Nonreentrant

Nonreentrant describes a function or routine that cannot safely be entered again before the first invocation has completed. In practice, if a nonreentrant function is interrupted (by a signal, an interrupt handler, or another thread) and called again, it may corrupt data, return incorrect results, or crash. Reentrancy is closely related to, but not identical to, thread safety; a reentrant function can be safely interrupted and re-entered, whereas a nonreentrant one is not.

Several common causes of nonreentrancy include the use of static or global state that is modified during execution, reliance on non-thread-safe library calls, and the use of static buffers or stateful resources. Functions that maintain internal pointers, buffers, or counters without ensuring isolation between invocations are typical examples. Interacting with shared I/O devices or memory without proper synchronization can also create nonreentrancy.

Common examples include certain string-tokenization routines, memory allocators, or I/O wrappers that hold internal state. In C, for instance, strtok is nonreentrant because it uses a static pointer to track its position; strtok_r provides a reentrant alternative. Nonreentrancy is a concern in both multithreaded programs and systems with interrupt handling, where a running function could be interrupted and subsequently invoked again.

Mitigations focus on designing functions to be stateless or to operate on per-call context. Approaches include removing internal static data, passing explicit context objects, allocating memory from the caller, using reentrant or thread-safe library variants, and employing appropriate synchronization when shared resources must be accessed concurrently.