org.sosy_lab.cpachecker.cpa.value.refiner.ValueAnalysisPathInterpolator.java Source code

Java tutorial

Introduction

Here is the source code for org.sosy_lab.cpachecker.cpa.value.refiner.ValueAnalysisPathInterpolator.java

Source

/*
 *  CPAchecker is a tool for configurable software verification.
 *  This file is part of CPAchecker.
 *
 *  Copyright (C) 2007-2014  Dirk Beyer
 *  All rights reserved.
 *
 *  Licensed under the Apache License, Version 2.0 (the "License");
 *  you may not use this file except in compliance with the License.
 *  You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 *  Unless required by applicable law or agreed to in writing, software
 *  distributed under the License is distributed on an "AS IS" BASIS,
 *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 *  See the License for the specific language governing permissions and
 *  limitations under the License.
 *
 *
 *  CPAchecker web page:
 *    http://cpachecker.sosy-lab.org
 */
package org.sosy_lab.cpachecker.cpa.value.refiner;

import java.io.PrintStream;
import java.util.Collections;
import java.util.HashMap;
import java.util.HashSet;
import java.util.LinkedHashMap;
import java.util.List;
import java.util.Map;
import java.util.Set;

import org.sosy_lab.common.Pair;
import org.sosy_lab.common.configuration.Configuration;
import org.sosy_lab.common.configuration.InvalidConfigurationException;
import org.sosy_lab.common.configuration.Option;
import org.sosy_lab.common.configuration.Options;
import org.sosy_lab.common.log.LogManager;
import org.sosy_lab.cpachecker.cfa.CFA;
import org.sosy_lab.cpachecker.cfa.model.CFAEdge;
import org.sosy_lab.cpachecker.cfa.model.CFANode;
import org.sosy_lab.cpachecker.cfa.types.Type;
import org.sosy_lab.cpachecker.core.CPAcheckerResult.Result;
import org.sosy_lab.cpachecker.core.ShutdownNotifier;
import org.sosy_lab.cpachecker.core.interfaces.Statistics;
import org.sosy_lab.cpachecker.core.reachedset.ReachedSet;
import org.sosy_lab.cpachecker.cpa.arg.ARGPath;
import org.sosy_lab.cpachecker.cpa.arg.ARGPath.PathIterator;
import org.sosy_lab.cpachecker.cpa.arg.ARGState;
import org.sosy_lab.cpachecker.cpa.arg.MutableARGPath;
import org.sosy_lab.cpachecker.cpa.conditions.path.AssignmentsInPathCondition.UniqueAssignmentsInPathConditionState;
import org.sosy_lab.cpachecker.cpa.value.ValueAnalysisState.MemoryLocation;
import org.sosy_lab.cpachecker.cpa.value.refiner.utils.AssumptionUseDefinitionCollector;
import org.sosy_lab.cpachecker.cpa.value.refiner.utils.ErrorPathClassifier;
import org.sosy_lab.cpachecker.cpa.value.refiner.utils.ErrorPathClassifier.ErrorPathPrefixPreference;
import org.sosy_lab.cpachecker.cpa.value.refiner.utils.InitialAssumptionUseDefinitionCollector;
import org.sosy_lab.cpachecker.cpa.value.refiner.utils.ValueAnalysisEdgeInterpolator;
import org.sosy_lab.cpachecker.cpa.value.refiner.utils.ValueAnalysisFeasibilityChecker;
import org.sosy_lab.cpachecker.cpa.value.type.Value;
import org.sosy_lab.cpachecker.cpa.value.type.Value.UnknownValue;
import org.sosy_lab.cpachecker.exceptions.CPAException;
import org.sosy_lab.cpachecker.exceptions.RefinementFailedException;
import org.sosy_lab.cpachecker.exceptions.RefinementFailedException.Reason;
import org.sosy_lab.cpachecker.util.AbstractStates;
import org.sosy_lab.cpachecker.util.statistics.StatCounter;
import org.sosy_lab.cpachecker.util.statistics.StatInt;
import org.sosy_lab.cpachecker.util.statistics.StatKind;
import org.sosy_lab.cpachecker.util.statistics.StatTimer;
import org.sosy_lab.cpachecker.util.statistics.StatisticsWriter;

import com.google.common.collect.HashMultimap;
import com.google.common.collect.Iterables;
import com.google.common.collect.Multimap;

