org.eclipse.sisu.space.asm.Label Maven / Gradle / Ivy
/*
* Copyright (c) 2010-2024 Sonatype, Inc.
* All rights reserved. This program and the accompanying materials
* are made available under the terms of the Eclipse Public License v2.0
* which accompanies this distribution, and is available at
* https://www.eclipse.org/legal/epl-2.0/
*
* Contributors:
* Stuart McCulloch (Sonatype, Inc.) - initial API and implementation
*/
// ASM: a very small and fast Java bytecode manipulation framework
// Copyright (c) 2000-2011 INRIA, France Telecom
// All rights reserved.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions
// are met:
// 1. Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
// 2. Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
// 3. Neither the name of the copyright holders nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
// ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
// LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
// CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
// SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
// INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
// CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
// ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
// THE POSSIBILITY OF SUCH DAMAGE.
package org.eclipse.sisu.space.asm;
/**
* A position in the bytecode of a method. Labels are used for jump, goto, and switch instructions,
* and for try catch blocks. A label designates the instruction that is just after. Note
* however that there can be other elements between a label and the instruction it designates (such
* as other labels, stack map frames, line numbers, etc.).
*
* @author Eric Bruneton
*/
public class Label {
/**
* A flag indicating that a label is only used for debug attributes. Such a label is not the start
* of a basic block, the target of a jump instruction, or an exception handler. It can be safely
* ignored in control flow graph analysis algorithms (for optimization purposes).
*/
static final int FLAG_DEBUG_ONLY = 1;
/**
* A flag indicating that a label is the target of a jump instruction, or the start of an
* exception handler.
*/
static final int FLAG_JUMP_TARGET = 2;
/** A flag indicating that the bytecode offset of a label is known. */
static final int FLAG_RESOLVED = 4;
/** A flag indicating that a label corresponds to a reachable basic block. */
static final int FLAG_REACHABLE = 8;
/**
* A flag indicating that the basic block corresponding to a label ends with a subroutine call. By
* construction in {@link MethodWriter#visitJumpInsn}, labels with this flag set have at least two
* outgoing edges:
*
*
* - the first one corresponds to the instruction that follows the jsr instruction in the
* bytecode, i.e. where execution continues when it returns from the jsr call. This is a
* virtual control flow edge, since execution never goes directly from the jsr to the next
* instruction. Instead, it goes to the subroutine and eventually returns to the instruction
* following the jsr. This virtual edge is used to compute the real outgoing edges of the
* basic blocks ending with a ret instruction, in {@link #addSubroutineRetSuccessors}.
*
- the second one corresponds to the target of the jsr instruction,
*
*/
static final int FLAG_SUBROUTINE_CALLER = 16;
/**
* A flag indicating that the basic block corresponding to a label is the start of a subroutine.
*/
static final int FLAG_SUBROUTINE_START = 32;
/** A flag indicating that the basic block corresponding to a label is the end of a subroutine. */
static final int FLAG_SUBROUTINE_END = 64;
/** A flag indicating that this label has at least one associated line number. */
static final int FLAG_LINE_NUMBER = 128;
/**
* The number of elements to add to the {@link #otherLineNumbers} array when it needs to be
* resized to store a new source line number.
*/
static final int LINE_NUMBERS_CAPACITY_INCREMENT = 4;
/**
* The number of elements to add to the {@link #forwardReferences} array when it needs to be
* resized to store a new forward reference.
*/
static final int FORWARD_REFERENCES_CAPACITY_INCREMENT = 6;
/**
* The bit mask to extract the type of a forward reference to this label. The extracted type is
* either {@link #FORWARD_REFERENCE_TYPE_SHORT} or {@link #FORWARD_REFERENCE_TYPE_WIDE}.
*
* @see #forwardReferences
*/
static final int FORWARD_REFERENCE_TYPE_MASK = 0xF0000000;
/**
* The type of forward references stored with two bytes in the bytecode. This is the case, for
* instance, of a forward reference from an ifnull instruction.
