<!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>[195466] 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/195466">195466</a></dd>
<dt>Author</dt> <dd>fpizlo@apple.com</dd>
<dt>Date</dt> <dd>2016-01-22 12:03:01 -0800 (Fri, 22 Jan 2016)</dd>
</dl>

<h3>Log Message</h3>
<pre>B3 should reduce obvious forms of Shl(SShr)
https://bugs.webkit.org/show_bug.cgi?id=153362

Reviewed by Mark Lam and Saam Barati.

This is a 40% speed-up in AsmBench-0.9/dry.c.js.

* b3/B3ReduceStrength.cpp:
* b3/testb3.cpp:
(JSC::B3::testStore16Load16Z):
(JSC::B3::testSShrShl32):
(JSC::B3::testSShrShl64):
(JSC::B3::zero):
(JSC::B3::run):</pre>

<h3>Modified Paths</h3>
<ul>
<li><a href="#trunkSourceJavaScriptCoreChangeLog">trunk/Source/JavaScriptCore/ChangeLog</a></li>
<li><a href="#trunkSourceJavaScriptCoreb3B3ReduceStrengthcpp">trunk/Source/JavaScriptCore/b3/B3ReduceStrength.cpp</a></li>
<li><a href="#trunkSourceJavaScriptCoreb3testb3cpp">trunk/Source/JavaScriptCore/b3/testb3.cpp</a></li>
</ul>