@Options(prefix = "cpa.value.refiner")
public class ValueAnalysisPathInterpolator implements Statistics {
    /**
     * whether or not to do lazy-abstraction, i.e., when true, the re-starting node
     * for the re-exploration of the ARG will be the node closest to the root
     * where new information is made available through the current refinement
     */
    @Option(secure = true, description = "whether or not to do lazy-abstraction")
    private boolean doLazyAbstraction = true;

    @Option(secure = true, description = "whether to perform (more precise) edge-based interpolation or (more efficient) path-based interpolation")
    private boolean performEdgeBasedInterpolation = true;

    @Option(secure = true, description = "which prefix of an actual counterexample trace should be used for interpolation")
    private ErrorPathPrefixPreference prefixPreference = ErrorPathPrefixPreference.DOMAIN_BEST_SHALLOW;

    /**
     * the offset in the path from where to cut-off the subtree, and restart the analysis
     */
    private int interpolationOffset = -1;

    /**
     * a reference to the assignment-counting state, to make the precision increment aware of thresholds
     */
    private UniqueAssignmentsInPathConditionState assignments = null;

    // statistics
    private StatCounter totalInterpolations = new StatCounter("Number of interpolations");
    private StatInt totalInterpolationQueries = new StatInt(StatKind.SUM, "Number of interpolation queries");
    private StatInt sizeOfInterpolant = new StatInt(StatKind.AVG, "Size of interpolant");
    private StatTimer timerInterpolation = new StatTimer("Time for interpolation");

    private final CFA cfa;
    private final LogManager logger;
    private final ShutdownNotifier shutdownNotifier;
    private final Configuration config;

    private final ValueAnalysisEdgeInterpolator interpolator;

    public ValueAnalysisPathInterpolator(Configuration pConfig, final LogManager pLogger,
            final ShutdownNotifier pShutdownNotifier, final CFA pCfa) throws InvalidConfigurationException {
        pConfig.inject(this);
        config = pConfig;

        logger = pLogger;
        cfa = pCfa;
        shutdownNotifier = pShutdownNotifier;
        interpolator = new ValueAnalysisEdgeInterpolator(pConfig, logger, shutdownNotifier, cfa);
    }

    protected Map<ARGState, ValueAnalysisInterpolant> performInterpolation(ARGPath errorPath,
            ValueAnalysisInterpolant interpolant) throws CPAException, InterruptedException {
        totalInterpolations.inc();
        timerInterpolation.start();

        interpolationOffset = -1;

        ARGPath errorPathPrefix = obtainErrorPathPrefix(errorPath, interpolant);

        Map<ARGState, ValueAnalysisInterpolant> pathInterpolants = (performEdgeBasedInterpolation)
                ? performEdgeBasedInterpolation(errorPathPrefix, interpolant)
                : performPathBasedInterpolation(errorPathPrefix);

        timerInterpolation.stop();
        return pathInterpolants;
    }

    /**
     * This method performs interpolation on each edge of the path, using the
     * {@link ValueAnalysisEdgeInterpolator}.
     *
     * @param errorPathPrefix the error path prefix to interpolate
     * @param interpolant an initial interpolant (only non-trivial when interpolating error path suffixes in global refinement)
     * @return the mapping of {@link ARGState}s to {@link ValueAnalysisInterpolant}s
     * @throws InterruptedException
     * @throws CPAException
     */
    private Map<ARGState, ValueAnalysisInterpolant> performEdgeBasedInterpolation(ARGPath errorPathPrefix,
            ValueAnalysisInterpolant interpolant) throws InterruptedException, CPAException {