*/
static final int FORWARD_REFERENCE_TYPE_SHORT = 0x10000000;
/**
* The type of forward references stored in four bytes in the bytecode. This is the case, for
* instance, of a forward reference from a lookupswitch instruction.
*/
static final int FORWARD_REFERENCE_TYPE_WIDE = 0x20000000;
/**
* The type of forward references stored in two bytes in the stack map table. This is the
* case of the labels of {@link Frame#ITEM_UNINITIALIZED} stack map frame elements, when the NEW
* instruction is after the <init> constructor call (in bytecode offset order).
*/
static final int FORWARD_REFERENCE_TYPE_STACK_MAP = 0x30000000;
/**
* The bit mask to extract the 'handle' of a forward reference to this label. The extracted handle
* is the bytecode offset where the forward reference value is stored (using either 2 or 4 bytes,
* as indicated by the {@link #FORWARD_REFERENCE_TYPE_MASK}).
*
* @see #forwardReferences
*/
static final int FORWARD_REFERENCE_HANDLE_MASK = 0x0FFFFFFF;
/**
* A sentinel element used to indicate the end of a list of labels.
*
* @see #nextListElement
*/
static final Label EMPTY_LIST = new Label();
/**
* A user managed state associated with this label. Warning: this field is used by the ASM tree
* package. In order to use it with the ASM tree package you must override the getLabelNode method
* in MethodNode.
*/
public Object info;
/**
* The type and status of this label or its corresponding basic block. Must be zero or more of
* {@link #FLAG_DEBUG_ONLY}, {@link #FLAG_JUMP_TARGET}, {@link #FLAG_RESOLVED}, {@link
* #FLAG_REACHABLE}, {@link #FLAG_SUBROUTINE_CALLER}, {@link #FLAG_SUBROUTINE_START}, {@link
* #FLAG_SUBROUTINE_END}.
*/
short flags;
/**
* The source line number corresponding to this label, if {@link #FLAG_LINE_NUMBER} is set. If
* there are several source line numbers corresponding to this label, the first one is stored in
* this field, and the remaining ones are stored in {@link #otherLineNumbers}.
*/
private short lineNumber;
/**
* The source line numbers corresponding to this label, in addition to {@link #lineNumber}, or
* null. The first element of this array is the number n of source line numbers it contains, which
* are stored between indices 1 and n (inclusive).
*/
private int[] otherLineNumbers;
/**
* The offset of this label in the bytecode of its method, in bytes. This value is set if and only
* if the {@link #FLAG_RESOLVED} flag is set.
*/
int bytecodeOffset;
/**
* The forward references to this label. The first element is the number of forward references,
* times 2 (this corresponds to the index of the last element actually used in this array). Then,
* each forward reference is described with two consecutive integers noted
* 'sourceInsnBytecodeOffset' and 'reference':
*
*
* - 'sourceInsnBytecodeOffset' is the bytecode offset of the instruction that contains the
* forward reference,
*
- 'reference' contains the type and the offset in the bytecode where the forward reference
* value must be stored, which can be extracted with {@link #FORWARD_REFERENCE_TYPE_MASK}
* and {@link #FORWARD_REFERENCE_HANDLE_MASK}.
*
*
* For instance, for an ifnull instruction at bytecode offset x, 'sourceInsnBytecodeOffset' is
* equal to x, and 'reference' is of type {@link #FORWARD_REFERENCE_TYPE_SHORT} with value x + 1
* (because the ifnull instruction uses a 2 bytes bytecode offset operand stored one byte after
* the start of the instruction itself). For the default case of a lookupswitch instruction at
* bytecode offset x, 'sourceInsnBytecodeOffset' is equal to x, and 'reference' is of type {@link
* #FORWARD_REFERENCE_TYPE_WIDE} with value between x + 1 and x + 4 (because the lookupswitch
* instruction uses a 4 bytes bytecode offset operand stored one to four bytes after the start of
* the instruction itself).
