<!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>[194873] 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/194873">194873</a></dd>
<dt>Author</dt> <dd>commit-queue@webkit.org</dd>
<dt>Date</dt> <dd>2016-01-11 16:11:25 -0800 (Mon, 11 Jan 2016)</dd>
</dl>

<h3>Log Message</h3>
<pre>[JSC] Add Div/Mod and fix Mul for B3 ARM64
https://bugs.webkit.org/show_bug.cgi?id=152978

Patch by Benjamin Poulain &lt;bpoulain@apple.com&gt; on 2016-01-11
Reviewed by Filip Pizlo.

Add the 3 operands forms of Mul.
Remove the form taking immediate on ARM64, there are no such instruction.

Add Div with sdiv.

Unfortunately, I discovered ChillMod's division by zero
makes it non-trivial on ARM64. I just made it into a macro like on x86.

* assembler/MacroAssemblerARM64.h:
(JSC::MacroAssemblerARM64::mul32):
(JSC::MacroAssemblerARM64::mul64):
(JSC::MacroAssemblerARM64::div32):
(JSC::MacroAssemblerARM64::div64):
* b3/B3LowerMacros.cpp:
* b3/B3LowerToAir.cpp:
(JSC::B3::Air::LowerToAir::lower):
* b3/air/AirOpcode.opcodes:</pre>

<h3>Modified Paths</h3>
<ul>
<li><a href="#trunkSourceJavaScriptCoreChangeLog">trunk/Source/JavaScriptCore/ChangeLog</a></li>
<li><a href="#trunkSourceJavaScriptCoreassemblerMacroAssemblerARM64h">trunk/Source/JavaScriptCore/assembler/MacroAssemblerARM64.h</a></li>
<li><a href="#trunkSourceJavaScriptCoreb3B3LowerMacroscpp">trunk/Source/JavaScriptCore/b3/B3LowerMacros.cpp</a></li>
<li><a href="#trunkSourceJavaScriptCoreb3B3LowerToAircpp">trunk/Source/JavaScriptCore/b3/B3LowerToAir.cpp</a></li>
<li><a href="#trunkSourceJavaScriptCoreb3airAirOpcodeopcodes">trunk/Source/JavaScriptCore/b3/air/AirOpcode.opcodes</a></li>
</ul>

