SMR-required
Group of algorithms that require safe memory reclamation.
Users are expected to couple the usage of these algorithms with an SMR scheme from vsync/smr e.g. vsync/smr/gdump.h . Detached/retired nodes of these algorithms must be retired to the SMR and not freed/recycled directly. Operations of these algorithms should be called within the critical section of the SMR within enter/exit, unless stated otherwise.
File Index
| File | Description | Linearizable | Lock-free | SMR-required | Unbounded-Queue |
|---|---|---|---|---|---|
| vsync/map/hashtable_standard.h | This is a lock-free listset based hashtable. | ✔ | ✔ | ✔ | ❌ |
| vsync/map/listset_coarse.h | This implementation of listset uses one lock to serialize all operations. | ✔ | ❌ | ✔ | ❌ |
| vsync/map/listset_fine.h | This implementation of listset uses fine grained locking. | ✔ | ❌ | ✔ | ❌ |
| vsync/map/listset_lazy.h | This is a partially parallel implementation with lock-free get. | ✔ | ❌ | ✔ | ❌ |
| vsync/map/listset_lf.h | Lock-free implementation of listset. | ✔ | ✔ | ✔ | ❌ |
| vsync/map/listset_opt.h | This implementation is an optimized verison of listset_fine. | ✔ | ❌ | ✔ | ❌ |
| vsync/map/skiplist_lf.h | Lock-free concurrent skiplist. | ✔ | ✔ | ✔ | ❌ |
| vsync/queue/unbounded_queue_lf.h | Lock-free unbounded queue. | ✔ | ✔ | ✔ | ✔ |
| vsync/queue/vqueue_prio_skiplist_based.h | Skiplist-based priority queue implementation. | ✔ | ✔ | ✔ | ✔ |
| vsync/queue/vqueue_prio_stack_array_based.h | Array-based bounded priority queue. | ✔ | ✔ | ✔ | ✔ |
| vsync/stack/elimination_stack.h | Unbounded lock-free stack with elimination backoff. | ✔ | ✔ | ✔ | ❌ |
| vsync/stack/xbo_stack.h | Unbounded lock-free stack with exponential backoff. | ✔ | ✔ | ✔ | ❌ |