*/
private int[] forwardReferences;
// -----------------------------------------------------------------------------------------------
// Fields for the control flow and data flow graph analysis algorithms (used to compute the
// maximum stack size or the stack map frames). A control flow graph contains one node per "basic
// block", and one edge per "jump" from one basic block to another. Each node (i.e., each basic
// block) is represented with the Label object that corresponds to the first instruction of this
// basic block. Each node also stores the list of its successors in the graph, as a linked list of
// Edge objects.
//
// The control flow analysis algorithms used to compute the maximum stack size or the stack map
// frames are similar and use two steps. The first step, during the visit of each instruction,
// builds information about the state of the local variables and the operand stack at the end of
// each basic block, called the "output frame", relatively to the frame state at the
// beginning of the basic block, which is called the "input frame", and which is unknown
// during this step. The second step, in {@link MethodWriter#computeAllFrames} and {@link
// MethodWriter#computeMaxStackAndLocal}, is a fix point algorithm
// that computes information about the input frame of each basic block, from the input state of
// the first basic block (known from the method signature), and by the using the previously
// computed relative output frames.
//
// The algorithm used to compute the maximum stack size only computes the relative output and
// absolute input stack heights, while the algorithm used to compute stack map frames computes
// relative output frames and absolute input frames.
/**
* The number of elements in the input stack of the basic block corresponding to this label. This
* field is computed in {@link MethodWriter#computeMaxStackAndLocal}.
*/
short inputStackSize;
/**
* The number of elements in the output stack, at the end of the basic block corresponding to this
* label. This field is only computed for basic blocks that end with a RET instruction.
*/
short outputStackSize;
/**
* The maximum height reached by the output stack, relatively to the top of the input stack, in
* the basic block corresponding to this label. This maximum is always positive or {@literal
* null}.
*/
short outputStackMax;
/**
* The id of the subroutine to which this basic block belongs, or 0. If the basic block belongs to
* several subroutines, this is the id of the "oldest" subroutine that contains it (with the
* convention that a subroutine calling another one is "older" than the callee). This field is
* computed in {@link MethodWriter#computeMaxStackAndLocal}, if the method contains JSR
* instructions.
*/
short subroutineId;
/**
* The input and output stack map frames of the basic block corresponding to this label. This
* field is only used when the {@link MethodWriter#COMPUTE_ALL_FRAMES} or {@link
* MethodWriter#COMPUTE_INSERTED_FRAMES} option is used.
*/
Frame frame;
/**
* The successor of this label, in the order they are visited in {@link MethodVisitor#visitLabel}.
* This linked list does not include labels used for debug info only. If the {@link
* MethodWriter#COMPUTE_ALL_FRAMES} or {@link MethodWriter#COMPUTE_INSERTED_FRAMES} option is used
* then it does not contain either successive labels that denote the same bytecode offset (in this
* case only the first label appears in this list).
*/
Label nextBasicBlock;
/**
* The outgoing edges of the basic block corresponding to this label, in the control flow graph of
* its method. These edges are stored in a linked list of {@link Edge} objects, linked to each
* other by their {@link Edge#nextEdge} field.
*/
Edge outgoingEdges;
/**
* The next element in the list of labels to which this label belongs, or {@literal null} if it
* does not belong to any list. All lists of labels must end with the {@link #EMPTY_LIST}
* sentinel, in order to ensure that this field is null if and only if this label does not belong
* to a list of labels. Note that there can be several lists of labels at the same time, but that
* a label can belong to at most one list at a time (unless some lists share a common tail, but
* this is not used in practice).
*
*
List of labels are used in {@link MethodWriter#computeAllFrames} and {@link
* MethodWriter#computeMaxStackAndLocal} to compute stack map frames and the maximum stack size,
* respectively, as well as in {@link #markSubroutine} and {@link #addSubroutineRetSuccessors} to
* compute the basic blocks belonging to subroutines and their outgoing edges. Outside of these
* methods, this field should be null (this property is a precondition and a postcondition of
* these methods).
*/
Label nextListElement;
// -----------------------------------------------------------------------------------------------
// Constructor and accessors
// -----------------------------------------------------------------------------------------------
/** Constructs a new label. */
public Label() {
// Nothing to do.
