A race condition arises when the result depends on
Answer: A
Two tasks reading, modifying and writing the same variable can interleave so that one update is lost, and whether they do depends on when the preemption falls. The bug therefore appears intermittently and often only under load.
Please log in or register to report this question.
A read-modify-write on a shared variable is unsafe because
Answer: B
Correct answer: a preemption between the read and the write loses one update
Incrementing a variable is three operations - read, modify, write - and a preemption between them lets another task read the same old value, so one of the two increments disappears. The remedy is a lock, a critical section, or an atomic instruction, according to how long the sequence is.
Please log in or register to report this question.
Two tasks each execute this on a shared counter, with no protection.
int t = count; t = t + 1; count = t;
Starting from zero, the final value may be
Answer: C
If either task completes all three lines before the other begins, the answer is two; if both read before either writes, one increment is lost and the answer is one. That the answer is not determinate is the whole meaning of a race.
Please log in or register to report this question.
Making that counter volatile would
Answer: D
The qualifier stops the compiler caching the value in a register; it says nothing about a preemption falling between two machine instructions. Believing otherwise is one of the most persistent misconceptions in embedded C.
Please log in or register to report this question.
A memory barrier is needed when
Answer: A
A processor or compiler may complete accesses out of program order, which matters when another core or a peripheral observes them. A barrier forbids the reordering across a chosen point, and it is quite separate from the mutual exclusion a lock provides.
Please log in or register to report this question.
The bounded buffer or producer-consumer pattern needs
Answer: B
One semaphore counts empty slots and blocks the producer when the buffer is full; the other counts full slots and blocks the consumer when it is empty. A mutex is added where several producers or consumers share the structure.
Please log in or register to report this question.
A circular buffer is preferred to a linear one because it
Answer: C
Correct answer: needs no data to be moved when an item is removed
A linear buffer must shuffle its contents down when the front item is taken; a ring simply advances the read index, which is one operation whatever the buffer holds. That makes the cost of insertion and removal constant, which is what an interrupt handler needs.
Please log in or register to report this question.
A circular buffer with one producer and one consumer can often be made lock-free because
Answer: D
The producer advances the write index and reads the read index; the consumer does the reverse. With aligned single-word indices and the right barriers no lock is needed, and adding a second producer destroys the property immediately.
Please log in or register to report this question.
A ring buffer of \( n \) slots is often used with one slot left empty in order to
Answer: A
Correct answer: distinguish the full condition from the empty one
With the indices equal in both cases, full and empty are indistinguishable. Leaving one slot unused means the buffer is full when advancing the write index would reach the read index, and empty when the two are equal - two conditions now told apart, at the cost of one slot. The alternative is to keep a separate count.
Please log in or register to report this question.
A ring buffer declared with 16 slots and the one-empty-slot convention can hold at most
Answer: B
One slot is given up so that full and empty remain distinguishable, so fifteen is the usable capacity. Sizing the array to a power of two lets the index wrap by a mask rather than a division, which is why sixteen was chosen.
Please log in or register to report this question.
