src/main/scala/midterm22/Part2.scala

+package midterm22
+
+import scala.annotation.nowarn
+
+// Questions 4-7
+
+// See tests in midterm22.Part2Test.
+// Run with `sbt testOnly midterm22.Part2Test`.
+
+/*
+Answers to the exam questions:
+  When called with a Vector:
+    The total amount of work is O(n), as it is dominated by the time needed to
+    read the array. More precisely W(n) = c + 2*W(n/2) = O(n).
+
+    The depth is O(log(n)), because every recursion takes constant time
+    and we divide the size of the input by 2 every time, i.e. D(n) = c + D(n/2) = O(log(n)).
+
+    Note however that in practice it is often still faster to manipulate
+    start and end indices rather than using take and drop.
+
+  When called with a List:
+    Every recursion takes up to time O(n) rather than constant time.
+
+    The total amount of work is O(n) times the number of recursion, because
+    take and drop takes time O(n) on lists. Precisely, W(n) = n + 2*W(n/2) = O(log(n)*n)
+
+    The depth is computed similarly: D(n) = n + D(n/2) = O(n), i.e.
+
+Note: these are theoretical results. In practice, you should always double-check
+that kind of conclusions with benchmarks. We did so in
+`midterm-code/src/main/scala/bench`. Results are available in `bench-results`.
+From these results, we can conclude that
+1. Vectors are indeed faster in this case, and
+2. parallelization of `contains` yields a 2x speedup.
+ */
+@nowarn
+def contains[A](l: Iterable[A], elem: A): Boolean =
+  val n = l.size
+  if n <= 5 then
+    for i <- l do
+      if i == elem then
+        return true
+    false
+  else
+    val (p0, p1) = parallel(
+      contains(l.take(n / 2), elem),
+      contains(l.drop(n / 2), elem)
+    )
+    p0 || p1

src/main/scala/midterm22/Part3.scala

+package midterm22
+
+// Question 8
+
+// Run with `sbt runMain midterm22.part3`
+
+@main def part3() =
+  def thread(b: => Unit) =
+    val t = new Thread:
+      override def run() = b
+    t
+  val t = thread { println(s"Hello World") }
+  t.join()
+  println(s"Hello")

src/main/scala/midterm22/Part4.scala

+package midterm22
+
+import java.util.concurrent.atomic.AtomicInteger
+import instrumentation.*
+
+import instrumentation.Monitor
+// Questions 9-15
+
+// See tests in midterm22.Part4Test.
+// Run with `sbt testOnly midterm22.Part4Test`.
+
+class Node(
+    // Globally unique identifier. Different for each instance.
+    val guid: Int
+) extends Monitor
+
+// This function might be called concurrently.
+def lockFun(nodes: List[Node], fn: (e: Int) => Unit): Unit =
+  if nodes.size > 0 then
+    nodes.head.synchronized {
+      fn(nodes(0).guid)
+      lockFun(nodes.tail, fn)
+    }

src/main/scala/midterm22/Part6.scala

+package midterm22
+
+import instrumentation.Monitor
+
+// Question 21
+
+// See tests in midterm22.Part6Test.
+// Run with `sbt testOnly midterm22.Part6Test`.
+
+class TicketsManager(totalTickets: Int) extends Monitor:
+  var remainingTickets = totalTickets
+
+  // This method might be called concurrently
+  def getTicket(): Boolean =
+    if remainingTickets > 0 then
+      this.synchronized {
+        remainingTickets -= 1
+      }
+      true
+    else false

src/main/scala/midterm22/Part7.scala

+package midterm22
+
+import instrumentation.Monitor
+
+// Questions 22-24
+
+// See tests in midterm22.Part7Test.
+// Run with `sbt testOnly midterm22.Part7Test`.
+
+class NIC(private val _index: Int, private var _assigned: Boolean)
+    extends Monitor:
+  def index = _index
+  def assigned = _assigned
+  def assigned_=(v: Boolean) = _assigned = v
+
+class NICManager(n: Int):
+  // Creates a list with n NICs
+  val nics = (for i <- 0 until n yield NIC(i, false)).toList
+
+  // This method might be called concurrently
+  def assignNICs(limitRecvNICs: Boolean = false): (Int, Int) =
+    var recvNIC: Int = 0
+    var sendNIC: Int = 0
+    var gotRecvNIC: Boolean = false
+    var gotSendNIC: Boolean = false
+
+    /// Obtaining receiving NIC...
+    while !gotRecvNIC do
+      nics(recvNIC).synchronized {
+        if !nics(recvNIC).assigned then
+          nics(recvNIC).assigned = true
+          gotRecvNIC = true
+        else recvNIC = (recvNIC + 1) % (if limitRecvNICs then n - 1 else n)
+      }
+    // Successfully obtained receiving NIC
+
+    // Obtaining sending NIC...
+    while !gotSendNIC do
+      nics(sendNIC).synchronized {
+        if !nics(sendNIC).assigned then
+          nics(sendNIC).assigned = true
+          gotSendNIC = true
+        else sendNIC = (sendNIC + 1) % n
+      }
+    // Successfully obtained sending NIC
+
+    return (recvNIC, sendNIC)

src/main/scala/midterm22/Part8.scala

+package midterm22
+
+import scala.collection.concurrent.{TrieMap, Map}
+import java.util.concurrent.atomic.AtomicInteger
+import scala.annotation.tailrec
+
+// Question 25
+
+// See tests in midterm22.Part8Test.
+// Run with `sbt testOnly midterm22.Part8Test`.
+
+// Represent a social network where user can follow each other. Each user is
+// represented by an id that is an `Int`.
+abstract class AbstractInstagram:
+  // The map storing the "following" relation of our social network.
+  // `graph(a)` contains the list of user ids that user `a` follows.
+  val graph: Map[Int, List[Int]] = new TrieMap[Int, List[Int]]()
+
+  // The maximum user id allocated until now. This value should be incremented
+  // by one each time a new user is added.
+  val maxId = new AtomicInteger(0)
+
+  // Allocates a new user and returns its unique id. Internally, this should
+  // also create an empty list at the corresponding id in `graph`. The
+  // implementation must be thread-safe.
+  def add(): Int
+
+  // Make `a` follow `b`. The implementation must be thread-safe.
+  def follow(a: Int, b: Int): Unit
+
+  // Makes `a` unfollow `b`. The implementation must be thread-safe.
+  def unfollow(a: Int, b: Int): Unit
+
+  // Removes user with id `a`. This should also remove all references to `a`
+  // in `graph`. The implementation must be thread-safe.
+  def remove(a: Int): Unit
+
+class Instagram extends AbstractInstagram:
+  // This method is worth 6 points.
+  def add(): Int =
+    // It is important to increment and read the value in one atomic step. See
+    // test `testParallelWrongAdd` for an alternative wrong implementation.
+    val id = maxId.incrementAndGet
+    // Note: it is also correct to use `graph.putIfAbsent`, but not needed as
+    // `id` is always new and therefore absent from the map at this point.
+    graph.update(id, Nil)
+    id
+
+  // This method is worth 8 points.
+  def remove(a: Int): Unit =
+    graph.remove(a)
+    // Iterate through all keys to make sure that nobody follows `a` anymore.
+    // For each key, we need to unfollow a in a thread-safe manner. Calling
+    // `unfollow` is the simplest way to so, as it is already guaranteed to be
+    // thread-safe. See test `testParallelWrongRemove` for an example of wrong
+    // implementation.
+    for b <- graph.keys do unfollow(b, a)
+
+  // This method is worth 10 points.
+  def unfollow(a: Int, b: Int) =
+    // Here, it is important to read the value only once. First calling
+    // `.contains(a)` and then `graph(a)` does not work, as `a` might be removed
+    // between the two calls. See `testParallelWrongUnfollow` for an example of
+    // this wrong implementation.
+    val prev = graph.get(a)
+    // Returns silently if `a` does not exist.
+    if prev.isEmpty then return
+    // We replace the list of users that `a` follows in an atomic manner. If the
+    // list of followed users changed concurrently, we start over.
+    if !graph.replace(a, prev.get, prev.get.filter(_ != b)) then unfollow(a, b)
+
+  // This method is worth 12 points.
+  def follow(a: Int, b: Int) =
+    val prev = graph.get(a)
+    // Returns silently if `a` does not exist.
+    if prev.isEmpty then return
+    // We replace the list of users that `a` follows in an atomic manner. If the
+    // list of followed users changed concurrently, we start over.
+    if !graph.replace(a, prev.get, b :: prev.get) then follow(a, b)
+    // Difficult: this handles the case where `b` is concurrently removed by
+    // another thread. To detect this case, we must check if `b` still exists
+    // after we have followed it, and unfollow it if it is not the case. See
+    // test `testParallelFollowABRemoveB`. This last step is worth 4 points.
+    else if !graph.contains(b) then unfollow(a, b)

