Page Comparison

A semaphore was originally a mechanical signaling mechanism. The railroad industry used the device to provide a form of mutual exclusion for railroads tracks shared by more than one train. In this form, the semaphore signaled trains by closing a set of mechanical arms to block a train from a section of track that was currently in use. When the track became available, the arm would swing up and the waiting train would then proceed.

The notion of using a semaphore in software as a means of mutual exclusion was invented by the Dutch computer scientist Edgser Dijkstra in 1959. In computer software, a semaphore is a protocol mechanism offered by most multitasking kernels. Semaphores, originally used to control access to shared resources, but now they are used for synchronization as described in Synchronization. However, it is useful to describe how semaphores can be used to share resources. The pitfalls of semaphores will be discussed in a later section.

A semaphore was originally a “lock mechanism” and code acquired the key to this lock to continue execution. Acquiring the key means that the executing task has permission to enter the section of otherwise locked code. Entering a section of locked code causes the task to wait until the key becomes available.

Typically, two types of semaphores exist: binary semaphores and counting semaphores. As its name implies, a binary semaphore can only take two values: 0 or 1. A counting semaphore allows for values between 0 and 255, 65,535, or 4,294,967,295, depending on whether the semaphore mechanism is implemented using 8, 16, or 32 bits, respectively. For µC/OS-III, the maximum value of a semaphore is determined by the data type OS_SEM_CTR (see os_type.h), which can be changed as needed. Along with the semaphore’s value, µC/OS-III also keeps track of tasks waiting for the semaphore’s availability.

Only tasks are allowed to use semaphores when semaphores are used for sharing resources; ISRs are not allowed.

A semaphore is a kernel object defined by the OS_SEM data type, which is defined by the structure os_sem (see os.h). The application can have any number of semaphores (limited only by the amount of RAM available).

There are a number of operations the application is able to perform on semaphores, summarized in Table 13-2. In this chapter, only three functions used most often are discussed: OSSemCreate(), OSSemPend(), and OSSemPost(). Other functions are described in Appendix A, “µC/OS-III API Reference”. When semaphores are used for sharing resources, every semaphore function must be called from a task and never from an ISR. The same limitation does not apply when using semaphores for signaling, as described later in Chapter 13.

...

1. 1. Anchor
      1092499 1092499

      Semaphore Management API

      1. Anchor
         1080238 1080238

         In µC/OS-III, there is no “accept” API since this feature is built into the OSSemPend() by specifying the OS_OPT_PEND_NON_BLOCKING option.

      2. Anchor
         1080239 1080239

         In µC/OS-II, OSSemCreate() returns the address of an OS_EVENT, which is used as the “handle” to the semaphore. In µC/OS-III, the application must allocate storage for an OS_SEM object, which serves the same purpose as the OS_EVENT. The benefit in µC/OS-III is that it is not necessary to predetermine the number of semaphores at compile time.

      3. Anchor
         1080240 1080240

         µC/OS-III returns additional information when a semaphore is signaled. The ISR or task that signals the semaphore takes a snapshot of the current timestamp and stores this in the OS_SEM object signaled. The receiver of the signal therefore knows when the signal was sent.

      4. Anchor
         1080241 1080241

         µC/OS-III does not provide query services, as they were rarely used.