</div>
<div id="patch">
<h3>Diff</h3>
<a id="trunkSourceJavaScriptCoreChangeLog"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/ChangeLog (194872 => 194873)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/ChangeLog        2016-01-11 23:44:03 UTC (rev 194872)
+++ trunk/Source/JavaScriptCore/ChangeLog        2016-01-12 00:11:25 UTC (rev 194873)
</span><span class="lines">@@ -1,3 +1,28 @@
</span><ins>+2016-01-11  Benjamin Poulain  &lt;bpoulain@apple.com&gt;
+
+        [JSC] Add Div/Mod and fix Mul for B3 ARM64
+        https://bugs.webkit.org/show_bug.cgi?id=152978
+
+        Reviewed by Filip Pizlo.
+
+        Add the 3 operands forms of Mul.
+        Remove the form taking immediate on ARM64, there are no such instruction.
+
+        Add Div with sdiv.
+
+        Unfortunately, I discovered ChillMod's division by zero
+        makes it non-trivial on ARM64. I just made it into a macro like on x86.
+
+        * assembler/MacroAssemblerARM64.h:
+        (JSC::MacroAssemblerARM64::mul32):
+        (JSC::MacroAssemblerARM64::mul64):
+        (JSC::MacroAssemblerARM64::div32):
+        (JSC::MacroAssemblerARM64::div64):
+        * b3/B3LowerMacros.cpp:
+        * b3/B3LowerToAir.cpp:
+        (JSC::B3::Air::LowerToAir::lower):
+        * b3/air/AirOpcode.opcodes:
+
</ins><span class="cx"> 2016-01-11  Keith Miller  &lt;keith_miller@apple.com&gt;
</span><span class="cx"> 
</span><span class="cx">         Arrays should use the InternalFunctionAllocationProfile when constructing new Arrays
</span></span></pre></div>
<a id="trunkSourceJavaScriptCoreassemblerMacroAssemblerARM64h"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/assembler/MacroAssemblerARM64.h (194872 => 194873)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/assembler/MacroAssemblerARM64.h        2016-01-11 23:44:03 UTC (rev 194872)
+++ trunk/Source/JavaScriptCore/assembler/MacroAssemblerARM64.h        2016-01-12 00:11:25 UTC (rev 194873)
</span><span class="lines">@@ -472,23 +472,43 @@
</span><span class="cx">     {
</span><span class="cx">         lshift64(dest, imm, dest);
</span><span class="cx">     }
</span><ins>+
+    void mul32(RegisterID left, RegisterID right, RegisterID dest)
+    {
+        m_assembler.mul&lt;32&gt;(dest, left, right);
+    }
</ins><span class="cx">     
</span><span class="cx">     void mul32(RegisterID src, RegisterID dest)
</span><span class="cx">     {
</span><span class="cx">         m_assembler.mul&lt;32&gt;(dest, dest, src);
</span><span class="cx">     }
</span><del>-    
</del><ins>+
+    void mul32(TrustedImm32 imm, RegisterID src, RegisterID dest)
+    {
+        move(imm, getCachedDataTempRegisterIDAndInvalidate());
+        m_assembler.mul&lt;32&gt;(dest, src, dataTempRegister);
+    }
+
</ins><span class="cx">     void mul64(RegisterID src, RegisterID dest)
</span><span class="cx">     {
</span><span class="cx">         m_assembler.mul&lt;64&gt;(dest, dest, src);
</span><span class="cx">     }
</span><span class="cx"> 
</span><del>-    void mul32(TrustedImm32 imm, RegisterID src, RegisterID dest)
</del><ins>+    void mul64(RegisterID left, RegisterID right, RegisterID dest)
</ins><span class="cx">     {
</span><del>-        move(imm, getCachedDataTempRegisterIDAndInvalidate());
-        m_assembler.mul&lt;32&gt;(dest, src, dataTempRegister);
</del><ins>+        m_assembler.mul&lt;64&gt;(dest, left, right);
</ins><span class="cx">     }
</span><span class="cx"> 
</span><ins>+    void div32(RegisterID dividend, RegisterID divisor, RegisterID dest)
+    {
+        m_assembler.sdiv&lt;32&gt;(dest, dividend, divisor);
+    }
+
+    void div64(RegisterID dividend, RegisterID divisor, RegisterID dest)
+    {
+        m_assembler.sdiv&lt;64&gt;(dest, dividend, divisor);
+    }
+
</ins><span class="cx">     void neg32(RegisterID dest)
</span><span class="cx">     {
</span><span class="cx">         m_assembler.neg&lt;32&gt;(dest, dest);