src/main/scala/midterm22/appendix/appendix.scala

+package midterm22.appendix
+
+// Represents optional values. Instances of Option are either an instance of
+// scala.Some or the object None.
+sealed abstract class Option[+A]:
+  // Returns the option's value if the option is an instance of scala.Some, or
+  // throws an exception if the option is None.
+  def get: A
+  // Returns true if the option is an instance of scala.Some, false otherwise.
+  // This is equivalent to:
+  //     option match
+  //         case Some(v) => true
+  //         case None    => false
+  def isDefined: Boolean
+  // Returns this scala.Option if it is nonempty, otherwise return the result of
+  // evaluating alternative.
+  def orElse[B >: A](alternative: => Option[B]): Option[B]
+
+abstract class Iterable[+A]:
+  // Selects all elements except first n ones.
+  def drop(n: Int): Iterable[A]
+  // The size of this collection.
+  def size: Int
+  // Selects the first n elements.
+  def take(n: Int): Iterable[A]
+
+abstract class List[+A] extends Iterable[A]:
+  // Adds an element at the beginning of this list.
+  def ::[B >: A](elem: B): List[B]
+  // A copy of this sequence with an element appended.
+  def appended[B >: A](elem: B): List[B]
+  // Get the element at the specified index.
+  def apply(n: Int): A
+  // Selects all elements of this list which satisfy a predicate.
+  def filter(pred: (A) => Boolean): List[A]
+  // Selects the first element of this list.
+  def head: A
+  // Sorts this sequence according to a comparison function.
+  def sortWith(lt: (A, A) => Boolean): List[A]
+  // Selects all elements except the first.
+  def tail: List[A]
+
+abstract class Array[+A] extends Iterable[A]:
+  // Get the element at the specified index.
+  def apply(n: Int): A
+
+abstract class Thread:
+  // Subclasses should override this method.
+  def run(): Unit
+  // Causes this thread to begin execution; the Java Virtual Machine calls the
+  // run method of this thread.
+  def start(): Unit
+  // Waits for this thread to die.
+  def join(): Unit
+
+// Creates and starts a new task ran concurrently.
+def task[T](body: => T): ForkJoinTask[T] = ???
+
+abstract class ForkJoinTask[T]:
+  // Returns the result of the computation when it is done.
+  def join(): T
+
+// A concurrent hash-trie or TrieMap is a concurrent thread-safe lock-free
+// implementation of a hash array mapped trie.
+abstract class TrieMap[K, V]:
+  // Retrieves the value which is associated with the given key. Throws a
+  // NoSuchElementException if there is no mapping from the given key to a
+  // value.
+  def apply(key: K): V
+  // Tests whether this map contains a binding for a key.
+  def contains(key: K): Boolean
+  // Applies a function f to all elements of this concurrent map. This function
+  // iterates over a snapshot of the map.
+  def foreach[U](f: ((K, V)) => U): Unit
+  // Optionally returns the value associated with a key.
+  def get(key: K): Option[V]
+  // Collects all key of this map in an iterable collection. The result is a
+  // snapshot of the values at a specific point in time.
+  def keys: Iterator[K]
+  // Transforms this map by applying a function to every retrieved value. This
+  // returns a new map.
+  def mapValues[W](f: V => W): TrieMap[K, W]
+  // Associates the given key with a given value, unless the key was already
+  // associated with some other value. This is an atomic operation.
+  def putIfAbsent(k: K, v: V): Option[V]
+  // Removes a key from this map, returning the value associated previously with
+  // that key as an option.
+  def remove(k: K): Option[V]
+  // Removes the entry for the specified key if it's currently mapped to the
+  // specified value. This is an atomic operation.
+  def remove(k: K, v: V): Boolean
+  // Replaces the entry for the given key only if it was previously mapped to a
+  // given value. Returns true if the change is successful, or false otherwise.
+  // This is an atomic operation.
+  def replace(k: K, oldvalue: V, newvalue: V): Boolean
+  // Adds a new key/value pair to this map.
+  def update(k: K, v: V): Unit
+  // Collects all values of this map in an iterable collection. The result is a
+  // snapshot of the values at a specific point in time.
+  def values: Iterator[V]
+
+// An int value that may be updated atomically.
+// The constructor takes the initial value at its only argument. For example,
+// this create an `AtomicInteger` with an initial value of `42`:
+//     val myAtomicInteger = new AtomicInteger(42)
+abstract class AtomicInteger:
+  // Atomically adds the given value to the current value and returns the
+  // updated value.
+  def addAndGet(delta: Int): Int
+  // Atomically sets the value to the given updated value if the current value
+  // == the expected value. Returns true if the change is successful, or false
+  // otherwise. This is an atomic operation.
+  def compareAndSet(oldvalue: Int, newvalue: Int): Boolean
+  // Gets the current value. This is an atomic operation.
+  def get(): Int
+  // Atomically increments by one the current value. This is an atomic operation.
+  def incrementAndGet(): Int
+
+// ---------------------------------------------
+
+// Needed so that we can compile successfully, but not included for students.
+// See Option class doc instead.
+abstract class Some[+A](value: A) extends Option[A]
+object Some:
+  def apply[A](value: A): Some[A] = ???
+val None: Option[Nothing] = ???
+
+object List:
+  def apply[A](values: A*): List[A] = ???
+val Nil: List[Nothing] = ???
+
+object Array:
+  def apply[A](values: A*): Array[A] = ???

src/main/scala/midterm22/common.scala

+package midterm22
+
+import java.util.concurrent.ForkJoinTask
+import java.util.concurrent.RecursiveTask
+import java.util.concurrent.ForkJoinWorkerThread
+import java.util.concurrent.ForkJoinPool
+import java.util.concurrent.atomic.AtomicInteger
+
+val forkJoinPool = ForkJoinPool()
+var parallelismEnabled = true
+var tasksCreated: AtomicInteger = AtomicInteger(0)
+
+def schedule[T](body: => T): ForkJoinTask[T] =
+  val t = new RecursiveTask[T]:
+    def compute = body
+  Thread.currentThread match
+    case wt: ForkJoinWorkerThread => t.fork()
+    case _                        => forkJoinPool.execute(t)
+  t
+
+def task[T](body: => T): ForkJoinTask[T] =
+  tasksCreated.incrementAndGet
+  schedule(body)
+
+def parallel[A, B](op1: => A, op2: => B): (A, B) =
+  if parallelismEnabled then
+    val res1 = task { op1 }
+    val res2 = op2
+    (res1.join(), res2)
+  else (op1, op2)