}
/**
* Returns the bytecode offset corresponding to this label. This offset is computed from the start
* of the method's bytecode. This method is intended for {@link Attribute} sub classes, and is
* normally not needed by class generators or adapters.
*
* @return the bytecode offset corresponding to this label.
* @throws IllegalStateException if this label is not resolved yet.
*/
public int getOffset() {
if ((flags & FLAG_RESOLVED) == 0) {
throw new IllegalStateException("Label offset position has not been resolved yet");
}
return bytecodeOffset;
}
/**
* Returns the "canonical" {@link Label} instance corresponding to this label's bytecode offset,
* if known, otherwise the label itself. The canonical instance is the first label (in the order
* of their visit by {@link MethodVisitor#visitLabel}) corresponding to this bytecode offset. It
* cannot be known for labels which have not been visited yet.
*
*
This method should only be used when the {@link MethodWriter#COMPUTE_ALL_FRAMES} option
* is used.
*
* @return the label itself if {@link #frame} is null, otherwise the Label's frame owner. This
* corresponds to the "canonical" label instance described above thanks to the way the label
* frame is set in {@link MethodWriter#visitLabel}.
*/
final Label getCanonicalInstance() {
return frame == null ? this : frame.owner;
}
// -----------------------------------------------------------------------------------------------
// Methods to manage line numbers
// -----------------------------------------------------------------------------------------------
/**
* Adds a source line number corresponding to this label.
*
* @param lineNumber a source line number (which should be strictly positive).
*/
final void addLineNumber(final int lineNumber) {
if ((flags & FLAG_LINE_NUMBER) == 0) {
flags |= FLAG_LINE_NUMBER;
this.lineNumber = (short) lineNumber;
} else {
if (otherLineNumbers == null) {
otherLineNumbers = new int[LINE_NUMBERS_CAPACITY_INCREMENT];
}
int otherLineNumberIndex = ++otherLineNumbers[0];
if (otherLineNumberIndex >= otherLineNumbers.length) {
int[] newLineNumbers = new int[otherLineNumbers.length + LINE_NUMBERS_CAPACITY_INCREMENT];
System.arraycopy(otherLineNumbers, 0, newLineNumbers, 0, otherLineNumbers.length);
otherLineNumbers = newLineNumbers;
}
otherLineNumbers[otherLineNumberIndex] = lineNumber;
}
}
/**
* Makes the given visitor visit this label and its source line numbers, if applicable.
*
* @param methodVisitor a method visitor.
* @param visitLineNumbers whether to visit of the label's source line numbers, if any.
*/
final void accept(final MethodVisitor methodVisitor, final boolean visitLineNumbers) {
methodVisitor.visitLabel(this);
if (visitLineNumbers && (flags & FLAG_LINE_NUMBER) != 0) {
methodVisitor.visitLineNumber(lineNumber & 0xFFFF, this);
if (otherLineNumbers != null) {
for (int i = 1; i <= otherLineNumbers[0]; ++i) {
methodVisitor.visitLineNumber(otherLineNumbers[i], this);
}
}
}
}
// -----------------------------------------------------------------------------------------------
// Methods to compute offsets and to manage forward references
// -----------------------------------------------------------------------------------------------
/**
* Puts a reference to this label in the bytecode of a method. If the bytecode offset of the label
* is known, the relative bytecode offset between the label and the instruction referencing it is
* computed and written directly. Otherwise, a null relative offset is written and a new forward
* reference is declared for this label.
*
* @param code the bytecode of the method. This is where the reference is appended.
* @param sourceInsnBytecodeOffset the bytecode offset of the instruction that contains the
* reference to be appended.
* @param wideReference whether the reference must be stored in 4 bytes (instead of 2 bytes).
