Safely add to a variable, returning the previous value
#include <atomic.h> unsigned atomic_add_value( volatile unsigned * loc, unsigned incr );
Use the -l c option to qcc to link against this library. This library is usually included automatically.
The atomic_add_value() function is a thread-safe way of doing an (*loc) += incr operation, even in a symmetric-multiprocessing system.
|The atomic_add_value() function may be slower than atomic_add().|
The atomic_*() functions are guaranteed to complete without being preempted by another thread. When modifying a variable shared between a thread and an interrupt, you must either disable interrupts or use the atomic_*() functions.
The atomic_*() functions are also useful for modifying variables that are referenced by more than one thread (that aren't necessarily in the same process) without having to use a mutex.
|Perform atomic operations only on objects that were allocated in normal memory mappings. On certain processors (e.g. some PPC ones), atomic operations will cause a fault if the object is allocated in uncached memory.|
The previous value of loc's contents.
To safely increment a counter shared between multiple threads:
#include <atomic.h> … volatile unsigned count; unsigned previous; … previous = atomic_add_value( &count, 1 );
atomic_add(), atomic_clr(), atomic_clr_value(), atomic_set(), atomic_set_value(), atomic_sub(), atomic_sub_value(), atomic_toggle(), atomic_toggle_value()