<!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>[170017] branches/ftlopt/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/170017">170017</a></dd>
<dt>Author</dt> <dd>fpizlo@apple.com</dd>
<dt>Date</dt> <dd>2014-06-16 11:20:02 -0700 (Mon, 16 Jun 2014)</dd>
</dl>
<h3>Log Message</h3>
<pre>[ftlopt] Remove the DFG optimization fixpoint and remove some obvious reasons why we previously benefited from it
https://bugs.webkit.org/show_bug.cgi?id=133931
Reviewed by Oliver Hunt.
* dfg/DFGAbstractInterpreterInlines.h:
(JSC::DFG::AbstractInterpreter<AbstractStateType>::executeEffects): Trigger constant-folding for GetMyArgumentByVal (which means turning it into GetLocalUnlinked) and correct the handling of Upsilon so we don't fold them away.
* dfg/DFGConstantFoldingPhase.cpp:
(JSC::DFG::ConstantFoldingPhase::foldConstants): Implement constant-folding for GetMyArgumentByVal.
* dfg/DFGPlan.cpp:
(JSC::DFG::Plan::compileInThreadImpl): Remove the fixpoint.</pre>
<h3>Modified Paths</h3>
<ul>
<li><a href="#branchesftloptSourceJavaScriptCoreChangeLog">branches/ftlopt/Source/JavaScriptCore/ChangeLog</a></li>
<li><a href="#branchesftloptSourceJavaScriptCoredfgDFGAbstractInterpreterInlinesh">branches/ftlopt/Source/JavaScriptCore/dfg/DFGAbstractInterpreterInlines.h</a></li>
<li><a href="#branchesftloptSourceJavaScriptCoredfgDFGConstantFoldingPhasecpp">branches/ftlopt/Source/JavaScriptCore/dfg/DFGConstantFoldingPhase.cpp</a></li>
<li><a href="#branchesftloptSourceJavaScriptCoredfgDFGPlancpp">branches/ftlopt/Source/JavaScriptCore/dfg/DFGPlan.cpp</a></li>
</ul>
</div>
<div id="patch">
<h3>Diff</h3>
<a id="branchesftloptSourceJavaScriptCoreChangeLog"></a>
<div class="modfile"><h4>Modified: branches/ftlopt/Source/JavaScriptCore/ChangeLog (170016 => 170017)</h4>
<pre class="diff"><span>
<span class="info">--- branches/ftlopt/Source/JavaScriptCore/ChangeLog 2014-06-16 18:03:04 UTC (rev 170016)
+++ branches/ftlopt/Source/JavaScriptCore/ChangeLog 2014-06-16 18:20:02 UTC (rev 170017)
</span><span class="lines">@@ -1,5 +1,19 @@
</span><span class="cx"> 2014-06-15 Filip Pizlo <fpizlo@apple.com>
</span><span class="cx">
</span><ins>+ [ftlopt] Remove the DFG optimization fixpoint and remove some obvious reasons why we previously benefited from it
+ https://bugs.webkit.org/show_bug.cgi?id=133931
+
+ Reviewed by Oliver Hunt.
+
+ * dfg/DFGAbstractInterpreterInlines.h:
+ (JSC::DFG::AbstractInterpreter<AbstractStateType>::executeEffects): Trigger constant-folding for GetMyArgumentByVal (which means turning it into GetLocalUnlinked) and correct the handling of Upsilon so we don't fold them away.
+ * dfg/DFGConstantFoldingPhase.cpp:
+ (JSC::DFG::ConstantFoldingPhase::foldConstants): Implement constant-folding for GetMyArgumentByVal.
+ * dfg/DFGPlan.cpp:
+ (JSC::DFG::Plan::compileInThreadImpl): Remove the fixpoint.