src/main/scala/midterm23/BarberShopSolution.scala

+package midterm23
+
+import instrumentation.Monitor
+import scala.annotation.tailrec
+
+/** Runs one of the three provided solutions to the barber shop problem.
+  *
+  * Run using `sbt "runMain midterm23.barberSolution 2 3"` for example.
+  *
+  * @param solutionNumber
+  *   One of the 3 provided solutions to run: 1, 2, or 3.
+  * @param nCustomers
+  *   The number of customers to simulate.
+  */
+@main def barberSolution(solutionNumber: Int, nCustomers: Int): Unit =
+  val barberShop: AbstractBarberShopSolution =
+    solutionNumber match
+      case 1 => BarberShopSolution1(nCustomers)
+      case 2 => BarberShopSolution2(nCustomers)
+      case 3 => BarberShopSolution3(nCustomers)
+
+  barberSolutionMain(barberShop, solutionNumber, nCustomers)
+
+def barberSolutionMain(
+    barberShop: AbstractBarberShopSolution,
+    solutionNumber: Int,
+    nCustomers: Int
+): Unit =
+  val bthread = new Thread:
+    override def run() =
+      barberShop.barber()
+  bthread.start()
+
+  val customerThreads =
+    for i <- 1 to nCustomers yield
+      val cthread = new Thread:
+        override def run() =
+          barberShop.customer(i)
+      cthread.start()
+      cthread
+
+  bthread.join()
+  customerThreads.foreach(_.join())
+
+abstract class AbstractBarberShopSolution:
+  // var notifyBarber: Boolean = false
+  // var notifyCustomer: Boolean = false
+  // val lockObj = Object()
+
+  // In order to be able to mock the three attributes above in the tests, we
+  // replace them with three private attributes and public getters and setters.
+  // Without overrides, this is equivalent to the definitions above.
+
+  private var _notifyBarber: Boolean = false
+  private var _notifyCustomer: Boolean = false
+  private val _lockObject: Monitor = new Monitor {}
+  def notifyBarber = _notifyBarber
+  def notifyBarber_=(v: Boolean) = _notifyBarber = v
+  def notifyCustomer = _notifyCustomer
+  def notifyCustomer_=(v: Boolean) = _notifyCustomer = v
+  def lockObj: Monitor = _lockObject
+
+  def hairCut(): Unit =
+    log("start hair cut")
+    Thread.sleep(1000)
+    log("finish hair cut")
+
+  def customer(n: Int): Unit
+  def barber(): Unit
+
+  /** Logs a message to the console.
+    *
+    * Overridden in the tests to avoid printing gigabytes of output.
+    *
+    * @param s
+    */
+  def log(s: String) = println(s)
+
+/** This is the solution we expected for the `customer` method.
+  *
+  * However, now that you have learned about `@volatile`, you can see that the
+  * `customer` method is not thread-safe. Can you spot a problem with this
+  * solution?
+  *
+  * Note that the `log` calls were added to help you understand what is going
+  * on. They are not part of the expected solution.
+  */
+trait BarberShopCustomerSolution1 extends AbstractBarberShopSolution:
+  override def customer(n: Int) =
+    lockObj.synchronized {
+      log(f"customer ${n} sits on the chair and waits for the barber")
+      notifyBarber = true
+      while !notifyCustomer do {}
+      notifyCustomer = false
+      log(f"customer ${n} leaves the chair")
+    }
+
+/** This is the solution we expected for the `barber` method.
+  *
+  * Same question as above: can you spot a problem with this solution?
+  *
+  * @param customersNumber
+  *   The number of customers to simulate.
+  */
+trait BarberShopBarberSolution1(customersNumber: Int)
+    extends AbstractBarberShopSolution:
+  override def barber() =
+    // while true do
+
+    // We replace the infinite loop from the instructions with a loop doing a
+    // limited number of iterations so that we can test the code.
+    for _ <- 1 to customersNumber do
+      log("barber waits for a customer")
+      while !notifyBarber do {}
+      notifyBarber = false
+      hairCut()
+      notifyCustomer = true
+
+    log("barber stops working")
+
+class BarberShopSolution1(customersNumber: Int)
+    extends BarberShopCustomerSolution1,
+      BarberShopBarberSolution1(customersNumber)
+
+/** The problem with BarberShopSolution1 is that the `notifyBarber` and
+  * `notifyCustomer` are accessed from multiple threads without any
+  * synchronization. This means that the compiler is free to reorder the
+  * instructions in the `customer` and `barber` methods, which can lead to
+  * unexpected behaviors like infinite loops!
+  *
+  * To solve this problem, the solution below redefines the two attributes as
+  * `@volatile`.
+  *
+  * @param customersNumber
+  *   The number of customers to simulate.
+  */
+class BarberShopSolution2(customersNumber: Int)
+    extends BarberShopSolution1(customersNumber):
+  @volatile private var _notifyBarber: Boolean = false
+  @volatile private var _notifyCustomer: Boolean = false
+
+  override def notifyBarber = _notifyBarber
+  override def notifyBarber_=(v: Boolean) = _notifyBarber = v
+  override def notifyCustomer = _notifyCustomer
+  override def notifyCustomer_=(v: Boolean) = _notifyCustomer = v
+
+/** The solution below uses a different approach to solve the problem of
+  * BarberShopSolution1. Instead of using `@volatile`, it uses the lock object
+  * to synchronize the accesses to the two attributes. Note how each access to
+  * the attributes is now wrapped in a `lockObj.synchronized` block.
+  *
+  * @param customersNumber
+  *   The number of customers to simulate.
+  */
+class BarberShopSolution3(customersNumber: Int)
+    extends AbstractBarberShopSolution:
+  override def customer(n: Int) =
+    getHairCut(n)
+
+  @tailrec
+  private def getHairCut(n: Int): Unit =
+    val sited = lockObj.synchronized {
+      if notifyBarber then
+        false
+      else if notifyCustomer then
+        false
+      else
+        log(f"customer ${n} sits on the chair and waits for the barber")
+        notifyBarber = true
+        true
+    }
+    if sited then
+      while !lockObj.synchronized { notifyCustomer } do {}
+      lockObj.synchronized {
+        log(f"customer ${n} leaves the chair")
+        notifyCustomer = false
+      }
+    else
+      getHairCut(n)
+
+  override def barber() =
+    for _ <- 1 to customersNumber do
+      log("barber waits for a customer")
+
+      while !lockObj.synchronized { notifyBarber } do {}
+      hairCut()
+      lockObj.synchronized {
+        notifyCustomer = true
+        notifyBarber = false
+      }
+
+    log("barber stops working")

src/main/scala/midterm23/BarberShopStudentAnswer333.scala

+package midterm23
+
+import java.util.concurrent.atomic.AtomicInteger
+
+/** This is an example `BarberShopStudentAnswer`, like the one we made for your
+  * own solution and that you can find on Moodle. The solution below is boringly
+  * correct as it use the same implementation as the expected solution. See
+  * BarberShopSolution.scala for explanations and comments.
+  *
+  * @param customersNumber
+  *   The number of customers to simulate.
+  */
+class BarberShopStudentAnswer333(customersNumber: Int)
+    extends AbstractBarberShopSolution:
+
+  /** This attribute is used to count the number of hair cuts. It is needed for
+    * testing because we do not want to loop forever in the `barber` method!
+    */
+  val hairCutsCounter = AtomicInteger(customersNumber)
+
+  override def customer(n: Int) =
+    lockObj.synchronized {
+      notifyBarber = true
+      while !notifyCustomer do {}
+      notifyCustomer = false
+    }
+
+  override def barber() =
+    // while true
+
+    // We replace the infinite loop from the instructions with a loop doing a
+    // limited number of iterations so that we can test the code.
+    while hairCutsCounter.get() != 0 do
+      while !notifyBarber do {}
+      notifyBarber = false
+      hairCut()
+      hairCutsCounter.decrementAndGet() // You can ignore this line. It has been added for testing.
+      notifyCustomer = true
+
+/** Version of `BarberShopStudentAnswer333` with the methods `customer`
+  * overridden to use the expected implementation.
+  *
+  * @param n
+  *   The number of customers to simulate.
+  */
+class BarberShopStudentAnswer333WithCorrectCustomer(n: Int)
+    extends BarberShopStudentAnswer333(n)
+    with BarberShopCustomerSolution1
+
+/** Version of `BarberShopStudentAnswer333` with the methods `barber` overridden
+  * to use the expected implementation.
+  *
+  * @param n
+  *   The number of customers to simulate.
+  */
+class BarberShopStudentAnswer333WithCorrectBarber(n: Int)
+    extends BarberShopStudentAnswer333(n)
+    with BarberShopBarberSolution1(n)
+
+// To grade, we ran tests with each of these classes.

