xref: /frameworks/base/core/java/com/android/internal/os/KernelCpuThreadReader.java

/*
 * Copyright (C) 2018 The Android Open Source Project
 *
 * 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.
 */

package com.android.internal.os;

import android.annotation.Nullable;
import android.os.Process;
import android.util.IntArray;
import android.util.Slog;

import com.android.internal.annotations.VisibleForTesting;
import com.android.internal.util.Preconditions;

import java.io.IOException;
import java.nio.file.DirectoryIteratorException;
import java.nio.file.DirectoryStream;
import java.nio.file.Files;
import java.nio.file.Path;
import java.nio.file.Paths;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.function.Predicate;

/**
 * Iterates over processes, and all threads owned by those processes, and return the CPU usage for
 * each thread. The CPU usage statistics contain the amount of time spent in a frequency band. CPU
 * usage is collected using {@link ProcTimeInStateReader}.
 *
 * <p>We only collect CPU data for processes and threads that are owned by certain UIDs. These UIDs
 * are configured via {@link #setUidPredicate}.
 *
 * <p>Frequencies are bucketed together to reduce the amount of data created. This means that we
 * return less frequencies than provided by {@link ProcTimeInStateReader}. The number of frequencies
 * is configurable by {@link #setNumBuckets}. Frequencies are reported as the lowest frequency in
 * that range. Frequencies are spread as evenly as possible across the buckets. The buckets do not
 * cross over the little/big frequencies reported.
 *
 * <p>N.B.: In order to bucket across little/big frequencies correctly, we assume that the {@code
 * time_in_state} file contains every little core frequency in ascending order, followed by every
 * big core frequency in ascending order. This assumption might not hold for devices with different
 * kernel implementations of the {@code time_in_state} file generation.
 */
public class KernelCpuThreadReader {

    private static final String TAG = "KernelCpuThreadReader";

    private static final boolean DEBUG = false;

    /**
     * The name of the file to read CPU statistics from, must be found in {@code
     * /proc/$PID/task/$TID}
     */
    private static final String CPU_STATISTICS_FILENAME = "time_in_state";

    /**
     * The name of the file to read process command line invocation from, must be found in {@code
     * /proc/$PID/}
     */
    private static final String PROCESS_NAME_FILENAME = "cmdline";

    /**
     * The name of the file to read thread name from, must be found in {@code /proc/$PID/task/$TID}
     */
    private static final String THREAD_NAME_FILENAME = "comm";

    /** Glob pattern for the process directory names under {@code proc} */
    private static final String PROCESS_DIRECTORY_FILTER = "[0-9]*";

    /** Default process name when the name can't be read */
    private static final String DEFAULT_PROCESS_NAME = "unknown_process";

    /** Default thread name when the name can't be read */
    private static final String DEFAULT_THREAD_NAME = "unknown_thread";

    /** Default mount location of the {@code proc} filesystem */
    private static final Path DEFAULT_PROC_PATH = Paths.get("/proc");

    /** The initial {@code time_in_state} file for {@link ProcTimeInStateReader} */
    private static final Path DEFAULT_INITIAL_TIME_IN_STATE_PATH =
            DEFAULT_PROC_PATH.resolve("self/time_in_state");

    /** Value returned when there was an error getting an integer ID value (e.g. PID, UID) */
    private static final int ID_ERROR = -1;

    /**
     * When checking whether to report data for a thread, we check the UID of the thread's owner
     * against this predicate
     */
    private Predicate<Integer> mUidPredicate;

    /** Where the proc filesystem is mounted */
    private final Path mProcPath;

    /**
     * Frequencies read from the {@code time_in_state} file. Read from {@link
     * #mProcTimeInStateReader#getCpuFrequenciesKhz()} and cast to {@code int[]}
     */
    private int[] mFrequenciesKhz;

    /** Used to read and parse {@code time_in_state} files */
    private final ProcTimeInStateReader mProcTimeInStateReader;

    /** Used to sort frequencies and usage times into buckets */
    private FrequencyBucketCreator mFrequencyBucketCreator;

    private final Injector mInjector;