        // obtain use-def relation, containing variables relevant to the "failing" assumption
        Set<MemoryLocation> useDefRelation = new HashSet<>();
        /* TODO: does not work as long as AssumptionUseDefinitionCollector is incomplete (e.g., does not take structs into account)
        if (prefixPreference != ErrorPathPrefixPreference.DEFAULT) {
          AssumptionUseDefinitionCollector useDefinitionCollector = new InitialAssumptionUseDefinitionCollector();
          useDefRelation = from(useDefinitionCollector.obtainUseDefInformation(errorTrace)).
              transform(MemoryLocation.FROM_STRING_TO_MEMORYLOCATION).toSet();
        }*/

        Map<ARGState, ValueAnalysisInterpolant> pathInterpolants = new LinkedHashMap<>(errorPathPrefix.size());

        PathIterator pathIterator = errorPathPrefix.pathIterator();
        while (pathIterator.hasNext()) {
            shutdownNotifier.shutdownIfNecessary();

            // interpolate at each edge as long the previous interpolant is not false
            if (!interpolant.isFalse()) {
                interpolant = interpolator.deriveInterpolant(errorPathPrefix, pathIterator.getOutgoingEdge(),
                        pathIterator.getIndex(), interpolant, useDefRelation);
            }

            totalInterpolationQueries.setNextValue(interpolator.getNumberOfInterpolationQueries());

            if (!interpolant.isTrivial() && interpolationOffset == -1) {
                interpolationOffset = pathIterator.getIndex();
            }

            sizeOfInterpolant.setNextValue(interpolant.getSize());

            pathIterator.advance();

            pathInterpolants.put(pathIterator.getAbstractState(), interpolant);

            if (!pathIterator.hasNext()) {
                assert interpolant.isFalse() : "final interpolant is not false";
            }
        }

        return pathInterpolants;
    }

    /**
     * This method performs interpolation on the complete path, based on the
     * use-def-relation obtained by {@link AssumptionUseDefinitionCollector} or
     * its subclass. It creates fake interpolants that are not inductive.
     *
     * @param errorPathPrefix the error path prefix to interpolate
     * @return
     */
    private Map<ARGState, ValueAnalysisInterpolant> performPathBasedInterpolation(ARGPath errorPathPrefix) {

        assert (errorPathPrefix.getFirstState().getParents()
                .isEmpty()) : "static interpolation requires cpa.value.refinement.useTopDownInterpolationStrategy to be set to 'true'";

        AssumptionUseDefinitionCollector useDefinitionCollector = prefixPreference == ErrorPathPrefixPreference.DEFAULT
                ? new AssumptionUseDefinitionCollector()
                : new InitialAssumptionUseDefinitionCollector();

        Set<String> useDefRelation = useDefinitionCollector.obtainUseDefInformation(errorPathPrefix.asEdgesList());

        totalInterpolationQueries.setNextValue(1);
        sizeOfInterpolant.setNextValue(useDefRelation.size() * errorPathPrefix.size());

        Map<ARGState, ValueAnalysisInterpolant> pathInterpolants = new LinkedHashMap<>(errorPathPrefix.size());

        // add the "fake" interpolant for each state except the root of the ARG;
        // this makes the first child of the root the refinement root
        ValueAnalysisInterpolant fakeItp = createFakeInterpolant(useDefRelation);
        for (ARGState state : Iterables.skip(errorPathPrefix.asStatesList(), 1)) {
            pathInterpolants.put(state, fakeItp);
        }

        return pathInterpolants;
    }

    /**
     * This method creates a "fake" interpolant from a set of relevant variables.
     *
     * @param relevantVariables
     * @return the "fake" interpolant
     */
    private ValueAnalysisInterpolant createFakeInterpolant(Set<String> relevantVariables) {
        HashMap<MemoryLocation, Value> values = new HashMap<>();
        for (String relevantVariable : relevantVariables) {
            values.put(MemoryLocation.valueOf(relevantVariable), UnknownValue.getInstance());
        }

        return new ValueAnalysisInterpolant(values, Collections.<MemoryLocation, Type>emptyMap());
    }

    public Multimap<CFANode, MemoryLocation> determinePrecisionIncrement(MutableARGPath errorPath)
            throws CPAException, InterruptedException {

        assignments = AbstractStates.extractStateByType(errorPath.getLast().getFirst(),
                UniqueAssignmentsInPathConditionState.class);