src/main/scala/midterm23/convolution.worksheet.sc

+import lecture1.parallel
+
+def numContributeTo(input: Array[Int], kernel: Array[Int], output: Array[Int], i: Int): Int =
+  if i < 0 || i >= output.length then
+    0
+  else if (i - kernel.length < 0) then
+    i+1
+  else if (i >= input.length) then
+    output.length - i
+  else kernel.length
+
+def sequentialConvolve(
+  input: Array[Int],
+  kernel: Array[Int],
+  output: Array[Int],
+  from: Int,
+  until: Int
+  ): Unit = {
+    // Approach A, as described in the clarification announcement for this
+    // exercise, where we treat `from` and `until` as indices on `output`
+    // instead of `input` as in the given code.
+    var iOutput = from
+    while iOutput < until do
+      var iKernel = math.max(0, iOutput - input.length + 1)
+      while iKernel < kernel.length && iKernel <= iOutput do
+        // `output` is only ever written to between the indices `from` and
+        // `until`, the range of `iOutput`. The indices for `input` and
+        // `kernel` are computed accordingly.
+        output(iOutput) += input(iOutput - iKernel) * kernel(iKernel)
+        iKernel += 1
+      iOutput += 1
+
+    // ALTERNATE SOLUTION: Approach B, as described in the clarification
+    // announcement for this exercise, which is unchanged from the given
+    // code, i.e. we treat `from` and `until` as indices on `input`.
+    var iInput = from
+    while iInput < until do
+      var iKernel = 0
+      while iKernel < kernel.length do
+        output(iInput + iKernel) += input(iInput) * kernel(iKernel)
+        iKernel += 1
+      iInput += 1
+  }
+
+def parallelConvolve(
+    input: Array[Int],
+    kernel: Array[Int],
+    output: Array[Int],
+    from: Int,
+    until: Int,
+    threshold: Int
+): Unit =
+  // Approach A, as described in the clarification announcement for this
+  // exercise, where we treat `from` and `until` as indices on `output`
+  // instead of `input` as in the given code. This does not require us to
+  // change anything in this function.  Only receives full credit if used
+  // together with Approach A for `sequentialConvolve`.
+  if (until - from) <= threshold then
+    sequentialConvolve(input, kernel, output, from, until)
+  else
+    val mid = from + (until - from) / 2
+    parallel(
+      parallelConvolve(input, kernel, output, from, mid, threshold),
+      parallelConvolve(input, kernel, output, mid, until, threshold)
+    )
+
+  // ALTERNATE SOLUTION: Approach B, as described in the clarification
+  // announcement for this exercise, where we treat `from` and `until` as
+  // indices on `input` as in the given code. This requires up to leave a
+  // gap in the parallel calls to be filled in sequentially as described
+  // below. Only receives full credit if used together with Approach B for
+  // `sequentialConvolve`.
+  if (until - from) <= threshold then
+    sequentialConvolve(input, kernel, output, from, until)
+  else
+    val mid = from + (until - from) / 2
+    val gap = numContributeTo(input, kernel, output, mid)
+    // Leave a gap large enough that accesses to `output` from the first
+    // parallel call will not overlap with accesses from the second. This
+    // gap is of size equal to the number of elements of `input` that will
+    // contribute to the computation of the result at the lowest index of
+    // `output` in the second parallel call, namely, at index `mid`.
+    // However, we are careful not to access indices lower than `from`.
+    val gapBeforeMid = Math.max(from, mid - gap + 1)
+    parallel(
+      parallelConvolve(input, kernel, output, from, gapBeforeMid, threshold),
+      parallelConvolve(input, kernel, output, mid, until, threshold)
+    )
+    // Fill in the gap sequentially.
+    sequentialConvolve(input, kernel, output, gapBeforeMid, mid)
+
+object Original_WithOutputRace:
+
+  def sequentialConvolve(
+      input: Array[Int],
+      kernel: Array[Int],
+      output: Array[Int],
+      from: Int,
+      until: Int): Unit =
+    var iInput = from
+    while iInput < until do
+      var iKernel = 0
+      while iKernel < kernel.length do
+        output(iInput + iKernel) += input(iInput) * kernel(iKernel)
+        iKernel += 1
+      iInput += 1
+
+  def parallelConvolve(
+      input: Array[Int],
+      kernel: Array[Int],
+      output: Array[Int],
+      from: Int,
+      until: Int,
+      threshold: Int): Unit =
+    if (until - from) <= threshold then
+      sequentialConvolve(input, kernel, output, from, until)
+    else
+      val mid = from + (until - from)/2
+      parallel(
+        parallelConvolve(input, kernel, output, from, mid, threshold),
+        parallelConvolve(input, kernel, output, mid, until, threshold))

src/main/scala/midterm23/groupBy.worksheet.sc

+import lecture1.parallel
+import collection.concurrent.TrieMap
+import collection.parallel.CollectionConverters.*
+
+// Example run of `groupBy`:
+case class Course(semester: String, id: String)
+val cs210 = Course("BA3", "CS210")
+val cs250 = Course("BA3", "CS250")
+val cs206 = Course("BA4", "CS206")
+val courses = Vector(cs210, cs250, cs206)
+assertEquals(
+  courses.groupBy(_.semester),
+  Map(
+    "BA3" -> Vector(cs210, cs250),
+    "BA4" -> Vector(cs206)
+  )
+)
+
+/** Asymptotic run time: O(1) */
+def appendAtKey[K, V](m: Map[K, Vector[V]], k: K, v: V): Map[K, Vector[V]] =
+  val prev = m.getOrElse(k, Vector.empty)
+  m.updated(k, prev :+ v)
+
+// Example runs of `appendAtKey`:
+assertEquals(
+  appendAtKey(Map(), 0, 1),
+  Map(0 -> Vector(1))
+)
+assertEquals(
+  appendAtKey(Map("BA3" -> Vector(cs210)), "BA3", cs250),
+  Map("BA3" -> List(cs210, cs250))
+)
+
+/** Asymptotic run time: O(1) */
+def mergeMaps[K, V](a: Map[K, Vector[V]], b: Map[K, Vector[V]]) =
+  var res = a
+  for (k, vs) <- b do // O(1) iterations (we assume the number of keys to be O(1))
+    val prev = res.getOrElse(k, Vector.empty) // O(1) lookup
+    res = res.updated(k, prev ++ vs) // O(1) update, O(1) concatenation
+  res
+
+// Example run of `mergeMaps`:
+val m1 = Map(1 -> Vector(5, 7))
+val m2 = Map(1 -> Vector(6), 2 -> Vector(3))
+assertEquals(
+  mergeMaps(m1, m2),
+  Map(
+    1 -> Vector(5, 7, 6),
+    2 -> Vector(3)
+  )
+)
+
+def parGroupBy1[K, V](vs: Vector[V], key: V => K): Map[K, Vector[V]] =
+  if vs.size == 1 then
+    Map(key(vs.head) -> Vector(vs.head))
+  else
+    val mid = vs.size / 2
+    val (m1, m2) = parallel(
+      parGroupBy1(vs.slice(0, mid), key),
+      parGroupBy1(vs.slice(mid, vs.size), key)
+    )
+    mergeMaps(m1, m2)
+
+assertEquals(
+  parGroupBy1(courses, c => c.semester),
+  Map(
+    "BA3" -> List(cs210, cs250),
+    "BA4" -> List(cs206)
+  )
+)
+
+assertEquals(
+  parGroupBy1(courses, c => c.semester),
+  Map(
+    "BA3" -> List(cs210, cs250),
+    "BA4" -> List(cs206)
+  )
+)
+
+def parGroupBy2[K, V](vs: Vector[V], key: V => K): Map[K, Vector[V]] =
+  if vs.size == 1 then
+    Map(key(vs.head) -> Vector(vs.head))
+  else
+    val (m1, m2) = parallel(
+      parGroupBy2(Vector(vs.head), key),
+      parGroupBy2(vs.tail, key)
+    )
+    mergeMaps(m1, m2)
+
+assertEquals(
+  parGroupBy2(courses, c => c.semester),
+  Map(
+    "BA3" -> List(cs210, cs250),
+    "BA4" -> List(cs206)
+  )
+)
+
+assertEquals(
+  parGroupBy2(courses, c => c.semester),
+  Map(
+    "BA3" -> List(cs210, cs250),
+    "BA4" -> List(cs206)
+  )
+)
+
+def parGroupBy3[K, V](vs: Vector[V], key: V => K): Map[K, Vector[V]] =
+  val lock = Object()
+  var res = Map[K, Vector[V]]()
+  for v <- vs.par do
+    lock.synchronized { res = appendAtKey(res, key(v), v) }
+  res
+
+var values3 = Set[Map[String, Vector[Course]]]()
+for _ <- 1 to 100000 do
+  values3 += parGroupBy3(courses, _.semester)
+assertEquals(values3.size, 2)
+values3
+
+def parGroupBy4[K, V](vs: Vector[V], key: V => K): Map[K, Vector[V]] =
+  val res = TrieMap[K, Vector[V]]()
+  for v <- vs.par do
+    val k = key(v)
+    val prev = res.getOrElse(k, Vector.empty)
+    res.update(k, prev :+ v)
+  res.toMap
+
+var values4 = Set[Map[String, Vector[Course]]]()
+for _ <- 1 to 100000 do
+  values4 += parGroupBy4(courses, _.semester)
+assertEquals(values4.size, 4)
+values4
+
+def parGroupBySolution[K, V](vs: Vector[V], key: V => K): Map[K, Vector[V]] =
+  vs.par.aggregate(Map.empty)(
+    (m, v) => appendAtKey(m, key(v), v),
+    mergeMaps
+  )
+
+var valuesSol = Set[Map[String, Vector[Course]]]()
+for _ <- 1 to 100000 do
+  valuesSol += parGroupBySolution(courses, _.semester)
+assertEquals(valuesSol.size, 1)
+valuesSol
+
+def assertEquals(a: Any, b: Any) = assert(a == b)

