47.11. Disadvantages of System V Semaphores

System V semaphores have many of the same disadvantages as message queues (Section 46.9), including the following:

• Semaphores are referred to by identifiers, rather than the file descriptors used by most other UNIX I/O and IPC mechanisms. This makes it difficult to perform operations such as simultaneously waiting both on a semaphore and on input from a file descriptor. (It is possible to resolve this difficulty by creating a child process or thread that operates on the semaphore and writes messages to a pipe monitored, along with other file descriptors, using one of the methods described in Chapter 63.)

• The use of keys, rather than filenames, to identify semaphores results in additional programming complexity.

• The use of separate system calls for creating and initializing semaphores means that, in some cases, we must do extra programming work to avoid race conditions when initializing a semaphore.

• The kernel doesn’t maintain a count of the number of processes referring to a semaphore set. This complicates the decision about when it is appropriate to delete a semaphore set and makes it difficult to ensure that an unused set is deleted.

• The programming interface provided by System V is overly complex. In the common case, a program operates on a single semaphore. The ability to simultaneously operate on multiple semaphores in a set is unnecessary.

• There are various limits on the operation of semaphores. These limits are configurable, but if an application operates outside the range of the default limits, this nevertheless requires extra work when installing the application.