</div>
<div id="patch">
<h3>Diff</h3>
<a id="trunkSourceJavaScriptCoreChangeLog"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/ChangeLog (195465 => 195466)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/ChangeLog        2016-01-22 20:00:34 UTC (rev 195465)
+++ trunk/Source/JavaScriptCore/ChangeLog        2016-01-22 20:03:01 UTC (rev 195466)
</span><span class="lines">@@ -1,3 +1,20 @@
</span><ins>+2016-01-22  Filip Pizlo  &lt;fpizlo@apple.com&gt;
+
+        B3 should reduce obvious forms of Shl(SShr)
+        https://bugs.webkit.org/show_bug.cgi?id=153362
+
+        Reviewed by Mark Lam and Saam Barati.
+
+        This is a 40% speed-up in AsmBench-0.9/dry.c.js.
+
+        * b3/B3ReduceStrength.cpp:
+        * b3/testb3.cpp:
+        (JSC::B3::testStore16Load16Z):
+        (JSC::B3::testSShrShl32):
+        (JSC::B3::testSShrShl64):
+        (JSC::B3::zero):
+        (JSC::B3::run):
+
</ins><span class="cx"> 2016-01-22  Alex Christensen  &lt;achristensen@webkit.org&gt;
</span><span class="cx"> 
</span><span class="cx">         Fix internal Windows build
</span></span></pre></div>
<a id="trunkSourceJavaScriptCoreb3B3ReduceStrengthcpp"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/b3/B3ReduceStrength.cpp (195465 => 195466)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/b3/B3ReduceStrength.cpp        2016-01-22 20:00:34 UTC (rev 195465)
+++ trunk/Source/JavaScriptCore/b3/B3ReduceStrength.cpp        2016-01-22 20:03:01 UTC (rev 195466)
</span><span class="lines">@@ -743,6 +743,55 @@
</span><span class="cx">                 break;
</span><span class="cx">             }
</span><span class="cx"> 
</span><ins>+            if (m_value-&gt;child(1)-&gt;hasInt32()
+                &amp;&amp; m_value-&gt;child(0)-&gt;opcode() == Shl
+                &amp;&amp; m_value-&gt;child(0)-&gt;child(1)-&gt;hasInt32()
+                &amp;&amp; m_value-&gt;child(1)-&gt;asInt32() == m_value-&gt;child(0)-&gt;child(1)-&gt;asInt32()) {
+                switch (m_value-&gt;child(1)-&gt;asInt32()) {
+                case 16:
+                    if (m_value-&gt;type() == Int32) {
+                        // Turn this: SShr(Shl(value, 16), 16)
+                        // Into this: SExt16(value)
+                        replaceWithNewValue(
+                            m_proc.add&lt;Value&gt;(
+                                SExt16, m_value-&gt;origin(), m_value-&gt;child(0)-&gt;child(0)));
+                    }
+                    break;
+
+                case 24:
+                    if (m_value-&gt;type() == Int32) {
+                        // Turn this: SShr(Shl(value, 24), 24)
+                        // Into this: SExt8(value)
+                        replaceWithNewValue(
+                            m_proc.add&lt;Value&gt;(
+                                SExt8, m_value-&gt;origin(), m_value-&gt;child(0)-&gt;child(0)));
+                    }
+                    break;
+
+                case 32:
+                    if (m_value-&gt;type() == Int64) {
+                        // Turn this: SShr(Shl(value, 32), 32)
+                        // Into this: SExt32(Trunc(value))
+                        replaceWithNewValue(
+                            m_proc.add&lt;Value&gt;(
+                                SExt32, m_value-&gt;origin(),
+                                m_insertionSet.insert&lt;Value&gt;(
+                                    m_index, Trunc, m_value-&gt;origin(),
+                                    m_value-&gt;child(0)-&gt;child(0))));
+                    }
+                    break;
+
+                // FIXME: Add cases for 48 and 56, but that would translate to SExt32(SExt8) or
+                // SExt32(SExt16), which we don't currently lower efficiently.
+
+                default:
+                    break;
+                }
+
+                if (m_value-&gt;opcode() != SShr)
+                    break;
+            }
+
</ins><span class="cx">             if (handleShiftByZero())
</span><span class="cx">                 break;
</span><span class="cx"> 
</span></span></pre></div>
<a id="trunkSourceJavaScriptCoreb3testb3cpp"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/b3/testb3.cpp (195465 => 195466)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/b3/testb3.cpp        2016-01-22 20:00:34 UTC (rev 195465)
+++ trunk/Source/JavaScriptCore/b3/testb3.cpp        2016-01-22 20:03:01 UTC (rev 195466)
</span><span class="lines">@@ -9501,6 +9501,48 @@
</span><span class="cx">     CHECK(compileAndRun&lt;int32_t&gt;(proc, value) == static_cast&lt;uint16_t&gt;(value));
</span><span class="cx"> }
</span><span class="cx"> 
</span><ins>+void testSShrShl32(int32_t value, int32_t sshrAmount, int32_t shlAmount)
+{
+    Procedure proc;
+    BasicBlock* root = proc.addBlock();
+
+    root-&gt;appendNew&lt;ControlValue&gt;(
+        proc, Return, Origin(),
+        root-&gt;appendNew&lt;Value&gt;(
+            proc, SShr, Origin(),
+            root-&gt;appendNew&lt;Value&gt;(
+                proc, Shl, Origin(),
+                root-&gt;appendNew&lt;Value&gt;(
+                    proc, Trunc, Origin(),