    /**
     * Create with a path where `proc` is mounted. Used primarily for testing
     *
     * @param procPath where `proc` is mounted (to find, see {@code mount | grep ^proc})
     * @param initialTimeInStatePath where the initial {@code time_in_state} file exists to define
     *     format
     */
    @VisibleForTesting
    public KernelCpuThreadReader(
            int numBuckets,
            Predicate<Integer> uidPredicate,
            Path procPath,
            Path initialTimeInStatePath,
            Injector injector)
            throws IOException {
        mUidPredicate = uidPredicate;
        mProcPath = procPath;
        mProcTimeInStateReader = new ProcTimeInStateReader(initialTimeInStatePath);
        mInjector = injector;
        setNumBuckets(numBuckets);
    }

    /**
     * Create the reader and handle exceptions during creation
     *
     * @return the reader, null if an exception was thrown during creation
     */
    @Nullable
    public static KernelCpuThreadReader create(int numBuckets, Predicate<Integer> uidPredicate) {
        try {
            return new KernelCpuThreadReader(
                    numBuckets,
                    uidPredicate,
                    DEFAULT_PROC_PATH,
                    DEFAULT_INITIAL_TIME_IN_STATE_PATH,
                    new Injector());
        } catch (IOException e) {
            Slog.e(TAG, "Failed to initialize KernelCpuThreadReader", e);
            return null;
        }
    }

    /**
     * Get the per-thread CPU usage of all processes belonging to a set of UIDs
     *
     * <p>This function will crawl through all process {@code proc} directories found by the pattern
     * {@code /proc/[0-9]*}, and then check the UID using {@code /proc/$PID/status}. This takes
     * approximately 500ms on a 2017 device. Therefore, this method can be computationally
     * expensive, and should not be called more than once an hour.
     *
     * <p>Data is only collected for UIDs passing the predicate supplied in {@link
     * #setUidPredicate}.
     */
    @Nullable
    public ArrayList<ProcessCpuUsage> getProcessCpuUsage() {
        if (DEBUG) {
            Slog.d(TAG, "Reading CPU thread usages for processes owned by UIDs");
        }

        final ArrayList<ProcessCpuUsage> processCpuUsages = new ArrayList<>();

        try (DirectoryStream<Path> processPaths =
                Files.newDirectoryStream(mProcPath, PROCESS_DIRECTORY_FILTER)) {
            for (Path processPath : processPaths) {
                final int processId = getProcessId(processPath);
                final int uid = mInjector.getUidForPid(processId);
                if (uid == ID_ERROR || processId == ID_ERROR) {
                    continue;
                }
                if (!mUidPredicate.test(uid)) {
                    continue;
                }

                final ProcessCpuUsage processCpuUsage =
                        getProcessCpuUsage(processPath, processId, uid);
                if (processCpuUsage != null) {
                    processCpuUsages.add(processCpuUsage);
                }
            }
        } catch (IOException e) {
            Slog.w(TAG, "Failed to iterate over process paths", e);
            return null;
        }

        if (processCpuUsages.isEmpty()) {
            Slog.w(TAG, "Didn't successfully get any process CPU information for UIDs specified");
            return null;
        }

        if (DEBUG) {
            Slog.d(TAG, "Read usage for " + processCpuUsages.size() + " processes");
        }

        return processCpuUsages;
    }

    /**
     * Get the CPU frequencies that correspond to the times reported in {@link
     * ThreadCpuUsage#usageTimesMillis}
     */
    @Nullable
    public int[] getCpuFrequenciesKhz() {
        return mFrequenciesKhz;
    }

    /** Set the number of frequency buckets to use */
    void setNumBuckets(int numBuckets) {
        // If `numBuckets` hasn't changed since the last set, do nothing
        if (mFrequenciesKhz != null && mFrequenciesKhz.length == numBuckets) {
            return;
        }

        final long[] frequenciesKhz = mProcTimeInStateReader.getFrequenciesKhz();
        if (numBuckets != 0) {
            mFrequencyBucketCreator = new FrequencyBucketCreator(frequenciesKhz, numBuckets);
            mFrequenciesKhz = mFrequencyBucketCreator.bucketFrequencies(frequenciesKhz);
        } else {
            mFrequencyBucketCreator = null;
            mFrequenciesKhz = new int[frequenciesKhz.length];
            for (int i = 0; i < frequenciesKhz.length; i++) {
                mFrequenciesKhz[i] = (int) frequenciesKhz[i];
            }
        }
    }