src/test/scala/concpar21final01/Problem1Suite.scala

+package concpar21final01
+
+import scala.concurrent.duration.*
+import scala.concurrent.Future
+import scala.concurrent.ExecutionContext.Implicits.global
+
+class Problem1Suite extends munit.FunSuite:
+  test(
+    "Retrieves grades at the end of the exam (everyone pushed something) (10pts)"
+  ) {
+    class Problem1Done extends Problem1:
+      override def getGrade(sciper: Int): Future[Option[Grade]] =
+        Future {
+          Thread.sleep(100)
+          Some(Grade(sciper, sciper))
+        }
+      override def getScipers(): Future[List[Int]] =
+        Future {
+          Thread.sleep(100)
+          List(1, 2, 3, 4)
+        }
+
+    val expected: List[Grade] =
+      List(Grade(1, 1.0), Grade(2, 2.0), Grade(3, 3.0), Grade(4, 4.0))
+
+    (new Problem1Done).leaderboard().map { grades =>
+      assertEquals(grades.toSet, expected.toSet)
+    }
+  }
+
+  test("Retrieves grades mid exam (some students didn't push yet) (10pts)") {
+    class Problem1Partial extends Problem1:
+      override def getGrade(sciper: Int): Future[Option[Grade]] =
+        Future {
+          Thread.sleep(100)
+          if sciper % 2 == 0 then None
+          else Some(Grade(sciper, sciper))
+        }
+      override def getScipers(): Future[List[Int]] =
+        Future {
+          Thread.sleep(100)
+          List(1, 2, 3, 4)
+        }
+
+    val expected: List[Grade] =
+      List(Grade(1, 1.0), Grade(3, 3.0))
+
+    (new Problem1Partial).leaderboard().map { grades =>
+      assertEquals(grades.toSet, expected.toSet)
+    }
+  }
+
+  test("The output list is sorted by grade (10pts)") {
+    (new Problem1MockData).leaderboard().map { grades =>
+      assert(grades.size >= 176)
+      assert(grades.zipWithIndex.forall { case (g, i) =>
+        grades.drop(i).forall(x => g.grade >= x.grade)
+      })
+    }
+  }
+
+  test("GitLab API calls are done in parallel (2pts)") {
+    var inParallel: Boolean = false
+
+    class Problem1Par extends Problem1MockData:
+      var in: Boolean = false
+
+      override def getGrade(sciper: Int): Future[Option[Grade]] =
+        Future {
+          if in then inParallel = true
+          in = true
+          val out = super.getGrade(sciper)
+          in = false
+          concurrent.Await.result(out, Duration(10, SECONDS))
+        }
+
+    (new Problem1Par).leaderboard().map { grades =>
+      assert(grades.size >= 176)
+      assert(inParallel)
+    }
+  }
+
+  test("The IS-academia API is called exactly once (2pts)") {
+    var called: Int = 0
+
+    class Problem1Once extends Problem1MockData:
+      override def getScipers(): Future[List[Int]] =
+        called += 1
+        super.getScipers()
+
+    (new Problem1Once).leaderboard().map { grades =>
+      assert(grades.size >= 176)
+      assert(called == 1)
+    }
+  }

src/test/scala/concpar21final02/Problem2Suite.scala

+package concpar21final02
+
+import akka.actor.*
+import akka.testkit.*
+import scala.collection.mutable
+import concurrent.duration.*
+
+import Problem2.*
+
+class Problem2Suite extends munit.FunSuite:
+  import NotificationService.Protocol.*
+  import NotificationService.Responses.*
+  import DiscordChannel.Protocol.*
+  import DiscordChannel.Responses.*
+
+  test("Notification register (1pts)") {
+    new MyTestKit:
+      def tests() =
+        val actor = system.actorOf(Props[NotificationService]())
+        actor ! Register
+        expectMsg(Registered(true))
+  }
+
+  test("Notification register and un-register (1pts)") {
+    new MyTestKit:
+      def tests() =
+        val actor = system.actorOf(Props[NotificationService]())
+        actor ! Register
+        expectMsg(Registered(true))
+        actor ! UnRegister
+        expectMsg(Registered(false))
+        actor ! UnRegister
+        expectMsg(Registered(false))
+        actor ! Register
+        expectMsg(Registered(true))
+        actor ! UnRegister
+        expectMsg(Registered(false))
+  }
+
+  test("Notification notify (1pts)") {
+    new MyTestKit:
+      def tests() =
+        val actor = system.actorOf(Props[NotificationService]())
+        actor ! Register
+        expectMsg(Registered(true))
+        actor ! NotifyAll
+        expectMsg(Notification)
+        actor ! NotifyAll
+        expectMsg(Notification)
+        actor ! UnRegister
+        expectMsg(Registered(false))
+        actor ! NotifyAll
+        expectNoMessage()
+        actor ! Register
+        expectMsg(Registered(true))
+        actor ! NotifyAll
+        expectMsg(Notification)
+        actor ! UnRegister
+        expectMsg(Registered(false))
+        actor ! NotifyAll
+        expectNoMessage()
+  }
+
+  test("NotifyAll from other actor (1pts)") {
+    new MyTestKit:
+      def tests() =
+        val actor = system.actorOf(Props[NotificationService]())
+        val otherActor = system.actorOf(Props[DummyActor]())
+
+        def notifyFormAllFromOtherActor() =
+          given ActorRef = otherActor
+          actor ! NotifyAll
+
+        expectNoMessage()
+
+        actor ! Register
+        expectMsg(Registered(true))
+
+        notifyFormAllFromOtherActor()
+        expectMsg(Notification)
+  }
+
+  test("Channel init (1pts)") {
+    new MyTestKit:
+      def tests() =
+        val notificationService = system.actorOf(Props[NotificationService]())
+        val channel = system.actorOf(Props[DiscordChannel]())
+        channel ! Init(notificationService)
+        expectMsg(Active)
+  }
+
+  test("Channel post and get post (1pts)") {
+    new MyTestKit:
+      def tests() =
+        val notificationService = system.actorOf(Props[NotificationService]())
+        val channel = system.actorOf(Props[DiscordChannel]())
+        channel ! Init(notificationService)
+        expectMsg(Active)
+        channel ! Post("hello")
+        channel ! GetLastPosts(1)
+        expectMsg(Posts(List("hello")))
+        channel ! GetLastPosts(10)
+        expectMsg(Posts(List("hello")))
+        channel ! GetLastPosts(0)
+        expectMsg(Posts(Nil))
+  }
+
+  test("Channel multiple posts (1pts)") {
+    new MyTestKit:
+      def tests() =
+        val notificationService = system.actorOf(Props[NotificationService]())
+        val channel = system.actorOf(Props[DiscordChannel]())
+        channel ! Init(notificationService)
+        expectMsg(Active)
+        channel ! Post("hello")
+        channel ! Post("world")
+        channel ! GetLastPosts(2)
+        channel ! GetLastPosts(1)
+        channel ! Post("!")
+        channel ! GetLastPosts(3)
+        expectMsg(Posts(List("world", "hello")))
+        expectMsg(Posts(List("world")))
+        expectMsg(Posts(List("!", "world", "hello")))
+  }
+
+  test("Channel posts and notify (1pts)") {
+    new MyTestKit:
+      def tests() =
+        val notificationService = system.actorOf(Props[NotificationService]())
+        val channel = system.actorOf(Props[DiscordChannel]())
+        channel ! Init(notificationService)
+        expectMsg(Active)
+        notificationService ! Register
+        expectMsg(Registered(true))
+        channel ! Post("hello")
+        channel ! Post("world")
+        expectMsg(Notification)
+        expectMsg(Notification)
+  }
+
+  test("Channel init twice (1pts)") {
+    new MyTestKit:
+      def tests() =
+        val notificationService = system.actorOf(Props[NotificationService]())
+        val channel = system.actorOf(Props[DiscordChannel]())
+        channel ! Init(notificationService)
+        expectMsg(Active)
+        channel ! Init(notificationService)
+        expectMsg(AlreadyActive)
+        channel ! Init(notificationService)
+        expectMsg(AlreadyActive)
+  }
+
+  test("Channel not active (1pts)") {
+    new MyTestKit:
+      def tests() =
+        val channel1 = system.actorOf(Props[DiscordChannel]())
+        channel1 ! Post("hello")
+        expectMsg(NotActive)
+
+        val channel2 = system.actorOf(Props[DiscordChannel]())
+        channel2 ! GetLastPosts(0)
+        expectMsg(NotActive)
+  }
+
+  abstract class MyTestKit
+      extends TestKit(ActorSystem("TestSystem"))
+      with ImplicitSender:
+    def tests(): Unit
+    try tests()
+    finally shutdown(system)
+
+class DummyActor extends Actor:
+  def receive: Receive = { case _ =>
+    ()
+  }