*/
final void put(
final ByteVector code, final int sourceInsnBytecodeOffset, final boolean wideReference) {
if ((flags & FLAG_RESOLVED) == 0) {
if (wideReference) {
addForwardReference(sourceInsnBytecodeOffset, FORWARD_REFERENCE_TYPE_WIDE, code.length);
code.putInt(-1);
} else {
addForwardReference(sourceInsnBytecodeOffset, FORWARD_REFERENCE_TYPE_SHORT, code.length);
code.putShort(-1);
}
} else {
if (wideReference) {
code.putInt(bytecodeOffset - sourceInsnBytecodeOffset);
} else {
code.putShort(bytecodeOffset - sourceInsnBytecodeOffset);
}
}
}
/**
* Puts a reference to this label in the stack map table of a method. If the bytecode
* offset of the label is known, it is written directly. Otherwise, a null relative offset is
* written and a new forward reference is declared for this label.
*
* @param stackMapTableEntries the stack map table where the label offset must be added.
*/
final void put(final ByteVector stackMapTableEntries) {
if ((flags & FLAG_RESOLVED) == 0) {
addForwardReference(0, FORWARD_REFERENCE_TYPE_STACK_MAP, stackMapTableEntries.length);
}
stackMapTableEntries.putShort(bytecodeOffset);
}
/**
* Adds a forward reference to this label. This method must be called only for a true forward
* reference, i.e. only if this label is not resolved yet. For backward references, the relative
* bytecode offset of the reference can be, and must be, computed and stored directly.
*
* @param sourceInsnBytecodeOffset the bytecode offset of the instruction that contains the
* reference stored at referenceHandle.
* @param referenceType either {@link #FORWARD_REFERENCE_TYPE_SHORT} or {@link
* #FORWARD_REFERENCE_TYPE_WIDE}.
* @param referenceHandle the offset in the bytecode where the forward reference value must be
* stored.
*/
private void addForwardReference(
final int sourceInsnBytecodeOffset, final int referenceType, final int referenceHandle) {
if (forwardReferences == null) {
forwardReferences = new int[FORWARD_REFERENCES_CAPACITY_INCREMENT];
}
int lastElementIndex = forwardReferences[0];
if (lastElementIndex + 2 >= forwardReferences.length) {
int[] newValues = new int[forwardReferences.length + FORWARD_REFERENCES_CAPACITY_INCREMENT];
System.arraycopy(forwardReferences, 0, newValues, 0, forwardReferences.length);
forwardReferences = newValues;
}
forwardReferences[++lastElementIndex] = sourceInsnBytecodeOffset;
forwardReferences[++lastElementIndex] = referenceType | referenceHandle;
forwardReferences[0] = lastElementIndex;
}
/**
* Sets the bytecode offset of this label to the given value and resolves the forward references
* to this label, if any. This method must be called when this label is added to the bytecode of
* the method, i.e. when its bytecode offset becomes known. This method fills in the blanks that
* where left in the bytecode (and optionally in the stack map table) by each forward reference
* previously added to this label.
*
* @param code the bytecode of the method.
* @param stackMapTableEntries the 'entries' array of the StackMapTable code attribute of the
* method. Maybe {@literal null}.
* @param bytecodeOffset the bytecode offset of this label.
* @return {@literal true} if a blank that was left for this label was too small to store the
* offset. In such a case the corresponding jump instruction is replaced with an equivalent
* ASM specific instruction using an unsigned two bytes offset. These ASM specific
* instructions are later replaced with standard bytecode instructions with wider offsets (4
* bytes instead of 2), in ClassReader.
*/
final boolean resolve(
final byte[] code, final ByteVector stackMapTableEntries, final int bytecodeOffset) {
this.flags |= FLAG_RESOLVED;
this.bytecodeOffset = bytecodeOffset;
if (forwardReferences == null) {
return false;
}
boolean hasAsmInstructions = false;
for (int i = forwardReferences[0]; i > 0; i -= 2) {
final int sourceInsnBytecodeOffset = forwardReferences[i - 1];
final int reference = forwardReferences[i];
final int relativeOffset = bytecodeOffset - sourceInsnBytecodeOffset;
int handle = reference & FORWARD_REFERENCE_HANDLE_MASK;
if ((reference & FORWARD_REFERENCE_TYPE_MASK) == FORWARD_REFERENCE_TYPE_SHORT) {
if (relativeOffset < Short.MIN_VALUE || relativeOffset > Short.MAX_VALUE) {
// Change the opcode of the jump instruction, in order to be able to find it later in
// ClassReader. These ASM specific opcodes are similar to jump instruction opcodes, except
// that the 2 bytes offset is unsigned (and can therefore represent values from 0 to
// 65535, which is sufficient since the size of a method is limited to 65535 bytes).