+                    root-&gt;appendNew&lt;ArgumentRegValue&gt;(proc, Origin(), GPRInfo::argumentGPR0)),
+                root-&gt;appendNew&lt;Const32Value&gt;(proc, Origin(), shlAmount)),
+            root-&gt;appendNew&lt;Const32Value&gt;(proc, Origin(), sshrAmount)));
+
+    CHECK(
+        compileAndRun&lt;int32_t&gt;(proc, value)
+        == ((value &lt;&lt; (shlAmount &amp; 31)) &gt;&gt; (sshrAmount &amp; 31)));
+}
+
+void testSShrShl64(int64_t value, int32_t sshrAmount, int32_t shlAmount)
+{
+    Procedure proc;
+    BasicBlock* root = proc.addBlock();
+
+    root-&gt;appendNew&lt;ControlValue&gt;(
+        proc, Return, Origin(),
+        root-&gt;appendNew&lt;Value&gt;(
+            proc, SShr, Origin(),
+            root-&gt;appendNew&lt;Value&gt;(
+                proc, Shl, Origin(),
+                root-&gt;appendNew&lt;ArgumentRegValue&gt;(proc, Origin(), GPRInfo::argumentGPR0),
+                root-&gt;appendNew&lt;Const32Value&gt;(proc, Origin(), shlAmount)),
+            root-&gt;appendNew&lt;Const32Value&gt;(proc, Origin(), sshrAmount)));
+
+    CHECK(
+        compileAndRun&lt;int64_t&gt;(proc, value)
+        == ((value &lt;&lt; (shlAmount &amp; 63)) &gt;&gt; (sshrAmount &amp; 63)));
+}
+
</ins><span class="cx"> // Make sure the compiler does not try to optimize anything out.
</span><span class="cx"> NEVER_INLINE double zero()
</span><span class="cx"> {
</span><span class="lines">@@ -10785,6 +10827,95 @@
</span><span class="cx">     RUN(testStore16Load16Z(12345678));
</span><span class="cx">     RUN(testStore16Load16Z(-123));
</span><span class="cx"> 
</span><ins>+    RUN(testSShrShl32(42, 24, 24));
+    RUN(testSShrShl32(-42, 24, 24));
+    RUN(testSShrShl32(4200, 24, 24));
+    RUN(testSShrShl32(-4200, 24, 24));
+    RUN(testSShrShl32(4200000, 24, 24));
+    RUN(testSShrShl32(-4200000, 24, 24));
+
+    RUN(testSShrShl32(42, 16, 16));
+    RUN(testSShrShl32(-42, 16, 16));
+    RUN(testSShrShl32(4200, 16, 16));
+    RUN(testSShrShl32(-4200, 16, 16));
+    RUN(testSShrShl32(4200000, 16, 16));
+    RUN(testSShrShl32(-4200000, 16, 16));
+
+    RUN(testSShrShl32(42, 8, 8));
+    RUN(testSShrShl32(-42, 8, 8));
+    RUN(testSShrShl32(4200, 8, 8));
+    RUN(testSShrShl32(-4200, 8, 8));
+    RUN(testSShrShl32(4200000, 8, 8));
+    RUN(testSShrShl32(-4200000, 8, 8));
+    RUN(testSShrShl32(420000000, 8, 8));
+    RUN(testSShrShl32(-420000000, 8, 8));
+
+    RUN(testSShrShl64(42, 56, 56));
+    RUN(testSShrShl64(-42, 56, 56));
+    RUN(testSShrShl64(4200, 56, 56));
+    RUN(testSShrShl64(-4200, 56, 56));
+    RUN(testSShrShl64(4200000, 56, 56));
+    RUN(testSShrShl64(-4200000, 56, 56));
+    RUN(testSShrShl64(420000000, 56, 56));
+    RUN(testSShrShl64(-420000000, 56, 56));
+    RUN(testSShrShl64(42000000000, 56, 56));
+    RUN(testSShrShl64(-42000000000, 56, 56));
+
+    RUN(testSShrShl64(42, 48, 48));
+    RUN(testSShrShl64(-42, 48, 48));
+    RUN(testSShrShl64(4200, 48, 48));
+    RUN(testSShrShl64(-4200, 48, 48));
+    RUN(testSShrShl64(4200000, 48, 48));
+    RUN(testSShrShl64(-4200000, 48, 48));
+    RUN(testSShrShl64(420000000, 48, 48));
+    RUN(testSShrShl64(-420000000, 48, 48));
+    RUN(testSShrShl64(42000000000, 48, 48));
+    RUN(testSShrShl64(-42000000000, 48, 48));
+
+    RUN(testSShrShl64(42, 32, 32));
+    RUN(testSShrShl64(-42, 32, 32));
+    RUN(testSShrShl64(4200, 32, 32));
+    RUN(testSShrShl64(-4200, 32, 32));
+    RUN(testSShrShl64(4200000, 32, 32));
+    RUN(testSShrShl64(-4200000, 32, 32));
+    RUN(testSShrShl64(420000000, 32, 32));
+    RUN(testSShrShl64(-420000000, 32, 32));
+    RUN(testSShrShl64(42000000000, 32, 32));
+    RUN(testSShrShl64(-42000000000, 32, 32));
+
+    RUN(testSShrShl64(42, 24, 24));
+    RUN(testSShrShl64(-42, 24, 24));
+    RUN(testSShrShl64(4200, 24, 24));
+    RUN(testSShrShl64(-4200, 24, 24));
+    RUN(testSShrShl64(4200000, 24, 24));
+    RUN(testSShrShl64(-4200000, 24, 24));
+    RUN(testSShrShl64(420000000, 24, 24));
+    RUN(testSShrShl64(-420000000, 24, 24));
+    RUN(testSShrShl64(42000000000, 24, 24));
+    RUN(testSShrShl64(-42000000000, 24, 24));
+
+    RUN(testSShrShl64(42, 16, 16));
+    RUN(testSShrShl64(-42, 16, 16));
+    RUN(testSShrShl64(4200, 16, 16));
+    RUN(testSShrShl64(-4200, 16, 16));
+    RUN(testSShrShl64(4200000, 16, 16));
+    RUN(testSShrShl64(-4200000, 16, 16));
+    RUN(testSShrShl64(420000000, 16, 16));
+    RUN(testSShrShl64(-420000000, 16, 16));
+    RUN(testSShrShl64(42000000000, 16, 16));
+    RUN(testSShrShl64(-42000000000, 16, 16));
+
+    RUN(testSShrShl64(42, 8, 8));
+    RUN(testSShrShl64(-42, 8, 8));
+    RUN(testSShrShl64(4200, 8, 8));
+    RUN(testSShrShl64(-4200, 8, 8));
+    RUN(testSShrShl64(4200000, 8, 8));
+    RUN(testSShrShl64(-4200000, 8, 8));
+    RUN(testSShrShl64(420000000, 8, 8));
+    RUN(testSShrShl64(-420000000, 8, 8));
+    RUN(testSShrShl64(42000000000, 8, 8));
+    RUN(testSShrShl64(-42000000000, 8, 8));
+
</ins><span class="cx">     if (tasks.isEmpty())
</span><span class="cx">         usage();
</span><span class="cx"> 
</span></span></pre>
</div>
</div>

</body>
</html>