+
+2014-06-15 Filip Pizlo <fpizlo@apple.com>
+
</ins><span class="cx"> [ftlopt] DFG OSR entry should have a crystal-clear story for when it's safe to enter at a block with a set of values
</span><span class="cx"> https://bugs.webkit.org/show_bug.cgi?id=133935
</span><span class="cx">
</span></span></pre></div>
<a id="branchesftloptSourceJavaScriptCoredfgDFGAbstractInterpreterInlinesh"></a>
<div class="modfile"><h4>Modified: branches/ftlopt/Source/JavaScriptCore/dfg/DFGAbstractInterpreterInlines.h (170016 => 170017)</h4>
<pre class="diff"><span>
<span class="info">--- branches/ftlopt/Source/JavaScriptCore/dfg/DFGAbstractInterpreterInlines.h 2014-06-16 18:03:04 UTC (rev 170016)
+++ branches/ftlopt/Source/JavaScriptCore/dfg/DFGAbstractInterpreterInlines.h 2014-06-16 18:20:02 UTC (rev 170017)
</span><span class="lines">@@ -1380,13 +1380,24 @@
</span><span class="cx"> forNode(node).makeHeapTop();
</span><span class="cx"> break;
</span><span class="cx">
</span><del>- case GetMyArgumentByVal:
</del><ins>+ case GetMyArgumentByVal: {
</ins><span class="cx"> node->setCanExit(true);
</span><del>- // We know that this executable does not escape its arguments, so we can optimize
- // the arguments a bit. Note that this ends up being further optimized by the
- // ArgumentsSimplificationPhase.
</del><ins>+ InlineCallFrame* inlineCallFrame = node->origin.semantic.inlineCallFrame;
+ JSValue value = forNode(node->child1()).m_value;
+ if (inlineCallFrame && value && value.isInt32()) {
+ int32_t index = value.asInt32();
+ if (index >= 0
+ && static_cast<size_t>(index + 1) < inlineCallFrame->arguments.size()) {
+ forNode(node) = m_state.variables().operand(
+ inlineCallFrame->stackOffset +
+ m_graph.baselineCodeBlockFor(inlineCallFrame)->argumentIndexAfterCapture(index));
+ m_state.setFoundConstants(true);
+ break;
+ }
+ }
</ins><span class="cx"> forNode(node).makeHeapTop();
</span><span class="cx"> break;
</span><ins>+ }
</ins><span class="cx">
</span><span class="cx"> case GetMyArgumentByValSafe:
</span><span class="cx"> node->setCanExit(true);
</span><span class="lines">@@ -1860,9 +1871,7 @@
</span><span class="cx">
</span><span class="cx"> case Upsilon: {
</span><span class="cx"> m_state.createValueForNode(node->phi());
</span><del>- AbstractValue value = forNode(node->child1());
- forNode(node) = value;
- forNode(node->phi()) = value;
</del><ins>+ forNode(node->phi()) = forNode(node->child1());
</ins><span class="cx"> break;
</span><span class="cx"> }
</span><span class="cx">
</span></span></pre></div>
<a id="branchesftloptSourceJavaScriptCoredfgDFGConstantFoldingPhasecpp"></a>
<div class="modfile"><h4>Modified: branches/ftlopt/Source/JavaScriptCore/dfg/DFGConstantFoldingPhase.cpp (170016 => 170017)</h4>
<pre class="diff"><span>
<span class="info">--- branches/ftlopt/Source/JavaScriptCore/dfg/DFGConstantFoldingPhase.cpp 2014-06-16 18:03:04 UTC (rev 170016)
+++ branches/ftlopt/Source/JavaScriptCore/dfg/DFGConstantFoldingPhase.cpp 2014-06-16 18:20:02 UTC (rev 170017)
</span><span class="lines">@@ -246,6 +246,34 @@
</span><span class="cx"> node->convertToIdentity();
</span><span class="cx"> break;
</span><span class="cx"> }
</span><ins>+
+ case GetMyArgumentByVal: {
+ InlineCallFrame* inlineCallFrame = node->origin.semantic.inlineCallFrame;
+ JSValue value = m_state.forNode(node->child1()).m_value;
+ if (inlineCallFrame && value && value.isInt32()) {
+ int32_t index = value.asInt32();
+ if (index >= 0
+ && static_cast<size_t>(index + 1) < inlineCallFrame->arguments.size()) {
+ // Roll the interpreter over this.
