit();
    74469.340764847   2.626 us                  4006b1 | main() {
    74469.340765061   0.214 us                  400656 |   a() {
    74469.340765195   0.134 us                  400669 |     b() {
    74469.340765344   0.149 us                  40067c |       c() {
    74469.340765524   0.180 us   0.742 us       4004b0 |         getpid();
    74469.340766935   1.411 us   1.591 us       40067c |       } /* c */
    74469.340767195   0.260 us   2.000 us       400669 |     } /* b */
    74469.340767372   0.177 us   2.311 us       400656 |   } /* a */
    74469.340767541   0.169 us   2.694 us       4006b1 | } /* main */
\f[R]
.fi
.PP
Each field has following meaning:
.IP \[bu] 2
tid: task id (obtained by gettid(2))
.IP \[bu] 2
duration: function execution time
.IP \[bu] 2
time: timestamp at the execution
.IP \[bu] 2
delta: difference between two timestamp in a task
.IP \[bu] 2
elapsed: elapsed time from the first timestamp
.IP \[bu] 2
addr: address of the function
.IP \[bu] 2
task: task name (comm)
.IP \[bu] 2
module: library or executable name of the function
.PP
The default value is `duration,tid'.
If given field name starts with \[lq]+\[rq], then it\[cq]ll be appended
to the default fields.
So \[lq]-f +time\[rq] is as same as \[lq]-f duration,tid,time\[rq].
And it also accepts a special field name of `none' which disables the
field display and shows function output only.
.SH SEE ALSO
.PP
\f[C]uftrace\f[R](1), \f[C]uftrace-record\f[R](1),
\f[C]uftrace-report\f[R](1), \f[C]uftrace-info\f[R](1)
.SH AUTHORS
Namhyung Kim <namhyung@gmail.com>.