src/test/scala/concpar21final03/Problem3Suite.scala

+package concpar21final03
+
+import scala.annotation.tailrec
+import scala.concurrent.*
+import scala.concurrent.duration.*
+import scala.collection.mutable.HashMap
+import scala.util.Random
+import instrumentation.*
+import instrumentation.TestHelper.*
+import instrumentation.TestUtils.*
+
+class Problem3Suite extends munit.FunSuite:
+
+  test("Part 1: ThreadMap (3pts)") {
+    testManySchedules(
+      4,
+      sched =>
+        val tmap = new SchedulableThreadMap[Int](sched)
+
+        def writeThread(): Unit =
+          tmap.setCurrentThreadValue(0)
+          tmap.setCurrentThreadValue(-1)
+          val readBack = tmap.currentThreadValue
+          assertEquals(readBack, Some(-1))
+
+        def writeAndDeleteThread(): Unit =
+          tmap.setCurrentThreadValue(42)
+          tmap.deleteCurrentThreadValue()
+
+        @tailrec
+        def waitThread(): Unit =
+          tmap.waitForall(_ < 0)
+          val all = tmap.allValues
+          if all != List(-1) then waitThread()
+
+        val threads = List(
+          () => writeThread(),
+          () => writeAndDeleteThread(),
+          () => waitThread(),
+          () => waitThread()
+        )
+
+        (threads, _ => (true, ""))
+    )
+  }
+
+  test("Part 2: RCU (5pts)") {
+    testManySchedules(
+      3,
+      sched =>
+        val rcu = new SchedulableRCU(sched)
+
+        case class State(
+            value: Int,
+            isDeleted: AtomicLong = SchedulableAtomicLong(0, sched, "isDeleted")
+        )
+
+        val sharedState =
+          SchedulableAtomicReference(State(0), sched, "sharedState")
+
+        def readThread(): Unit =
+          rcu.startRead()
+          val state = sharedState.get
+          val stateWasDeleted = state.isDeleted.get != 0
+          assert(
+            !stateWasDeleted,
+            "RCU shared state deleted in the middle of a read."
+          )
+          rcu.stopRead()
+
+        def writeThread(): Unit =
+          val oldState = sharedState.get
+          sharedState.set(State(oldState.value + 1))
+          rcu.waitForOldReads()
+          oldState.isDeleted.set(1)
+
+        val threads = List(
+          () => readThread(),
+          () => readThread(),
+          () => writeThread()
+        )
+
+        (threads, _ => (true, ""))
+    )
+  }
+
+  test("Part 3: UpdateServer (2pts)") {
+    testManySchedules(
+      3,
+      sched =>
+        val fs = SchedulableInMemoryFileSystem(sched)
+        val server = new SchedulableUpdateServer(sched, fs)
+
+        def writeThread(): Unit =
+          server.newUpdate("update1.bin", "Update 1")
+          server.newUpdate("update2.bin", "Update 2")
+          assertEquals(fs.fsMap.toSet, Set("update2.bin" -> "Update 2"))
+
+        def fetchThread(): Unit =
+          val res = server.fetchUpdate()
+          assert(
+            List(None, Some("Update 1"), Some("Update 2")).contains(res),
+            s"fetchUpdate returned unexpected value $res"
+          )
+
+        val threads = List(
+          () => writeThread(),
+          () => fetchThread(),
+          () => fetchThread()
+        )
+
+        (threads, _ => (true, ""))
+    )
+  }
+
+  import scala.concurrent.duration.*
+  override val munitTimeout = 200.seconds
+end Problem3Suite

src/test/scala/concpar21final03/instrumentation/AtomicReference.scala

+package concpar21final03.instrumentation
+
+class AtomicReference[T](initial: T):
+
+  private val atomic =
+    new java.util.concurrent.atomic.AtomicReference[T](initial)
+
+  def get: T = atomic.get()
+
+  def set(value: T): Unit = atomic.set(value)

src/test/scala/concpar21final03/instrumentation/MockedMonitor.scala

+package concpar21final03.instrumentation
+
+trait MockedMonitor extends Monitor:
+  def scheduler: Scheduler
+
+  // Can be overriden.
+  override def waitDefault() =
+    scheduler.log("wait")
+    scheduler updateThreadState Wait(this, scheduler.threadLocks.tail)
+  override def synchronizedDefault[T](toExecute: => T): T =
+    scheduler.log("synchronized check")
+    val prevLocks = scheduler.threadLocks
+    scheduler updateThreadState Sync(
+      this,
+      prevLocks
+    ) // If this belongs to prevLocks, should just continue.
+    scheduler.log("synchronized -> enter")
+    try toExecute
+    finally
+      scheduler updateThreadState Running(prevLocks)
+      scheduler.log("synchronized -> out")
+  override def notifyDefault() =
+    scheduler mapOtherStates { state =>
+      state match
+        case Wait(lockToAquire, locks) if lockToAquire == this =>
+          SyncUnique(this, state.locks)
+        case e => e
+    }
+    scheduler.log("notify")
+  override def notifyAllDefault() =
+    scheduler mapOtherStates { state =>
+      state match
+        case Wait(lockToAquire, locks) if lockToAquire == this =>
+          Sync(this, state.locks)
+        case SyncUnique(lockToAquire, locks) if lockToAquire == this =>
+          Sync(this, state.locks)
+        case e => e
+    }
+    scheduler.log("notifyAll")
+
+trait LockFreeMonitor extends Monitor:
+  override def waitDefault() =
+    throw new Exception("Please use lock-free structures and do not use wait()")
+  override def synchronizedDefault[T](toExecute: => T): T =
+    throw new Exception(
+      "Please use lock-free structures and do not use synchronized()"
+    )
+  override def notifyDefault() =
+    throw new Exception(
+      "Please use lock-free structures and do not use notify()"
+    )
+  override def notifyAllDefault() =
+    throw new Exception(
+      "Please use lock-free structures and do not use notifyAll()"
+    )
+
+abstract class ThreadState:
+  def locks: Seq[AnyRef]
+trait CanContinueIfAcquiresLock extends ThreadState:
+  def lockToAquire: AnyRef
+case object Start extends ThreadState:
+  def locks: Seq[AnyRef] = Seq.empty
+case object End extends ThreadState:
+  def locks: Seq[AnyRef] = Seq.empty
+case class Wait(lockToAquire: AnyRef, locks: Seq[AnyRef]) extends ThreadState
+case class SyncUnique(lockToAquire: AnyRef, locks: Seq[AnyRef])
+    extends ThreadState
+    with CanContinueIfAcquiresLock
+case class Sync(lockToAquire: AnyRef, locks: Seq[AnyRef])
+    extends ThreadState
+    with CanContinueIfAcquiresLock
+case class Running(locks: Seq[AnyRef]) extends ThreadState
+case class VariableReadWrite(locks: Seq[AnyRef]) extends ThreadState

