Synchronization issues in RTOS

In an RTOS, several tasks run concurrently and often share global variables, buffers, functions and I/O devices. If access is not coordinated, the result depends on the order of execution, which causes the following issues.

Issues

1. Race condition: two tasks read-modify-write a shared variable at the same time; the result depends on timing. Example: both tasks do counter = counter + 1 and one update is lost.
2. Critical section problem: the code that accesses a shared resource must run by only one task at a time; otherwise data becomes inconsistent.
3. Non-reentrant functions: a shared function that uses global or static data gives wrong results if a task is pre-empted inside it and another task calls it.
4. Deadlock: tasks wait forever for resources held by each other.
5. Priority inversion: a high-priority task is blocked by a low-priority task holding a lock, while a medium-priority task runs (Mars Pathfinder problem).
6. Starvation: a low-priority task never gets the resource.

Solutions

• Mutual exclusion: protect critical sections with mutex/binary semaphore, or disable interrupts for very short sections.
• Semaphores (counting) to control access to a pool of resources and for signalling.
• Make functions reentrant: use local variables and parameters, not globals.
• Priority inheritance / priority ceiling to avoid priority inversion.
• Resource ordering / request all at once to avoid deadlock.
• Message queues, mailboxes and events for safe communication instead of shared variables.