int opcode = code[sourceInsnBytecodeOffset] & 0xFF;
if (opcode < Opcodes.IFNULL) {
// Change IFEQ ... JSR to ASM_IFEQ ... ASM_JSR.
code[sourceInsnBytecodeOffset] = (byte) (opcode + Constants.ASM_OPCODE_DELTA);
} else {
// Change IFNULL and IFNONNULL to ASM_IFNULL and ASM_IFNONNULL.
code[sourceInsnBytecodeOffset] = (byte) (opcode + Constants.ASM_IFNULL_OPCODE_DELTA);
}
hasAsmInstructions = true;
}
code[handle++] = (byte) (relativeOffset >>> 8);
code[handle] = (byte) relativeOffset;
} else if ((reference & FORWARD_REFERENCE_TYPE_MASK) == FORWARD_REFERENCE_TYPE_WIDE) {
code[handle++] = (byte) (relativeOffset >>> 24);
code[handle++] = (byte) (relativeOffset >>> 16);
code[handle++] = (byte) (relativeOffset >>> 8);
code[handle] = (byte) relativeOffset;
} else {
stackMapTableEntries.data[handle++] = (byte) (bytecodeOffset >>> 8);
stackMapTableEntries.data[handle] = (byte) bytecodeOffset;
}
}
return hasAsmInstructions;
}
// -----------------------------------------------------------------------------------------------
// Methods related to subroutines
// -----------------------------------------------------------------------------------------------
/**
* Finds the basic blocks that belong to the subroutine starting with the basic block
* corresponding to this label, and marks these blocks as belonging to this subroutine. This
* method follows the control flow graph to find all the blocks that are reachable from the
* current basic block WITHOUT following any jsr target.
*
*
Note: a precondition and postcondition of this method is that all labels must have a null
* {@link #nextListElement}.
*
* @param subroutineId the id of the subroutine starting with the basic block corresponding to
* this label.
*/
final void markSubroutine(final short subroutineId) {
// Data flow algorithm: put this basic block in a list of blocks to process (which are blocks
// belonging to subroutine subroutineId) and, while there are blocks to process, remove one from
// the list, mark it as belonging to the subroutine, and add its successor basic blocks in the
// control flow graph to the list of blocks to process (if not already done).
Label listOfBlocksToProcess = this;
listOfBlocksToProcess.nextListElement = EMPTY_LIST;
while (listOfBlocksToProcess != EMPTY_LIST) {
// Remove a basic block from the list of blocks to process.
Label basicBlock = listOfBlocksToProcess;
listOfBlocksToProcess = listOfBlocksToProcess.nextListElement;
basicBlock.nextListElement = null;
// If it is not already marked as belonging to a subroutine, mark it as belonging to
// subroutineId and add its successors to the list of blocks to process (unless already done).
if (basicBlock.subroutineId == 0) {
basicBlock.subroutineId = subroutineId;
listOfBlocksToProcess = basicBlock.pushSuccessors(listOfBlocksToProcess);
}
}
}
/**
* Finds the basic blocks that end a subroutine starting with the basic block corresponding to
* this label and, for each one of them, adds an outgoing edge to the basic block following the
* given subroutine call. In other words, completes the control flow graph by adding the edges
* corresponding to the return from this subroutine, when called from the given caller basic
* block.
*
*
Note: a precondition and postcondition of this method is that all labels must have a null
* {@link #nextListElement}.
*
* @param subroutineCaller a basic block that ends with a jsr to the basic block corresponding to
* this label. This label is supposed to correspond to the start of a subroutine.