src/test/scala/concpar21final03/instrumentation/Scheduler.scala

+package concpar21final03.instrumentation
+
+import java.util.concurrent.*;
+import scala.concurrent.duration.*
+import scala.collection.mutable.*
+import Stats.*
+
+import java.util.concurrent.atomic.AtomicInteger
+
+sealed abstract class Result
+case class RetVal(rets: List[Any]) extends Result
+case class Except(msg: String, stackTrace: Array[StackTraceElement])
+    extends Result
+case class Timeout(msg: String) extends Result
+
+/** A class that maintains schedule and a set of thread ids. The schedules are
+  * advanced after an operation of a SchedulableBuffer is performed. Note: the
+  * real schedule that is executed may deviate from the input schedule due to
+  * the adjustments that had to be made for locks
+  */
+class Scheduler(sched: List[Int]):
+  val maxOps =
+    500 // a limit on the maximum number of operations the code is allowed to perform
+
+  private var schedule = sched
+  private var numThreads = 0
+  private val realToFakeThreadId = Map[Long, Int]()
+  private val opLog =
+    ListBuffer[String]() // a mutable list (used for efficient concat)
+  private val threadStates = Map[Int, ThreadState]()
+
+  /** Runs a set of operations in parallel as per the schedule. Each operation
+    * may consist of many primitive operations like reads or writes to shared
+    * data structure each of which should be executed using the function `exec`.
+    * @timeout
+    *   in milliseconds
+    * @return
+    *   true - all threads completed on time, false -some tests timed out.
+    */
+  def runInParallel(timeout: Long, ops: List[() => Any]): Result =
+    numThreads = ops.length
+    val threadRes = Array.fill(numThreads) { None: Any }
+    var exception: Option[Except] = None
+    val syncObject = new Object()
+    var completed = new AtomicInteger(0)
+    // create threads
+    val threads = ops.zipWithIndex.map { case (op, i) =>
+      new Thread(
+        new Runnable():
+          def run(): Unit =
+            val fakeId = i + 1
+            setThreadId(fakeId)
+            try
+              updateThreadState(Start)
+              val res = op()
+              updateThreadState(End)
+              threadRes(i) = res
+              // notify the master thread if all threads have completed
+              if completed.incrementAndGet() == ops.length then
+                syncObject.synchronized { syncObject.notifyAll() }
+            catch
+              case e: Throwable
+                  if exception != None => // do nothing here and silently fail
+              case e: Throwable =>
+                log(s"throw ${e.toString}")
+                exception = Some(
+                  Except(
+                    s"Thread $fakeId crashed on the following schedule: \n" + opLog
+                      .mkString("\n"),
+                    e.getStackTrace
+                  )
+                )
+                syncObject.synchronized { syncObject.notifyAll() }
+            // println(s"$fakeId: ${e.toString}")
+            // Runtime.getRuntime().halt(0) //exit the JVM and all running threads (no other way to kill other threads)
+      )
+    }
+    // start all threads
+    threads.foreach(_.start())
+    // wait for all threads to complete, or for an exception to be thrown, or for the time out to expire
+    var remTime = timeout
+    syncObject.synchronized {
+      timed { if completed.get() != ops.length then syncObject.wait(timeout) } {
+        time => remTime -= time
+      }
+    }
+    if exception.isDefined then exception.get
+    else if remTime <= 1
+    then // timeout ? using 1 instead of zero to allow for some errors
+      Timeout(opLog.mkString("\n"))
+    else
+      // every thing executed normally
+      RetVal(threadRes.toList)
+
+  // Updates the state of the current thread
+  def updateThreadState(state: ThreadState): Unit =
+    val tid = threadId
+    synchronized {
+      threadStates(tid) = state
+    }
+    state match
+      case Sync(lockToAquire, locks) =>
+        if locks.indexOf(lockToAquire) < 0 then waitForTurn
+        else
+          // Re-aqcuiring the same lock
+          updateThreadState(Running(lockToAquire +: locks))
+      case Start      => waitStart()
+      case End        => removeFromSchedule(tid)
+      case Running(_) =>
+      case _          => waitForTurn // Wait, SyncUnique, VariableReadWrite
+
+  def waitStart(): Unit =
+    // while (threadStates.size < numThreads) {
+    // Thread.sleep(1)
+    // }
+    synchronized {
+      if threadStates.size < numThreads then wait()
+      else notifyAll()
+    }
+
+  def threadLocks =
+    synchronized {
+      threadStates(threadId).locks
+    }
+
+  def threadState =
+    synchronized {
+      threadStates(threadId)
+    }
+
+  def mapOtherStates(f: ThreadState => ThreadState) =
+    val exception = threadId
+    synchronized {
+      for k <- threadStates.keys if k != exception do
+        threadStates(k) = f(threadStates(k))
+    }
+
+  def log(str: String) =
+    if (realToFakeThreadId contains Thread.currentThread().getId()) then
+      val space = (" " * ((threadId - 1) * 2))
+      val s =
+        space + threadId + ":" + "\n".r.replaceAllIn(str, "\n" + space + "  ")
+      opLog += s
+
+  /** Executes a read or write operation to a global data structure as per the
+    * given schedule
+    * @param msg
+    *   a message corresponding to the operation that will be logged
+    */
+  def exec[T](
+      primop: => T
+  )(msg: => String, postMsg: => Option[T => String] = None): T =
+    if !(realToFakeThreadId contains Thread.currentThread().getId()) then primop
+    else
+      updateThreadState(VariableReadWrite(threadLocks))
+      val m = msg
+      if m != "" then log(m)
+      if opLog.size > maxOps then
+        throw new Exception(
+          s"Total number of reads/writes performed by threads exceed $maxOps. A possible deadlock!"
+        )
+      val res = primop
+      postMsg match
+        case Some(m) => log(m(res))
+        case None    =>
+      res
+
+  private def setThreadId(fakeId: Int) = synchronized {
+    realToFakeThreadId(Thread.currentThread.getId) = fakeId
+  }
+
+  def threadId =
+    try realToFakeThreadId(Thread.currentThread().getId())
+    catch
+      case e: NoSuchElementException =>
+        throw new Exception(
+          "You are accessing shared variables in the constructor. This is not allowed. The variables are already initialized!"
+        )
+
+  private def isTurn(tid: Int) = synchronized {
+    (!schedule.isEmpty && schedule.head != tid)
+  }
+
+  def canProceed(): Boolean =
+    val tid = threadId
+    canContinue match
+      case Some((i, state)) if i == tid =>
+        // println(s"$tid: Runs ! Was in state $state")
+        canContinue = None
+        state match
+          case Sync(lockToAquire, locks) =>
+            updateThreadState(Running(lockToAquire +: locks))
+          case SyncUnique(lockToAquire, locks) =>
+            mapOtherStates {
+              _ match
+                case SyncUnique(lockToAquire2, locks2)
+                    if lockToAquire2 == lockToAquire =>
+                  Wait(lockToAquire2, locks2)
+                case e => e
+            }
+            updateThreadState(Running(lockToAquire +: locks))
+          case VariableReadWrite(locks) => updateThreadState(Running(locks))
+        true
+      case Some((i, state)) =>
+        // println(s"$tid: not my turn but $i !")
+        false
+      case None =>
+        false
+
+  var threadPreference =
+    0 // In the case the schedule is over, which thread should have the preference to execute.
+
+  /** returns true if the thread can continue to execute, and false otherwise */
+  def decide(): Option[(Int, ThreadState)] =
+    if !threadStates.isEmpty
+    then // The last thread who enters the decision loop takes the decision.
+      // println(s"$threadId: I'm taking a decision")
+      if threadStates.values.forall {
+          case e: Wait => true
+          case _       => false
+        }
+      then
+        val waiting = threadStates.keys.map(_.toString).mkString(", ")
+        val s = if threadStates.size > 1 then "s" else ""
+        val are = if threadStates.size > 1 then "are" else "is"
+        throw new Exception(
+          s"Deadlock: Thread$s $waiting $are waiting but all others have ended and cannot notify them."
+        )
+      else
+        // Threads can be in Wait, Sync, SyncUnique, and VariableReadWrite mode.
+        // Let's determine which ones can continue.
+        val notFree =
+          threadStates.collect { case (id, state) => state.locks }.flatten.toSet
+        val threadsNotBlocked = threadStates.toSeq.filter {
+          case (id, v: VariableReadWrite) => true
+          case (id, v: CanContinueIfAcquiresLock) =>
+            !notFree(v.lockToAquire) || (v.locks contains v.lockToAquire)
+          case _ => false
+        }
+        if threadsNotBlocked.isEmpty then
+          val waiting = threadStates.keys.map(_.toString).mkString(", ")
+          val s = if threadStates.size > 1 then "s" else ""
+          val are = if threadStates.size > 1 then "are" else "is"
+          val whoHasLock = threadStates.toSeq.flatMap { case (id, state) =>
+            state.locks.map(lock => (lock, id))
+          }.toMap
+          val reason = threadStates
+            .collect {
+              case (id, state: CanContinueIfAcquiresLock)
+                  if !notFree(state.lockToAquire) =>
+                s"Thread $id is waiting on lock ${state.lockToAquire} held by thread ${whoHasLock(state.lockToAquire)}"
+            }
+            .mkString("\n")
+          throw new Exception(
+            s"Deadlock: Thread$s $waiting are interlocked. Indeed:\n$reason"
+          )
+        else if threadsNotBlocked.size == 1
+        then // Do not consume the schedule if only one thread can execute.
+          Some(threadsNotBlocked(0))
+        else
+          val next = schedule.indexWhere(t =>
+            threadsNotBlocked.exists { case (id, state) => id == t }
+          )
+          if next != -1 then
+            // println(s"$threadId: schedule is $schedule, next chosen is ${schedule(next)}")
+            val chosenOne = schedule(
+              next
+            ) // TODO: Make schedule a mutable list.
+            schedule = schedule.take(next) ++ schedule.drop(next + 1)
+            Some((chosenOne, threadStates(chosenOne)))
+          else
+            threadPreference = (threadPreference + 1) % threadsNotBlocked.size
+            val chosenOne = threadsNotBlocked(
+              threadPreference
+            ) // Maybe another strategy
+            Some(chosenOne)
+            // threadsNotBlocked.indexOf(threadId) >= 0
+            /*
+            val tnb = threadsNotBlocked.map(_._1).mkString(",")
+            val s = if (schedule.isEmpty) "empty" else schedule.mkString(",")
+            val only = if (schedule.isEmpty) "" else " only"
+            throw new Exception(s"The schedule is $s but$only threads ${tnb} can continue")*/
+    else canContinue
+
+  /** This will be called before a schedulable operation begins. This should not
+    * use synchronized
+    */
+  var numThreadsWaiting = new AtomicInteger(0)
+  // var waitingForDecision = Map[Int, Option[Int]]() // Mapping from thread ids to a number indicating who is going to make the choice.
+  var canContinue: Option[(Int, ThreadState)] =
+    None // The result of the decision thread Id of the thread authorized to continue.
+  private def waitForTurn =
+    synchronized {
+      if numThreadsWaiting.incrementAndGet() == threadStates.size then
+        canContinue = decide()
+        notifyAll()
+      // waitingForDecision(threadId) = Some(numThreadsWaiting)
+      // println(s"$threadId Entering waiting with ticket number $numThreadsWaiting/${waitingForDecision.size}")
+      while !canProceed() do wait()
+    }
+    numThreadsWaiting.decrementAndGet()
+
+  /** To be invoked when a thread is about to complete
+    */
+  private def removeFromSchedule(fakeid: Int) = synchronized {
+    // println(s"$fakeid: I'm taking a decision because I finished")
+    schedule = schedule.filterNot(_ == fakeid)
+    threadStates -= fakeid
+    if numThreadsWaiting.get() == threadStates.size then
+      canContinue = decide()
+      notifyAll()
+  }
+
+  def getOperationLog() = opLog

