Wed, 07 Nov 2007 10:28:11 -0800
changeset 7665 eaa2b0789a8e20736a73fd10df2cc38ce61cce36
parent 1 9b2a99adc05e53cd4010de512f50118594756650
child 8794 60903a5c45c4ff470d2442fdcc08d40719e3011b
permissions -rw-r--r--
Bug 401741, bump nightly version to 3.0b2pre, r=ccooper, a=beltzner

<head><title>The Jprof Profiler</title></head>

<body bgcolor="#FFFFFF" text="#000000"
      link="#0000EE" vlink="#551A8B" alink="#FF0000">
<h1>The Jprof Profiler</h1>
<font size="-1">
<a href=""></a>

<a href="#introduction">Introduction</a> | <a href="#operation">Operation</a> |
<a href="#setup">Setup</a> | <a href="#usage">Usage</a> |
<a href="#interpretation">Interpretation</a>


<h3><a name="introduction">Introduction</a></h3>

Jprof is a profiling tool.  I am writing it because I need to find out
where mozilla is spending its time, and there do not seem to be any
profilers for Linux that can handle threads and/or shared libraries.
This code is based heavily on Kipp Hickman's leaky.

<h3><a name="operation">Operation</a></h3>

Jprof operates by installing a timer which periodically interrupts mozilla.
When this timer goes off, the jprof code inside mozilla walks the function call
stack to determine which code was executing and saves the results into the
<code>jprof-log</code> and <code>jprof-map</code> files.  By collecting a large
number of these call stacks, it is possible to deduce where mozilla is spending
its time.

<h3><a name="setup">Setup</a></h3>

First, check out the jprof source code since it is not a part of the
default pull.  To do this do:
  cvs co mozilla/tools/jprof

<p>Next, configure your mozilla with jprof support by adding
<code>--enable-jprof</code> to your configure options (eg adding
<code>ac_add_options --enable-jprof</code> to your <code>.mozconfig</code>) and
making sure that you do <strong>not</strong> have the
<code>--enable-strip</code> configure option set -- jprof needs symbols to

<p>Finally, build mozilla with your new configuration.  Now you can run jprof.</p>

<h3><a name="usage">Usage</a></h3>

The behavior of jprof is determined by the value of the JPROF_FLAGS environment
variable.  This environment variable can be composed of several substrings
which have the following meanings:
    <li> <b>JP_START</b> : Install the signal handler, and start sending the
    timer signals.
    <li> <b>JP_DEFER</b> : Install the signal handler, but don't start sending
    the timer signals.  The user must start the signals by sending the first
    one (with <code>kill -PROF</code>, or with <code>kill -ALRM</code> if
    JP_REALTIME is used, or with <code>kill -POLL</code> (also known as <code>kill -IO</code>) if JP_RTC_HZ is used).

    <li> <b>JP_FIRST=x</b> : Wait x seconds before starting the timer

    <li> <b>JP_PERIOD=y</b> : Set timer to interrupt every y seconds.  Only
    values of y strictly greater than 0.001 are supported.
    <li> <b>JP_REALTIME</b> : Do the profiling in intervals of real time rather
    than intervals of time used by the mozilla process (and the kernel
    when doing work for mozilla).  This could probably lead to weird
    results (you'll see whatever runs when mozilla is waiting for events),
    but is needed to see time spent in the X server.

    <li> <b>JP_RTC_HZ=freq</b> : This option, only available on Linux if the
    kernel is built with RTC support, makes jprof use the RTC timer instead of
    using its own timer.  This option, like JP_REALTIME, uses intervals of real
    time.  This option overrides JP_PERIOD.  <code>freq</code> is the frequency
    at which the timer should fire, measured in Hz. It must be a power of 2.
    The maximal frequency allowed by the kernel can be changed by writing to
    <code>/proc/sys/dev/rtc/max-user-freq</code>; the maximum value it can be
    set to is 8192.  Note that <code>/dev/rtc</code> will need to be readable
    by the Firefox process; making that file world-readable is a simple way to
    accomplish that.

<h4>Examples of JPROF_FLAGS usage</h4>

  <li>To make the timer start firing 3 seconds after the program is started and
  fire every 25 milliseconds of program time use:
        setenv JPROF_FLAGS "JP_START JP_FIRST=3 JP_PERIOD=0.025" </pre>

  <li>To make the timer start on your signal and fire every 1.5 milliseconds of
  program time use:  
        setenv JPROF_FLAGS "JP_DEFER JP_PERIOD=0.0015" </pre>

  <li>To make the timer start on your signal and fire every 10 milliseconds of
  wall-clock time use:  
        setenv JPROF_FLAGS "JP_DEFER JP_PERIOD=0.010 JP_REALTIME" </pre>

  <li>To make the timer start on your signal and fire at 8192 Hz in wall-clock
  time use:
        setenv JPROF_FLAGS "JP_DEFER JP_RTC_HZ=8192" </pre>

<h4>Pausing profiles</h4>

<P>jprof can be paused at any time by sending a SIGUSR1 to mozilla (<code>kill
-USR1</code>).  This will cause the timer signals to stop and jprof-map to be
written, but it will not close jprof-log.  Combining SIGUSR1 with the JP_DEFER
option allows profiling of one sequence of actions by starting the timer right
before starting the actions and stopping the timer right afterward.

<P>After a SIGUSR1, sending another timer signal (SIGPROF, SIGALRM, or SIGPOLL (aka SIGIO),
depending on the mode) can be used to continue writing data to the same output.