</span></span></pre></div>
<a id="trunkSourceJavaScriptCoreb3B3LowerMacroscpp"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/b3/B3LowerMacros.cpp (194872 => 194873)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/b3/B3LowerMacros.cpp        2016-01-11 23:44:03 UTC (rev 194872)
+++ trunk/Source/JavaScriptCore/b3/B3LowerMacros.cpp        2016-01-12 00:11:25 UTC (rev 194873)
</span><span class="lines">@@ -84,6 +84,7 @@
</span><span class="cx">                         m_value-&gt;child(0),
</span><span class="cx">                         m_value-&gt;child(1));
</span><span class="cx">                     m_value-&gt;replaceWithIdentity(result);
</span><ins>+                    m_changed = true;
</ins><span class="cx">                 } else if (m_value-&gt;type() == Float) {
</span><span class="cx">                     Value* numeratorAsDouble = m_insertionSet.insert&lt;Value&gt;(m_index, FloatToDouble, m_origin, m_value-&gt;child(0));
</span><span class="cx">                     Value* denominatorAsDouble = m_insertionSet.insert&lt;Value&gt;(m_index, FloatToDouble, m_origin, m_value-&gt;child(1));
</span><span class="lines">@@ -95,6 +96,13 @@
</span><span class="cx">                         denominatorAsDouble);
</span><span class="cx">                     Value* result = m_insertionSet.insert&lt;Value&gt;(m_index, DoubleToFloat, m_origin, doubleMod);
</span><span class="cx">                     m_value-&gt;replaceWithIdentity(result);
</span><ins>+                    m_changed = true;
+                } else if (isARM64()) {
+                    Value* divResult = m_insertionSet.insert&lt;Value&gt;(m_index, ChillDiv, m_origin, m_value-&gt;child(0), m_value-&gt;child(1));
+                    Value* multipliedBack = m_insertionSet.insert&lt;Value&gt;(m_index, Mul, m_origin, divResult, m_value-&gt;child(1));
+                    Value* result = m_insertionSet.insert&lt;Value&gt;(m_index, Sub, m_origin, m_value-&gt;child(0), multipliedBack);
+                    m_value-&gt;replaceWithIdentity(result);
+                    m_changed = true;
</ins><span class="cx">                 }
</span><span class="cx">                 break;
</span><span class="cx">             }
</span><span class="lines">@@ -104,7 +112,34 @@
</span><span class="cx">             }
</span><span class="cx"> 
</span><span class="cx">             case ChillMod: {
</span><del>-                makeDivisionChill(Mod);
</del><ins>+                if (isARM64()) {
+                    BasicBlock* before = m_blockInsertionSet.splitForward(m_block, m_index, &amp;m_insertionSet);
+                    BasicBlock* zeroDenCase = m_blockInsertionSet.insertBefore(m_block);
+                    BasicBlock* normalModCase = m_blockInsertionSet.insertBefore(m_block);
+
+                    before-&gt;replaceLastWithNew&lt;ControlValue&gt;(
+                        m_proc, Branch, m_origin, m_value-&gt;child(1),
+                        FrequentedBlock(normalModCase, FrequencyClass::Normal),
+                        FrequentedBlock(zeroDenCase, FrequencyClass::Rare));
+
+                    Value* divResult = normalModCase-&gt;appendNew&lt;Value&gt;(m_proc, ChillDiv, m_origin, m_value-&gt;child(0), m_value-&gt;child(1));
+                    Value* multipliedBack = normalModCase-&gt;appendNew&lt;Value&gt;(m_proc, Mul, m_origin, divResult, m_value-&gt;child(1));
+                    Value* result = normalModCase-&gt;appendNew&lt;Value&gt;(m_proc, Sub, m_origin, m_value-&gt;child(0), multipliedBack);
+                    UpsilonValue* normalResult = normalModCase-&gt;appendNew&lt;UpsilonValue&gt;(m_proc, m_origin, result);
+                    normalModCase-&gt;appendNew&lt;ControlValue&gt;(m_proc, Jump, m_origin, FrequentedBlock(m_block));
+
+                    UpsilonValue* zeroResult = zeroDenCase-&gt;appendNew&lt;UpsilonValue&gt;(
