List's node has two fields: data and atomic pointer to the next element next.

List always next fields:

• fake head node head_, which refers to the first informative element
• last informative element last_
• fake tail node tail_, to which refers last_
• number of the concurrent threads n_threads_
• size of the local hazard pointers n_hazard_ptr_
• map of hazard nodes for each proccess hazard_ptrs_
• map of nodes want to delete retired_nodes_
• mutex to make maps protectedthreadMutex

Stack is implemented with two-way adding methods push_front() and push_back() and front deleting pop_front(). Due to the Iterator definition we could make range based loop for our list or output it with output function.

To check ABA problem was implemented swap methods with sleep and yield realisations swapSleep and swapYield respectively

swap()
pop() // Delete A
pop() // Delete B
push(A) // Add A

Deleting is implemented with Hazard pointers. These approach required methods for retiring nodes retire_ and then scanning all hazard pointers' arrays for save removal.

The moment our nodes could become "hazard" occurs while swap operation, so we mark nodes as HP by addToHp method and then make them normal again by deleteFromHp method.

To make --always-make -s ARGS="$1"