    /** Set the UID predicate for {@link #getProcessCpuUsage} */
    @VisibleForTesting
    public void setUidPredicate(Predicate<Integer> uidPredicate) {
        mUidPredicate = uidPredicate;
    }

    /**
     * Read all of the CPU usage statistics for each child thread of a process
     *
     * @param processPath the {@code /proc} path of the thread
     * @param processId the ID of the process
     * @param uid the ID of the user who owns the process
     * @return process CPU usage containing usage of all child threads. Null if the process exited
     *     and its {@code proc} directory was removed while collecting information
     */
    @Nullable
    private ProcessCpuUsage getProcessCpuUsage(Path processPath, int processId, int uid) {
        if (DEBUG) {
            Slog.d(
                    TAG,
                    "Reading CPU thread usages with directory "
                            + processPath
                            + " process ID "
                            + processId
                            + " and user ID "
                            + uid);
        }

        final Path allThreadsPath = processPath.resolve("task");
        final ArrayList<ThreadCpuUsage> threadCpuUsages = new ArrayList<>();
        try (DirectoryStream<Path> threadPaths = Files.newDirectoryStream(allThreadsPath)) {
            for (Path threadDirectory : threadPaths) {
                ThreadCpuUsage threadCpuUsage = getThreadCpuUsage(threadDirectory);
                if (threadCpuUsage == null) {
                    continue;
                }
                threadCpuUsages.add(threadCpuUsage);
            }
        } catch (IOException | DirectoryIteratorException e) {
            // Expected when a process finishes
            return null;
        }

        // If we found no threads, then the process has exited while we were reading from it
        if (threadCpuUsages.isEmpty()) {
            return null;
        }
        if (DEBUG) {
            Slog.d(TAG, "Read CPU usage of " + threadCpuUsages.size() + " threads");
        }
        return new ProcessCpuUsage(processId, getProcessName(processPath), uid, threadCpuUsages);
    }

    /**
     * Get a thread's CPU usage
     *
     * @param threadDirectory the {@code /proc} directory of the thread
     * @return thread CPU usage. Null if the thread exited and its {@code proc} directory was
     *     removed while collecting information
     */
    @Nullable
    private ThreadCpuUsage getThreadCpuUsage(Path threadDirectory) {
        // Get the thread ID from the directory name
        final int threadId;
        try {
            final String directoryName = threadDirectory.getFileName().toString();
            threadId = Integer.parseInt(directoryName);
        } catch (NumberFormatException e) {
            Slog.w(TAG, "Failed to parse thread ID when iterating over /proc/*/task", e);
            return null;
        }

        // Get the thread name from the thread directory
        final String threadName = getThreadName(threadDirectory);

        // Get the CPU statistics from the directory
        final Path threadCpuStatPath = threadDirectory.resolve(CPU_STATISTICS_FILENAME);
        final long[] cpuUsagesLong = mProcTimeInStateReader.getUsageTimesMillis(threadCpuStatPath);
        if (cpuUsagesLong == null) {
            return null;
        }
        final int[] cpuUsages;
        if (mFrequencyBucketCreator != null) {
            cpuUsages = mFrequencyBucketCreator.bucketValues(cpuUsagesLong);
        } else {
            cpuUsages = new int[cpuUsagesLong.length];
            for (int i = 0; i < cpuUsagesLong.length; i++) {
                cpuUsages[i] = (int) cpuUsagesLong[i];
            }
        }
        return new ThreadCpuUsage(threadId, threadName, cpuUsages);
    }

    /** Get the command used to start a process */
    private String getProcessName(Path processPath) {
        final Path processNamePath = processPath.resolve(PROCESS_NAME_FILENAME);

        final String processName = ProcStatsUtil.readSingleLineProcFile(processNamePath.toString());
        if (processName != null) {
            return processName;
        }
        return DEFAULT_PROCESS_NAME;
    }