*/
final void addSubroutineRetSuccessors(final Label subroutineCaller) {
// Data flow algorithm: put this basic block in a list blocks to process (which are blocks
// belonging to a subroutine starting with this label) and, while there are blocks to process,
// remove one from the list, put it in a list of blocks that have been processed, add a return
// edge to the successor of subroutineCaller if applicable, and add its successor basic blocks
// in the control flow graph to the list of blocks to process (if not already done).
Label listOfProcessedBlocks = EMPTY_LIST;
Label listOfBlocksToProcess = this;
listOfBlocksToProcess.nextListElement = EMPTY_LIST;
while (listOfBlocksToProcess != EMPTY_LIST) {
// Move a basic block from the list of blocks to process to the list of processed blocks.
Label basicBlock = listOfBlocksToProcess;
listOfBlocksToProcess = basicBlock.nextListElement;
basicBlock.nextListElement = listOfProcessedBlocks;
listOfProcessedBlocks = basicBlock;
// Add an edge from this block to the successor of the caller basic block, if this block is
// the end of a subroutine and if this block and subroutineCaller do not belong to the same
// subroutine.
if ((basicBlock.flags & FLAG_SUBROUTINE_END) != 0
&& basicBlock.subroutineId != subroutineCaller.subroutineId) {
basicBlock.outgoingEdges =
new Edge(
basicBlock.outputStackSize,
// By construction, the first outgoing edge of a basic block that ends with a jsr
// instruction leads to the jsr continuation block, i.e. where execution continues
// when ret is called (see {@link #FLAG_SUBROUTINE_CALLER}).
subroutineCaller.outgoingEdges.successor,
basicBlock.outgoingEdges);
}
// Add its successors to the list of blocks to process. Note that {@link #pushSuccessors} does
// not push basic blocks which are already in a list. Here this means either in the list of
// blocks to process, or in the list of already processed blocks. This second list is
// important to make sure we don't reprocess an already processed block.
listOfBlocksToProcess = basicBlock.pushSuccessors(listOfBlocksToProcess);
}
// Reset the {@link #nextListElement} of all the basic blocks that have been processed to null,
// so that this method can be called again with a different subroutine or subroutine caller.
while (listOfProcessedBlocks != EMPTY_LIST) {
Label newListOfProcessedBlocks = listOfProcessedBlocks.nextListElement;
listOfProcessedBlocks.nextListElement = null;
listOfProcessedBlocks = newListOfProcessedBlocks;
}
}
/**
* Adds the successors of this label in the method's control flow graph (except those
* corresponding to a jsr target, and those already in a list of labels) to the given list of
* blocks to process, and returns the new list.
*
* @param listOfLabelsToProcess a list of basic blocks to process, linked together with their
* {@link #nextListElement} field.
* @return the new list of blocks to process.
*/
private Label pushSuccessors(final Label listOfLabelsToProcess) {
Label newListOfLabelsToProcess = listOfLabelsToProcess;
Edge outgoingEdge = outgoingEdges;
while (outgoingEdge != null) {
// By construction, the second outgoing edge of a basic block that ends with a jsr instruction
// leads to the jsr target (see {@link #FLAG_SUBROUTINE_CALLER}).
boolean isJsrTarget =
(flags & Label.FLAG_SUBROUTINE_CALLER) != 0 && outgoingEdge == outgoingEdges.nextEdge;
if (!isJsrTarget && outgoingEdge.successor.nextListElement == null) {
// Add this successor to the list of blocks to process, if it does not already belong to a
// list of labels.
outgoingEdge.successor.nextListElement = newListOfLabelsToProcess;
newListOfLabelsToProcess = outgoingEdge.successor;
}
outgoingEdge = outgoingEdge.nextEdge;
}
return newListOfLabelsToProcess;
}
// -----------------------------------------------------------------------------------------------
// Overridden Object methods
// -----------------------------------------------------------------------------------------------
/**
* Returns a string representation of this label.
*
* @return a string representation of this label.
*/
@Override
public String toString() {
return "L" + System.identityHashCode(this);
}
}