<h4>Looking at the results</h4>

Now that we have <code>jprof-log</code> and <code>jprof-map</code> files, we
can use the jprof executable is used to turn them into readable output.  To do
this jprof needs the name of the mozilla binary and the log file.  It deduces
the name of the map file:

  ./jprof /home/user/mozilla/debug/dist/bin/mozilla-bin ./jprof-log > tmp.html

This will generate the file <code>tmp.html</code> which you should view in a
web browser.

<h3><a name="interpretation">Interpretation</a></h3>

The Jprof output is split into a flat portion and a hierarchical portion.
There are links to each section at the top of the page.  It is typically
easier to analyze the profile by starting with the flat output and following
the links contained in the flat output up to the hierarchical output.

<h4><a name="flat">Flat output</a></h3>

The flat portion of the profile indicates which functions were executing
when the timer was going off.  It is displayed as a list of functions names
on the right and the number of times that function was interrupted on the
left.  The list is sorted by decreasing interrupt count.  For example:

<blockquote> <pre>
Total hit count: 151603
Count %Total  Function Name

<a href="#23081">8806   5.8     __libc_poll</a>
<a href="#40008">2254   1.5     __i686.get_pc_thunk.bx</a>
<a href="#21390">2053   1.4     _int_malloc</a>
<a href="#49013">1777   1.2     nsStyleContext::GetStyleData(nsStyleStructID)</a>
<a href="#21380">1600   1.1     __libc_malloc</a>
<a href="#603">1552   1.0     nsCOMPtr_base::~nsCOMPtr_base()</a>
</pre> </blockquote>

This shows that of the 151603 times the timer fired, 1777 (1.2% of the total) were inside nsStyleContext::GetStyleData() and 1552 (1.0% of the total) were in the nsCOMPtr_base destructor.

In general, the functions with the highest count are the functions which
are taking the most time.

The function names are linked to the entry for that function in the
hierarchical profile, which is described in the next section.

<h4><a name="hier">Hierarchical output</a></h4>

The hierarchical output is divided up into sections, with each section
corresponding to one function.  A typical section looks something like

             <A href="#29355">141300 PL_ProcessPendingEvents</A>
             <A href="#29372">   927 PL_ProcessEventsBeforeID</A>
 29358   0 <a name=29358>  142227</a> <b>PL_HandleEvent</b>
             <A href="#28546"> 92394 nsInputStreamReadyEvent::EventHandler(PLEvent*)</A>
             <A href="#41572"> 49181 HandlePLEvent(ReflowEvent*)</A>
             <A href="#29537">   481 handleTimerEvent(TimerEventType*)</A>
             <A href="#34494">   158 nsTransportStatusEvent::HandleEvent(PLEvent*)</A>
             <A href="#29359">     9 PL_DestroyEvent</A>

             <A href="#20319">     4 __restore_rt</A>

The information this block tells us is:

<li>There were 0 profiler hits <em>in</em> <code>PL_HandleEvent</code>
<li>There were 142227 profiler hits <em>under</em> <code>PL_HandleEvent</code>.  Of these:
  <li>92394 were in or under <code>nsInputStreamReadyEvent::EventHandler</code>
  <li>49181 were in or under <code>HandlePLEvent(ReflowEvent*)</code>
  <li>481 were in or under <code>handleTimerEvent</code>
  <li>158 were in or under <code>nsTransportStatusEvent::HandleEvent</code>
  <li>9 were in or under <code>PL_DestroyEvent</code>
  <li>4 were in or under <code>__restore_rt</code>
<li>Of these 142227 calls into <code>PL_HandleEvent</code>:
  <li>141300 came from <code>PL_ProcessPendingEvents</code>
  <li>927 came from <code>PL_ProcessEventsBeforeID</code>

The rest of this section explains how to read this information off from the jprof output.

<p>This block corresponds to the function <code>PL_HandleEvent</code>, which is
therefore bolded and not a link.  The name of this function is preceded by
three numbers which have the following meaning.  The number on the left (29358)
is the index number, and is not important.  The center number (0) is the number
of times this function was interrupted by the timer.  The last number (142227)
is the number of times this function was in the call stack when the timer went
off.  That is, the timer went off while we were in code that was ultimately
called from <code>PL_HandleEvent</code>.
<p>For our example we can see that our function was in the call stack for
142227 interrupt ticks, but we were never the function that was running when
the interrupt arrived.
The functions listed above the line for <code>PL_HandleEvent</code> are its
callers.  The numbers to the left of these function names are the numbers of
times these functions were in the call stack as callers of
<code>PL_HandleEvent</code>.  In our example, we were called 927 times by
<code>PL_ProcessEventsBeforeID</code> and 141300 times by
The functions listed below the line for <code>PL_HandleEvent</code> are its
callees.  The numbers to the left of the function names are the numbers of
times these functions were in the callstack as callees of <code>PL_HandleEvent</code>. In our example, of the 142227 profiler hits under <code>PL_HandleEvent</code> 92394 were under <code>nsInputStreamReadyEvent::EventHandler</code>, 49181 were under <code>HandlePLEvent(ReflowEvent*)</code>, and so forth.

Jprof has only been tested under Red Hat Linux 6.0, 6.1, and 6.2.  It does
not work under 6.0, though it is possible hack up the source code and make
it work there.  The way I determine the stack trace from inside the
signal handler is tightly bound to the version of glibc that is running.
If you know of a more portable way to get this information please let
me know.

  <li>Ben Bucksch reports that installing the Red Hat 6.1 glibc rpms on a Red Hat
6.0 system allows jprof to work, and does not seem to break anything except
gdm (the Gnome login program), and that can be fixed by installing the RH 6.1
gdb rpm.</li>
  <li>David Baron reports that jprof works under RedHat 6.0 if one uncomments
the <code>#define JPROF_PTHREAD_HACK</code> near the beginning of