    /** Get the name of a thread, given the {@code /proc} path of the thread */
    private String getThreadName(Path threadPath) {
        final Path threadNamePath = threadPath.resolve(THREAD_NAME_FILENAME);
        final String threadName = ProcStatsUtil.readNullSeparatedFile(threadNamePath.toString());
        if (threadName == null) {
            return DEFAULT_THREAD_NAME;
        }
        return threadName;
    }

    /**
     * Get the ID of a process from its path
     *
     * @param processPath {@code proc} path of the process
     * @return the ID, {@link #ID_ERROR} if the path could not be parsed
     */
    private int getProcessId(Path processPath) {
        String fileName = processPath.getFileName().toString();
        try {
            return Integer.parseInt(fileName);
        } catch (NumberFormatException e) {
            Slog.w(TAG, "Failed to parse " + fileName + " as process ID", e);
            return ID_ERROR;
        }
    }

    /**
     * Quantizes a list of N frequencies into a list of M frequencies (where M<=N)
     *
     * <p>In order to reduce data sent from the device, we discard precise frequency information for
     * an approximation. This is done by putting groups of adjacent frequencies into the same
     * bucket, and then reporting that bucket under the minimum frequency in that bucket.
     *
     * <p>Many devices have multiple core clusters. We do not want to report frequencies from
     * different clusters under the same bucket, so some complication arises.
     *
     * <p>Buckets are allocated evenly across all core clusters, i.e. they all have the same number
     * of buckets regardless of how many frequencies they contain. This is done to reduce code
     * complexity, and in practice the number of frequencies doesn't vary too much between core
     * clusters.
     *
     * <p>If the number of buckets is not a factor of the number of frequencies, the remainder of
     * the frequencies are placed into the last bucket.
     *
     * <p>It is possible to have less buckets than asked for, so any calling code can't assume that
     * initializing with N buckets will use return N values. This happens in two scenarios:
     *
     * <ul>
     *   <li>There are less frequencies available than buckets asked for.
     *   <li>There are less frequencies in a core cluster than buckets allocated to that core
     *       cluster.
     * </ul>
     */
    @VisibleForTesting
    public static class FrequencyBucketCreator {
        private final int mNumFrequencies;
        private final int mNumBuckets;
        private final int[] mBucketStartIndices;

        @VisibleForTesting
        public FrequencyBucketCreator(long[] frequencies, int targetNumBuckets) {
            mNumFrequencies = frequencies.length;
            int[] clusterStartIndices = getClusterStartIndices(frequencies);
            mBucketStartIndices =
                    getBucketStartIndices(clusterStartIndices, targetNumBuckets, mNumFrequencies);
            mNumBuckets = mBucketStartIndices.length;
        }

        /**
         * Put an array of values into buckets. This takes a {@code long[]} and returns {@code
         * int[]} as everywhere this method is used will have to do the conversion anyway, so we
         * save time by doing it here instead
         *
         * @param values the values to bucket
         * @return the bucketed usage times
         */
        @VisibleForTesting
        public int[] bucketValues(long[] values) {
            Preconditions.checkArgument(values.length == mNumFrequencies);
            int[] buckets = new int[mNumBuckets];
            for (int bucketIdx = 0; bucketIdx < mNumBuckets; bucketIdx++) {
                final int bucketStartIdx = getLowerBound(bucketIdx, mBucketStartIndices);
                final int bucketEndIdx =
                        getUpperBound(bucketIdx, mBucketStartIndices, values.length);
                for (int valuesIdx = bucketStartIdx; valuesIdx < bucketEndIdx; valuesIdx++) {
                    buckets[bucketIdx] += values[valuesIdx];
                }
            }
            return buckets;
        }

        /** Get the minimum frequency in each bucket */
        @VisibleForTesting
        public int[] bucketFrequencies(long[] frequencies) {
            Preconditions.checkArgument(frequencies.length == mNumFrequencies);
            int[] buckets = new int[mNumBuckets];
            for (int i = 0; i < buckets.length; i++) {
                buckets[i] = (int) frequencies[mBucketStartIndices[i]];
            }
            return buckets;
        }