+                        m_proc, m_origin,
+                        zeroDenCase-&gt;appendIntConstant(m_proc, m_value, 0));
+                    zeroDenCase-&gt;appendNew&lt;ControlValue&gt;(m_proc, Jump, m_origin, FrequentedBlock(m_block));
+
+                    Value* phi = m_insertionSet.insert&lt;Value&gt;(m_index, Phi, m_value-&gt;type(), m_origin);
+                    normalResult-&gt;setPhi(phi);
+                    zeroResult-&gt;setPhi(phi);
+                    m_value-&gt;replaceWithIdentity(phi);
+                    m_changed = true;
+                } else
+                    makeDivisionChill(Mod);
</ins><span class="cx">                 break;
</span><span class="cx">             }
</span><span class="cx"> 
</span></span></pre></div>
<a id="trunkSourceJavaScriptCoreb3B3LowerToAircpp"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/b3/B3LowerToAir.cpp (194872 => 194873)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/b3/B3LowerToAir.cpp        2016-01-11 23:44:03 UTC (rev 194872)
+++ trunk/Source/JavaScriptCore/b3/B3LowerToAir.cpp        2016-01-12 00:11:25 UTC (rev 194873)
</span><span class="lines">@@ -1641,21 +1641,25 @@
</span><span class="cx">             return;
</span><span class="cx">         }
</span><span class="cx"> 
</span><ins>+        case ChillDiv:
+            RELEASE_ASSERT(isARM64());
+            FALLTHROUGH;
</ins><span class="cx">         case Div: {
</span><ins>+#if CPU(X86) || CPU(X86_64)
</ins><span class="cx">             if (isInt(m_value-&gt;type())) {
</span><del>-#if CPU(X86) || CPU(X86_64)
</del><span class="cx">                 lowerX86Div();
</span><span class="cx">                 append(Move, Tmp(X86Registers::eax), tmp(m_value));
</span><del>-#endif
</del><span class="cx">                 return;
</span><span class="cx">             }
</span><del>-            ASSERT(isFloat(m_value-&gt;type()));
</del><ins>+#endif
+            ASSERT(!isX86() || isFloat(m_value-&gt;type()));
</ins><span class="cx"> 
</span><del>-            appendBinOp&lt;Air::Oops, Air::Oops, DivDouble, DivFloat&gt;(m_value-&gt;child(0), m_value-&gt;child(1));
</del><ins>+            appendBinOp&lt;Div32, Div64, DivDouble, DivFloat&gt;(m_value-&gt;child(0), m_value-&gt;child(1));
</ins><span class="cx">             return;
</span><span class="cx">         }
</span><span class="cx"> 
</span><span class="cx">         case Mod: {
</span><ins>+            RELEASE_ASSERT(isX86());
</ins><span class="cx"> #if CPU(X86) || CPU(X86_64)
</span><span class="cx">             lowerX86Div();
</span><span class="cx">             append(Move, Tmp(X86Registers::edx), tmp(m_value));
</span></span></pre></div>
<a id="trunkSourceJavaScriptCoreb3airAirOpcodeopcodes"></a>
<div class="modfile"><h4>Modified: trunk/Source/JavaScriptCore/b3/air/AirOpcode.opcodes (194872 => 194873)</h4>
<pre class="diff"><span>
<span class="info">--- trunk/Source/JavaScriptCore/b3/air/AirOpcode.opcodes        2016-01-11 23:44:03 UTC (rev 194872)
+++ trunk/Source/JavaScriptCore/b3/air/AirOpcode.opcodes        2016-01-12 00:11:25 UTC (rev 194873)
</span><span class="lines">@@ -186,11 +186,21 @@
</span><span class="cx">     x86: Addr, Tmp
</span><span class="cx"> 
</span><span class="cx"> Mul32 U:G:32, U:G:32, ZD:G:32
</span><del>-    Imm, Tmp, Tmp
</del><ins>+    arm64: Tmp, Tmp, Tmp
+    x86: Imm, Tmp, Tmp
</ins><span class="cx"> 
</span><span class="cx"> 64: Mul64 U:G:64, UD:G:64
</span><span class="cx">     Tmp, Tmp
</span><span class="cx"> 
</span><ins>+arm64: Mul64 U:G:64, U:G:64, D:G:64
+    Tmp, Tmp, Tmp
+
+arm64: Div32 U:G:32, U:G:32, ZD:G:32
+    Tmp, Tmp, Tmp
+
+arm64: Div64 U:G:64, U:G:64, D:G:64
+    Tmp, Tmp, Tmp
+
</ins><span class="cx"> arm64: MulDouble U:F:64, U:F:64, D:F:64
</span><span class="cx">     Tmp, Tmp, Tmp
</span><span class="cx"> 
</span></span></pre>
</div>
</div>

</body>
</html>