src/test/scala/concpar21final03/instrumentation/Stats.scala

+/* Copyright 2009-2015 EPFL, Lausanne */
+package concpar21final03.instrumentation
+
+import java.lang.management.*
+
+/** A collection of methods that can be used to collect run-time statistics */
+object Stats:
+  def timed[T](code: => T)(cont: Long => Unit): T =
+    var t1 = System.currentTimeMillis()
+    val r = code
+    cont((System.currentTimeMillis() - t1))
+    r
+
+  def withTime[T](code: => T): (T, Long) =
+    var t1 = System.currentTimeMillis()
+    val r = code
+    (r, (System.currentTimeMillis() - t1))

src/test/scala/concpar21final03/instrumentation/TestHelper.scala

+package concpar21final03.instrumentation
+
+import scala.util.Random
+import scala.collection.mutable.{Map as MutableMap}
+
+import Stats.*
+
+object TestHelper:
+  val noOfSchedules = 10000 // set this to 100k during deployment
+  val readWritesPerThread =
+    20 // maximum number of read/writes possible in one thread
+  val contextSwitchBound = 10
+  val testTimeout = 150 // the total time out for a test in seconds
+  val schedTimeout =
+    15 // the total time out for execution of a schedule in secs
+
+  // Helpers
+  /*def testManySchedules(op1: => Any): Unit = testManySchedules(List(() => op1))
+  def testManySchedules(op1: => Any, op2: => Any): Unit = testManySchedules(List(() => op1, () => op2))
+  def testManySchedules(op1: => Any, op2: => Any, op3: => Any): Unit = testManySchedules(List(() => op1, () => op2, () => op3))
+  def testManySchedules(op1: => Any, op2: => Any, op3: => Any, op4: => Any): Unit = testManySchedules(List(() => op1, () => op2, () => op3, () => op4))*/
+
+  def testSequential[T](
+      ops: Scheduler => Any
+  )(assertions: T => (Boolean, String)) =
+    testManySchedules(
+      1,
+      (sched: Scheduler) =>
+        (
+          List(() => ops(sched)),
+          (res: List[Any]) => assertions(res.head.asInstanceOf[T])
+        )
+    )
+
+  /** @numThreads
+    *   number of threads
+    * @ops
+    *   operations to be executed, one per thread
+    * @assertion
+    *   as condition that will executed after all threads have completed
+    *   (without exceptions) the arguments are the results of the threads
+    */
+  def testManySchedules(
+      numThreads: Int,
+      ops: Scheduler => (
+          List[() => Any], // Threads
+          List[Any] => (Boolean, String)
+      ) // Assertion
+  ) =
+    var timeout = testTimeout * 1000L
+    val threadIds = (1 to numThreads)
+    // (1 to scheduleLength).flatMap(_ => threadIds).toList.permutations.take(noOfSchedules).foreach {
+    val schedules = (new ScheduleGenerator(numThreads)).schedules()
+    var schedsExplored = 0
+    schedules
+      .takeWhile(_ => schedsExplored <= noOfSchedules && timeout > 0)
+      .foreach {
+        // case _ if timeout <= 0 => // break
+        case schedule =>
+          schedsExplored += 1
+          val schedr = new Scheduler(schedule)
+          // println("Exploring Sched: "+schedule)
+          val (threadOps, assertion) = ops(schedr)
+          if threadOps.size != numThreads then
+            throw new IllegalStateException(
+              s"Number of threads: $numThreads, do not match operations of threads: $threadOps"
+            )
+          timed { schedr.runInParallel(schedTimeout * 1000, threadOps) } { t =>
+            timeout -= t
+          } match
+            case Timeout(msg) =>
+              throw new java.lang.AssertionError(
+                "assertion failed\n" + "The schedule took too long to complete. A possible deadlock! \n" + msg
+              )
+            case Except(msg, stkTrace) =>
+              val traceStr = "Thread Stack trace: \n" + stkTrace
+                .map(" at " + _.toString)
+                .mkString("\n")
+              throw new java.lang.AssertionError(
+                "assertion failed\n" + msg + "\n" + traceStr
+              )
+            case RetVal(threadRes) =>
+              // check the assertion
+              val (success, custom_msg) = assertion(threadRes)
+              if !success then
+                val msg =
+                  "The following schedule resulted in wrong results: \n" + custom_msg + "\n" + schedr
+                    .getOperationLog()
+                    .mkString("\n")
+                throw new java.lang.AssertionError("Assertion failed: " + msg)
+      }
+    if timeout <= 0 then
+      throw new java.lang.AssertionError(
+        "Test took too long to complete! Cannot check all schedules as your code is too slow!"
+      )
+
+  /** A schedule generator that is based on the context bound
+    */
+  class ScheduleGenerator(numThreads: Int):
+    val scheduleLength = readWritesPerThread * numThreads
+    val rands =
+      (1 to scheduleLength).map(i =>
+        new Random(0xcafe * i)
+      ) // random numbers for choosing a thread at each position
+    def schedules(): LazyList[List[Int]] =
+      var contextSwitches = 0
+      var contexts =
+        List[Int]() // a stack of thread ids in the order of context-switches
+      val remainingOps = MutableMap[Int, Int]()
+      remainingOps ++= (1 to numThreads).map(i =>
+        (i, readWritesPerThread)
+      ) // num ops remaining in each thread
+      val liveThreads = (1 to numThreads).toSeq.toBuffer
+
+      /** Updates remainingOps and liveThreads once a thread is chosen for a
+        * position in the schedule
+        */
+      def updateState(tid: Int): Unit =
+        val remOps = remainingOps(tid)
+        if remOps == 0 then liveThreads -= tid
+        else remainingOps += (tid -> (remOps - 1))
+      val schedule = rands.foldLeft(List[Int]()) {
+        case (acc, r) if contextSwitches < contextSwitchBound =>
+          val tid = liveThreads(r.nextInt(liveThreads.size))
+          contexts match
+            case prev :: tail
+                if prev != tid => // we have a new context switch here
+              contexts +:= tid
+              contextSwitches += 1
+            case prev :: tail =>
+            case _ => // init case
+              contexts +:= tid
+          updateState(tid)
+          acc :+ tid
+        case (
+              acc,
+              _
+            ) => // here context-bound has been reached so complete the schedule without any more context switches
+          if !contexts.isEmpty then
+            contexts = contexts.dropWhile(remainingOps(_) == 0)
+          val tid = contexts match
+            case top :: tail => top
+            case _ =>
+              liveThreads(
+                0
+              ) // here, there has to be threads that have not even started
+          updateState(tid)
+          acc :+ tid
+      }
+      schedule #:: schedules()