        /**
         * Get the index in frequencies where each core cluster starts
         *
         * <p>The frequencies for each cluster are given in ascending order, appended to each other.
         * This means that every time there is a decrease in frequencies (instead of increase) a new
         * cluster has started.
         */
        private static int[] getClusterStartIndices(long[] frequencies) {
            IntArray indices = new IntArray();
            indices.add(0);
            for (int i = 0; i < frequencies.length - 1; i++) {
                if (frequencies[i] >= frequencies[i + 1]) {
                    indices.add(i + 1);
                }
            }
            return indices.toArray();
        }

        /** Get the index in frequencies where each bucket starts */
        private static int[] getBucketStartIndices(
                int[] clusterStartIndices, int targetNumBuckets, int numFrequencies) {
            int numClusters = clusterStartIndices.length;

            // If we haven't got enough buckets for every cluster, we instead have one bucket per
            // cluster, with the last bucket containing the remaining clusters
            if (numClusters > targetNumBuckets) {
                return Arrays.copyOfRange(clusterStartIndices, 0, targetNumBuckets);
            }

            IntArray bucketStartIndices = new IntArray();
            for (int clusterIdx = 0; clusterIdx < numClusters; clusterIdx++) {
                final int clusterStartIdx = getLowerBound(clusterIdx, clusterStartIndices);
                final int clusterEndIdx =
                        getUpperBound(clusterIdx, clusterStartIndices, numFrequencies);

                final int numBucketsInCluster;
                if (clusterIdx != numClusters - 1) {
                    numBucketsInCluster = targetNumBuckets / numClusters;
                } else {
                    // If we're in the last cluster, the bucket will contain the remainder of the
                    // frequencies
                    int previousBucketsInCluster = targetNumBuckets / numClusters;
                    numBucketsInCluster =
                            targetNumBuckets - (previousBucketsInCluster * (numClusters - 1));
                }

                final int numFrequenciesInCluster = clusterEndIdx - clusterStartIdx;
                // If there are less frequencies than buckets in a cluster, we have one bucket per
                // frequency, and do not use the remaining buckets
                final int numFrequenciesInBucket =
                        Math.max(1, numFrequenciesInCluster / numBucketsInCluster);
                for (int bucketIdx = 0; bucketIdx < numBucketsInCluster; bucketIdx++) {
                    int bucketStartIdx = clusterStartIdx + bucketIdx * numFrequenciesInBucket;
                    // If we've gone over the end index, ignore the rest of the buckets for this
                    // cluster
                    if (bucketStartIdx >= clusterEndIdx) {
                        break;
                    }
                    bucketStartIndices.add(bucketStartIdx);
                }
            }
            return bucketStartIndices.toArray();
        }

        private static int getLowerBound(int index, int[] startIndices) {
            return startIndices[index];
        }

        private static int getUpperBound(int index, int[] startIndices, int max) {
            if (index != startIndices.length - 1) {
                return startIndices[index + 1];
            } else {
                return max;
            }
        }
    }

    /** CPU usage of a process */
    public static class ProcessCpuUsage {
        public final int processId;
        public final String processName;
        public final int uid;
        public ArrayList<ThreadCpuUsage> threadCpuUsages;

        @VisibleForTesting
        public ProcessCpuUsage(
                int processId,
                String processName,
                int uid,
                ArrayList<ThreadCpuUsage> threadCpuUsages) {
            this.processId = processId;
            this.processName = processName;
            this.uid = uid;
            this.threadCpuUsages = threadCpuUsages;
        }
    }

    /** CPU usage of a thread */
    public static class ThreadCpuUsage {
        public final int threadId;
        public final String threadName;
        public int[] usageTimesMillis;

        @VisibleForTesting
        public ThreadCpuUsage(int threadId, String threadName, int[] usageTimesMillis) {
            this.threadId = threadId;
            this.threadName = threadName;
            this.usageTimesMillis = usageTimesMillis;
        }
    }

    /** Used to inject static methods from {@link Process} */
    @VisibleForTesting
    public static class Injector {
        /** Get the UID for the process with ID {@code pid} */
        public int getUidForPid(int pid) {
            return Process.getUidForPid(pid);
        }
    }
}