Embedded Systems Architecture

Limiting the heap

In all the allocation functions seen so far, there is no limit imposed by the software on the amount of memory reserved for the heap. While overflowing the stack is often hard to prevent, and extremely difficult to recover from, running out of available heap memory can more often be gracefully handled by the application, for example, by canceling or postponing the operation that required the allocation. In more complex multi-threaded systems, the operating system could actively react to the memory shortage by terminating non-vital processes to free up memory for new allocations. Some advanced systems using page-swapping mechanisms, like Linux, may implement overcommit on the available memory. This mechanism guarantees that ...