blob: 6f66ea34c2c1908f2aee1444e4315ba826a31351 [file] [log] [blame]
/*
* Copyright (c) 2015, Oracle and/or its affiliates. All rights reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
* This code is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 only, as
* published by the Free Software Foundation. Oracle designates this
* particular file as subject to the "Classpath" exception as provided
* by Oracle in the LICENSE file that accompanied this code.
*
* This code is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
* version 2 for more details (a copy is included in the LICENSE file that
* accompanied this code).
*
* You should have received a copy of the GNU General Public License version
* 2 along with this work; if not, write to the Free Software Foundation,
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
*
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
* or visit www.oracle.com if you need additional information or have any
* questions.
*/
package jdk.nashorn.internal.codegen;
import java.util.ArrayDeque;
import java.util.Collections;
import java.util.Deque;
import jdk.nashorn.internal.ir.FunctionNode;
import jdk.nashorn.internal.ir.Node;
import jdk.nashorn.internal.ir.Statement;
import jdk.nashorn.internal.ir.visitor.SimpleNodeVisitor;
import jdk.nashorn.internal.runtime.RecompilableScriptFunctionData;
class CacheAst extends SimpleNodeVisitor {
private final Deque<RecompilableScriptFunctionData> dataStack = new ArrayDeque<>();
private final Compiler compiler;
CacheAst(final Compiler compiler) {
this.compiler = compiler;
assert !compiler.isOnDemandCompilation();
}
@Override
public boolean enterFunctionNode(final FunctionNode functionNode) {
final int id = functionNode.getId();
// It isn't necessary to keep a stack of RecompilableScriptFunctionData, but then we'd need to do a
// potentially transitive lookup with compiler.getScriptFunctionData(id) for deeper functions; this way
// we keep it constant time.
dataStack.push(dataStack.isEmpty() ? compiler.getScriptFunctionData(id) : dataStack.peek().getScriptFunctionData(id));
return true;
}
@Override
public Node leaveFunctionNode(final FunctionNode functionNode) {
final RecompilableScriptFunctionData data = dataStack.pop();
if (functionNode.isSplit()) {
// NOTE: cache only split function ASTs from eager pass. Caching non-split functions would require
// some additional work, namely creating the concept of "uncacheable" function and reworking
// ApplySpecialization to ensure that functions undergoing apply-to-call transformations are not
// cacheable as well as recomputing Symbol.useCount when caching the eagerly parsed AST.
// Recomputing Symbol.useCount would be needed so it will only reflect uses from within the
// function being cached (and not reflect uses from its own nested functions or functions it is
// nested in). This is consistent with the count an on-demand recompilation of the function would
// produce. This is important as the decision to emit shared scope calls is based on this count,
// and if it is not matched between a previous version of the code and its deoptimizing rest-of
// compilation, it can result in rest-of not emitting a shared scope call where a previous version
// of the code (compiled from a cached eager pre-pass seeing higher (global) useCount) would emit
// it, causing a mismatch in stack shapes between previous code and its rest-of.
data.setCachedAst(functionNode);
}
if (!dataStack.isEmpty() && ((dataStack.peek().getFunctionFlags() & FunctionNode.IS_SPLIT) != 0)) {
// Return a function node with no body so that caching outer functions doesn't hold on to nested
// functions' bodies. Note we're doing this only for functions directly nested inside split
// functions, since we're only caching the split ones. It is not necessary to limit body removal
// to just these functions, but it's a cheap way to prevent unnecessary AST mutations.
return functionNode.setBody(lc, functionNode.getBody().setStatements(null, Collections.<Statement>emptyList()));
}
return functionNode;
}
}