{
"q1": {
"type": "blank",
"question": "\n\n\nTo take advantage of modern CPUs with multiple cores, application\ndevelopers must ____ their ____ programs into separate streams of execution\ncalled ____. Each of these streams of execution co-exist within the same\n____, which means they share the same address space and thus can access the\nsame data. One common use of threads is to ____ I/O with other activities\nwithin a program.\n\n\n"
},
"q2": {
"type": "multiple",
"question": "\n\n\nWhich of the following are true statements about threads (choose all that apply)?\n\n\n",
"responses": {
"registers": "Threads share the same set of registers and program counters.",
"tcb": "The state of each thread is stored in a thread control block.",
"address": "All threads within a process share the same code, data, and heap, but not stack.",
"onethread": "Only one thread per process is allowed to be executed at a time.",
"main": "Every process has at least one thread.",
"unlimited": "A process can have an unlimited number of threads."
}
},
"q3": {
"type": "blank",
"question": "\n\n\nWith processes, we use fork to create a new process, and\nwait to wait for the process to finish. With POSIX threads, we\nuse the ____ function to start a thread and the ____\nfunction to wait for one to finish.\n\nTo guard a critical section of code, we can use a ____ from the POSIX\nthreads library as follows:\n\n\n\n\npthread_mutex_t lock = PTHREAD_MUTEX_INITIALIZER;\n____(&lock); // 3\nx = x + 1; // Critical section\n____(&lock); // 4\n
\n\n\n\nIf we need some sort of signaling between threads, we can use a ____ in the\nfollowing format:\n\n\n\n\npthread_mutex_t lock = PTHREAD_MUTEX_INITIALIZER;\npthread_cond_t cond = PTHREAD_COND_INITIALIZER;\npthread_mutex_lock(&lock);\nwhile (ready == 0)\n ____(&cond, &lock); // 6\npthread_mutex_unlock(&lock);\n
\n\n\n\nTo perform the actual signaling, we would put the following code in\nanother thread:\n\n\n\n\npthread_mutex_lock(&lock);\nready = 1;\n____(&cond); // 7\npthread_mutex_unlock(&lock);\n
\n"
},
"q4": {
"type": "order",
"question": "\n\n\nMatch the following terms with the following definitions (pick the terms in\norder):\n\n\n\n\n\n - Arises if multiple threads attempt to update a shared resource,\n leading to a surprising (and perhaps undesirable) outcome.
\n\n - A piece of code that accesses a shared resource (usually a variable or\n data structure).
\n\n - A primitive that guarantees only a single thread has access to a\n resource.
\n\n
\n",
"responses": {
"critical": "Critical Section",
"mutex": "Mutual Exclusion",
"racecondition": "Race Condition"
}
},
"q5": {
"type": "blank",
"question": "\n\n\nWhen using threads, we can protect a critical section by employing a ____\nvariable. This object is either ____, which means no thread holds the\nobject, or ____, which means exactly one thread (aka the ____) holds the\nobject and is in the critical section.\n\n\n"
},
"q6": {
"type": "blank",
"question": "\nTo evaluate different locking mechanisms, we need to consider three basic criteria:\n\n\n\n- ____: Whether or not the lock actually prevents multiple threads from\nentering a critical section.
\n\n- ____: Whether or not each thread contending for the lock gets a fair\nshot at acquiring it.
\n\n- ____: How much overhead is added by using the locking mechanism.
\n\n
\n\n\n"
},
"q7": {
"type": "multiple",
"question": "\n\nTo implement locks, we need support from the hardware. Which of the\nfollowing statements about different locking mechanisms are true (select\nall that apply)?\n\n\n",
"responses": {
"interrupts": "Although simple, disabling interrupts has numerous problems as a locking mechanism.",
"testandset": "Test-and-Set provides a machine instruction that allows us to test the old value at a memory location and while simultaneously setting the memory location to a new value.",
"compareandswap": "Compare-and-Swap provides a machine instruction that allows us to test a value, update it if it is the expected value, and return the actual value at that memory location.",
"loadlink": "Load-Linked and Store-Conditional instructions work together to atomically fetch a value and update it.",
"fetchandadd": "Fetch-And-Add can be used to implement a ticket lock, which guarantees progress for all threads."
}
},
"q8": {
"type": "multiple",
"question": "\n\nRegarding spin locks, which of the following statements are true (select\nall that apply)?\n\n",
"responses": {
"waiting": "Spin-waiting is inefficient because a thread wastes CPU resources simply waiting for another thread to release a lock",
"fairness": "Spin locks are both correct and guarantee fairness.",
"yielding": "Yielding allows a thread to deschedule itself when it discovers the lock it wants is being held.",
"parking": "Parking involves putting a thread to sleep temporarily and into a queue that will be used to select the next thread to wake up.",
"futex": "A futex is a mutex implemented on the filesystem."
}
},
"q9": {
"type": "blank",
"question": "\n\n\nIn ____ concurrency, we simply wait for something to occur; when it does,\nwe do some small amount of work based on the type of event. This\nprocessing usually happens in a simple construct known as an ____.\n\nOn a Unix system we can use either the ____ or ____ system calls to perform\nnon-blocking or ____ I/O. The advantage of the event-based approach is\nthat we don't need to deal with any ____ normally found in threaded\nprograms.\n\n\n"
}
}