<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN"
"http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head><meta http-equiv="content-type" content="text/html; charset=utf-8" />
<title>[202021] trunk/Source/JavaScriptCore</title>
</head>
<body>
<style type="text/css"><!--
#msg dl.meta { border: 1px #006 solid; background: #369; padding: 6px; color: #fff; }
#msg dl.meta dt { float: left; width: 6em; font-weight: bold; }
#msg dt:after { content:':';}
#msg dl, #msg dt, #msg ul, #msg li, #header, #footer, #logmsg { font-family: verdana,arial,helvetica,sans-serif; font-size: 10pt; }
#msg dl a { font-weight: bold}
#msg dl a:link { color:#fc3; }
#msg dl a:active { color:#ff0; }
#msg dl a:visited { color:#cc6; }
h3 { font-family: verdana,arial,helvetica,sans-serif; font-size: 10pt; font-weight: bold; }
#msg pre { overflow: auto; background: #ffc; border: 1px #fa0 solid; padding: 6px; }
#logmsg { background: #ffc; border: 1px #fa0 solid; padding: 1em 1em 0 1em; }
#logmsg p, #logmsg pre, #logmsg blockquote { margin: 0 0 1em 0; }
#logmsg p, #logmsg li, #logmsg dt, #logmsg dd { line-height: 14pt; }
#logmsg h1, #logmsg h2, #logmsg h3, #logmsg h4, #logmsg h5, #logmsg h6 { margin: .5em 0; }
#logmsg h1:first-child, #logmsg h2:first-child, #logmsg h3:first-child, #logmsg h4:first-child, #logmsg h5:first-child, #logmsg h6:first-child { margin-top: 0; }
#logmsg ul, #logmsg ol { padding: 0; list-style-position: inside; margin: 0 0 0 1em; }
#logmsg ul { text-indent: -1em; padding-left: 1em; }#logmsg ol { text-indent: -1.5em; padding-left: 1.5em; }
#logmsg > ul, #logmsg > ol { margin: 0 0 1em 0; }
#logmsg pre { background: #eee; padding: 1em; }
#logmsg blockquote { border: 1px solid #fa0; border-left-width: 10px; padding: 1em 1em 0 1em; background: white;}
#logmsg dl { margin: 0; }
#logmsg dt { font-weight: bold; }
#logmsg dd { margin: 0; padding: 0 0 0.5em 0; }
#logmsg dd:before { content:'\00bb';}
#logmsg table { border-spacing: 0px; border-collapse: collapse; border-top: 4px solid #fa0; border-bottom: 1px solid #fa0; background: #fff; }
#logmsg table th { text-align: left; font-weight: normal; padding: 0.2em 0.5em; border-top: 1px dotted #fa0; }
#logmsg table td { text-align: right; border-top: 1px dotted #fa0; padding: 0.2em 0.5em; }
#logmsg table thead th { text-align: center; border-bottom: 1px solid #fa0; }
#logmsg table th.Corner { text-align: left; }
#logmsg hr { border: none 0; border-top: 2px dashed #fa0; height: 1px; }
#header, #footer { color: #fff; background: #636; border: 1px #300 solid; padding: 6px; }
#patch { width: 100%; }
#patch h4 {font-family: verdana,arial,helvetica,sans-serif;font-size:10pt;padding:8px;background:#369;color:#fff;margin:0;}
#patch .propset h4, #patch .binary h4 {margin:0;}
#patch pre {padding:0;line-height:1.2em;margin:0;}
#patch .diff {width:100%;background:#eee;padding: 0 0 10px 0;overflow:auto;}
#patch .propset .diff, #patch .binary .diff {padding:10px 0;}
#patch span {display:block;padding:0 10px;}
#patch .modfile, #patch .addfile, #patch .delfile, #patch .propset, #patch .binary, #patch .copfile {border:1px solid #ccc;margin:10px 0;}
#patch ins {background:#dfd;text-decoration:none;display:block;padding:0 10px;}
#patch del {background:#fdd;text-decoration:none;display:block;padding:0 10px;}
#patch .lines, .info {color:#888;background:#fff;}
--></style>
<div id="msg">
<dl class="meta">
<dt>Revision</dt> <dd><a href="http://trac.webkit.org/projects/webkit/changeset/202021">202021</a></dd>
<dt>Author</dt> <dd>sbarati@apple.com</dd>
<dt>Date</dt> <dd>2016-06-13 19:29:26 -0700 (Mon, 13 Jun 2016)</dd>
</dl>
<h3>Log Message</h3>
<pre>The sampling profiler should further protect itself against certain forms of sampling bias that arise due to the sampling interval being in sync with some other system process
https://bugs.webkit.org/show_bug.cgi?id=158678
Reviewed by Benjamin Poulain.
I first became aware of this problem when I read this paper:
http://plv.colorado.edu/papers/mytkowicz-pldi10.pdf
To provide background for this change, I'll quote a paragraph
from section 6.2:
"One statically sound method for collecting random samples is to collect a
sample at every t + r milliseconds, where t is the desired sampling interval
and r is a random number between −t and t. One might think that sampling every
t seconds is enough (i.e., drop the r component) but it is not: specifically,
if a profiler samples every t seconds, the sampling rate would be synchronized
with any program or system activity that occurs at regular time intervals [17].