+ m_interpreter.execute(indexInBlock);
+ eliminated = true;
+
+ int operand =
+ inlineCallFrame->stackOffset +
+ m_graph.baselineCodeBlockFor(inlineCallFrame)->argumentIndexAfterCapture(index);
+
+ m_insertionSet.insertNode(
+ indexInBlock, SpecNone, CheckArgumentsNotCreated, node->origin);
+ m_insertionSet.insertNode(
+ indexInBlock, SpecNone, Phantom, node->origin, node->children);
+
+ node->convertToGetLocalUnlinked(VirtualRegister(operand));
+ break;
+ }
+ }
+
+ break;
+ }
</ins><span class="cx">
</span><span class="cx"> default:
</span><span class="cx"> break;
</span></span></pre></div>
<a id="branchesftloptSourceJavaScriptCoredfgDFGPlancpp"></a>
<div class="modfile"><h4>Modified: branches/ftlopt/Source/JavaScriptCore/dfg/DFGPlan.cpp (170016 => 170017)</h4>
<pre class="diff"><span>
<span class="info">--- branches/ftlopt/Source/JavaScriptCore/dfg/DFGPlan.cpp 2014-06-16 18:03:04 UTC (rev 170016)
+++ branches/ftlopt/Source/JavaScriptCore/dfg/DFGPlan.cpp 2014-06-16 18:20:02 UTC (rev 170017)
</span><span class="lines">@@ -238,32 +238,35 @@
</span><span class="cx"> performInvalidationPointInjection(dfg);
</span><span class="cx"> performTypeCheckHoisting(dfg);
</span><span class="cx">
</span><del>- unsigned count = 1;
</del><span class="cx"> dfg.m_fixpointState = FixpointNotConverged;
</span><del>- for (;; ++count) {
- if (logCompilationChanges(mode))
- dataLogF("DFG beginning optimization fixpoint iteration #%u.\n", count);
- bool changed = false;
</del><ins>+
+ // For now we're back to avoiding a fixpoint. Note that we've ping-ponged on this decision
+ // many times. For maximum throughput, it's best to fixpoint. But the throughput benefit is
+ // small and not likely to show up in FTL anyway. On the other hand, not fixpointing means
+ // that the compiler compiles more quickly. We want the third tier to compile quickly, which
+ // not fixpointing accomplishes; and the fourth tier shouldn't need a fixpoint.
+ if (validationEnabled())
+ validate(dfg);
</ins><span class="cx">
</span><del>- if (validationEnabled())
- validate(dfg);
-
- changed |= performStrengthReduction(dfg);
</del><ins>+ performStrengthReduction(dfg);
+ performCSE(dfg);
+ performArgumentsSimplification(dfg);
+ performCPSRethreading(dfg);
+ performCFA(dfg);
+ performConstantFolding(dfg);
+ bool changed = false;
+ changed |= performCFGSimplification(dfg);
+ changed |= performCSE(dfg);
+
+ if (validationEnabled())
+ validate(dfg);
+
+ performCPSRethreading(dfg);
+ if (changed) {
</ins><span class="cx"> performCFA(dfg);
</span><del>- changed |= performConstantFolding(dfg);
- changed |= performArgumentsSimplification(dfg);
- changed |= performCFGSimplification(dfg);
- changed |= performCSE(dfg);
-
- if (!changed)
- break;
-
- performCPSRethreading(dfg);
</del><ins>+ performConstantFolding(dfg);
</ins><span class="cx"> }
</span><span class="cx">
</span><del>- if (logCompilationChanges(mode))
- dataLogF("DFG optimization fixpoint converged in %u iterations.\n", count);
-
</del><span class="cx"> // If we're doing validation, then run some analyses, to give them an opportunity
</span><span class="cx"> // to self-validate. Now is as good a time as any to do this.
</span><span class="cx"> if (validationEnabled()) {
</span></span></pre>
</div>
</div>
</body>
</html>