        Multimap<CFANode, MemoryLocation> increment = HashMultimap.create();

        Map<ARGState, ValueAnalysisInterpolant> itps = performInterpolation(errorPath.immutableCopy(),
                ValueAnalysisInterpolant.createInitial());

        for (Map.Entry<ARGState, ValueAnalysisInterpolant> itp : itps.entrySet()) {
            addToPrecisionIncrement(increment, AbstractStates.extractLocation(itp.getKey()), itp.getValue());
        }

        return increment;
    }

    /**
     * This method adds the given variable at the given location to the increment.
     *
     * @param increment the current increment
     * @param currentNode the current node for which to add a new variable
     * @param memoryLocation the name of the variable to add to the increment at the given edge
     */
    private void addToPrecisionIncrement(Multimap<CFANode, MemoryLocation> increment, CFANode currentNode,
            ValueAnalysisInterpolant itp) {
        for (MemoryLocation memoryLocation : itp.getMemoryLocations()) {
            if (assignments == null || !assignments.exceedsHardThreshold(memoryLocation)) {
                increment.put(currentNode, memoryLocation);
            }
        }
    }

    /**
     * This method determines the new refinement root.
     *
     * @param errorPath the error path from where to determine the refinement root
     * @param increment the current precision increment
     * @param isRepeatedRefinement the flag to determine whether or not this is a repeated refinement
     * @return the new refinement root
     * @throws RefinementFailedException if no refinement root can be determined
     */
    public Pair<ARGState, CFAEdge> determineRefinementRoot(MutableARGPath errorPath,
            Multimap<CFANode, MemoryLocation> increment, boolean isRepeatedRefinement)
            throws RefinementFailedException {

        if (interpolationOffset == -1) {
            throw new RefinementFailedException(Reason.InterpolationFailed, errorPath.immutableCopy());
        }

        // if doing lazy abstraction, use the node closest to the root node where new information is present
        if (doLazyAbstraction) {
            return errorPath.get(interpolationOffset);
        }

        // otherwise, just use the successor of the root node
        else {
            return errorPath.get(1);
        }
    }

    /**
     * This path obtains a (sub)path of the error path which is given to the interpolation procedure.
     *
     * @param errorPath the original error path
     * @param interpolant the initial interpolant, i.e. the initial state, with which to check the error path.
     * @return a (sub)path of the error path which is given to the interpolation procedure
     * @throws CPAException
     * @throws InterruptedException
     */
    private ARGPath obtainErrorPathPrefix(ARGPath errorPath, ValueAnalysisInterpolant interpolant)
            throws CPAException, InterruptedException {

        try {
            ValueAnalysisFeasibilityChecker checker = new ValueAnalysisFeasibilityChecker(logger, cfa, config);
            List<ARGPath> prefixes = checker.getInfeasilbePrefixes(errorPath,
                    interpolant.createValueAnalysisState());

            ErrorPathClassifier classifier = new ErrorPathClassifier(cfa.getVarClassification(),
                    cfa.getLoopStructure());
            errorPath = classifier.obtainPrefix(prefixPreference, errorPath, prefixes);

        } catch (InvalidConfigurationException e) {
            throw new CPAException("Configuring ValueAnalysisFeasibilityChecker failed: " + e.getMessage(), e);
        }

        return errorPath;
    }

    @Override
    public String getName() {
        return "ValueAnalysisInterpolationBasedRefiner";
    }

    @Override
    public void printStatistics(PrintStream out, Result result, ReachedSet reached) {
        StatisticsWriter writer = StatisticsWriter.writingStatisticsTo(out).beginLevel();
        writer.put(totalInterpolations);
        writer.put(totalInterpolationQueries);
        writer.put(sizeOfInterpolant);
        writer.put(timerInterpolation);
    }

    public int getInterpolationOffset() {
        return interpolationOffset;
    }
}