For example, if the thread scheduler switches between threads every 10ms and our
sampling interval was also 10ms, then we may always take samples immediately after
a thread switch. Because performance is often different immediately after a thread
switch than at other points (e.g., due to cache and TLB warm-up effects) we would
get biased data. The random component, r, guards against such situations."
* runtime/SamplingProfiler.cpp:
(JSC::SamplingProfiler::timerLoop):</pre>
<h3>Modified Paths</h3>
<ul>
<li><a href="#trunkSourceJavaScriptCoreChangeLog">trunk/Source/JavaScriptCore/ChangeLog</a></li>
<li><a href="#trunkSourceJavaScriptCoreruntimeSamplingProfilercpp">trunk/Source/JavaScriptCore/runtime/SamplingProfiler.cpp</a></li>
</ul>
</div>
<div id="patch">
<h3>Diff</h3>
<a id="trunkSourceJavaScriptCoreChangeLog"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/ChangeLog (202020 => 202021)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/ChangeLog 2016-06-14 02:27:08 UTC (rev 202020)
+++ trunk/Source/JavaScriptCore/ChangeLog 2016-06-14 02:29:26 UTC (rev 202021)
</span><span class="lines">@@ -1,3 +1,30 @@
</span><ins>+2016-06-13 Saam Barati <sbarati@apple.com>
+
+ The sampling profiler should further protect itself against certain forms of sampling bias that arise due to the sampling interval being in sync with some other system process
+ https://bugs.webkit.org/show_bug.cgi?id=158678
+
+ Reviewed by Benjamin Poulain.
+
+ I first became aware of this problem when I read this paper:
+ http://plv.colorado.edu/papers/mytkowicz-pldi10.pdf
+
+ To provide background for this change, I'll quote a paragraph
+ from section 6.2:
+ "One statically sound method for collecting random samples is to collect a
+ sample at every t + r milliseconds, where t is the desired sampling interval
+ and r is a random number between −t and t. One might think that sampling every
+ t seconds is enough (i.e., drop the r component) but it is not: specifically,
+ if a profiler samples every t seconds, the sampling rate would be synchronized
+ with any program or system activity that occurs at regular time intervals [17].
+ For example, if the thread scheduler switches between threads every 10ms and our
+ sampling interval was also 10ms, then we may always take samples immediately after
+ a thread switch. Because performance is often different immediately after a thread
+ switch than at other points (e.g., due to cache and TLB warm-up effects) we would
+ get biased data. The random component, r, guards against such situations."
+
+ * runtime/SamplingProfiler.cpp:
+ (JSC::SamplingProfiler::timerLoop):
+
</ins><span class="cx"> 2016-06-13 Oliver Hunt <oliver@apple.com>
</span><span class="cx">
</span><span class="cx"> DFG Validation fails when performing a concatenation with only a single entry
</span></span></pre></div>
<a id="trunkSourceJavaScriptCoreruntimeSamplingProfilercpp"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/runtime/SamplingProfiler.cpp (202020 => 202021)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/runtime/SamplingProfiler.cpp 2016-06-14 02:27:08 UTC (rev 202020)
+++ trunk/Source/JavaScriptCore/runtime/SamplingProfiler.cpp 2016-06-14 02:29:26 UTC (rev 202021)
</span><span class="lines">@@ -48,6 +48,7 @@
</span><span class="cx"> #include "VM.h"
</span><span class="cx"> #include "VMEntryScope.h"
</span><span class="cx"> #include <wtf/HashSet.h>
</span><ins>+#include <wtf/RandomNumber.h>
</ins><span class="cx"> #include <wtf/RefPtr.h>
</span><span class="cx">
</span><span class="cx"> namespace JSC {
</span><span class="lines">@@ -233,7 +234,13 @@
</span><span class="cx"> m_lastTime = m_stopwatch->elapsedTime();
</span><span class="cx"> }
</span><span class="cx">
</span><del>- std::this_thread::sleep_for(m_timingInterval - std::min(m_timingInterval, stackTraceProcessingTime));
</del><ins>+ // Read section 6.2 of this paper for more elaboration of why we add a random
+ // fluctuation here. The main idea is to prevent our timer from being in sync
+ // with some system process such as a scheduled context switch.
+ // http://plv.colorado.edu/papers/mytkowicz-pldi10.pdf
+ double randomSignedNumber = (randomNumber() * 2.0) - 1.0; // A random number between [-1, 1).
+ std::chrono::microseconds randomFluctuation = std::chrono::microseconds(static_cast<uint64_t>(randomSignedNumber * static_cast<double>(m_timingInterval.count()) * 0.20l));
+ std::this_thread::sleep_for(m_timingInterval - std::min(m_timingInterval, stackTraceProcessingTime) + randomFluctuation);
</ins><span class="cx"> }
</span><span class="cx"> }
</span><span class="cx">
</span></span></pre>
</div>
</div>
</body>
</html>