Changed SkipListSet/Map<RCU> for new get() semantics (with raw_ptr)

[libcds.git] / cds / container / skip_list_set_rcu.h

//$$CDS-header$$

#ifndef CDSLIB_CONTAINER_SKIP_LIST_SET_RCU_H
#define CDSLIB_CONTAINER_SKIP_LIST_SET_RCU_H

#include <cds/intrusive/skip_list_rcu.h>
#include <cds/container/details/make_skip_list_set.h>

namespace cds { namespace container {

    /// Lock-free skip-list set (template specialization for \ref cds_urcu_desc "RCU")
    /** @ingroup cds_nonintrusive_set
        \anchor cds_nonintrusive_SkipListSet_rcu

        The implementation of well-known probabilistic data structure called skip-list
        invented by W.Pugh in his papers:
            - [1989] W.Pugh Skip Lists: A Probabilistic Alternative to Balanced Trees
            - [1990] W.Pugh A Skip List Cookbook

        A skip-list is a probabilistic data structure that provides expected logarithmic
        time search without the need of rebalance. The skip-list is a collection of sorted
        linked list. Nodes are ordered by key. Each node is linked into a subset of the lists.
        Each list has a level, ranging from 0 to 32. The bottom-level list contains
        all the nodes, and each higher-level list is a sublist of the lower-level lists.
        Each node is created with a random top level (with a random height), and belongs
        to all lists up to that level. The probability that a node has the height 1 is 1/2.
        The probability that a node has the height N is 1/2 ** N (more precisely,
        the distribution depends on an random generator provided, but our generators
        have this property).

        The lock-free variant of skip-list is implemented according to book
            - [2008] M.Herlihy, N.Shavit "The Art of Multiprocessor Programming",
                chapter 14.4 "A Lock-Free Concurrent Skiplist"

        Template arguments:
        - \p RCU - one of \ref cds_urcu_gc "RCU type".
        - \p T - type to be stored in the list.
        - \p Traits - set traits, default is skip_list::traits for explanation.

        It is possible to declare option-based list with cds::container::skip_list::make_traits metafunction istead of \p Traits template
        argument.
        Template argument list \p Options of cds::container::skip_list::make_traits metafunction are:
        - opt::compare - key comparison functor. No default functor is provided.
            If the option is not specified, the opt::less is used.
        - opt::less - specifies binary predicate used for key comparison. Default is \p std::less<T>.
        - opt::item_counter - the type of item counting feature. Default is \ref atomicity::empty_item_counter that is no item counting.
        - opt::memory_model - C++ memory ordering model. Can be opt::v::relaxed_ordering (relaxed memory model, the default)
            or opt::v::sequential_consistent (sequentially consisnent memory model).
        - skip_list::random_level_generator - random level generator. Can be skip_list::xorshift, skip_list::turbo_pascal or
            user-provided one. See skip_list::random_level_generator option description for explanation.
            Default is \p %skip_list::turbo_pascal.
        - opt::allocator - allocator for skip-list node. Default is \ref CDS_DEFAULT_ALLOCATOR.
        - opt::back_off - back-off strategy used. If the option is not specified, the cds::backoff::Default is used.
        - opt::stat - internal statistics. Available types: skip_list::stat, skip_list::empty_stat (the default)
        - opt::rcu_check_deadlock - a deadlock checking policy. Default is opt::v::rcu_throw_deadlock

        @note Before including <tt><cds/container/skip_list_set_rcu.h></tt> you should include appropriate RCU header file,
        see \ref cds_urcu_gc "RCU type" for list of existing RCU class and corresponding header files.

        <b>Iterators</b>

        The class supports a forward iterator (\ref iterator and \ref const_iterator).
        The iteration is ordered.

        You may iterate over skip-list set items only under RCU lock.
        Only in this case the iterator is thread-safe since
        while RCU is locked any set's item cannot be reclaimed.

        The requirement of RCU lock during iterating means that deletion of the elements (i.e. \ref erase)
        is not possible.

        @warning The iterator object cannot be passed between threads

        Example how to use skip-list set iterators:
        \code
        // First, you should include the header for RCU type you have chosen
        #include <cds/urcu/general_buffered.h>
        #include <cds/container/skip_list_set_rcu.h>

        typedef cds::urcu::gc< cds::urcu::general_buffered<> > rcu_type;

        struct Foo {
            // ...
        };

        // Traits for your skip-list.
        // At least, you should define cds::opt::less or cds::opt::compare for Foo struct
        struct my_traits: public cds::continer::skip_list::traits
        {
            // ...
        };
        typedef cds::container::SkipListSet< rcu_type, Foo, my_traits > my_skiplist_set;

        my_skiplist_set theSet;

        // ...

        // Begin iteration
        {
            // Apply RCU locking manually
            typename rcu_type::scoped_lock sl;

            for ( auto it = theList.begin(); it != theList.end(); ++it ) {
                // ...
            }

            // rcu_type::scoped_lock destructor releases RCU lock implicitly
        }
        \endcode

        \warning Due to concurrent nature of skip-list set it is not guarantee that you can iterate
        all elements in the set: any concurrent deletion can exclude the element
        pointed by the iterator from the set, and your iteration can be terminated
        before end of the set. Therefore, such iteration is more suitable for debugging purposes

        The iterator class supports the following minimalistic interface:
        \code
        struct iterator {
            // Default ctor
            iterator();

            // Copy ctor
            iterator( iterator const& s);

            value_type * operator ->() const;
            value_type& operator *() const;

            // Pre-increment
            iterator& operator ++();

            // Copy assignment
            iterator& operator = (const iterator& src);

            bool operator ==(iterator const& i ) const;
            bool operator !=(iterator const& i ) const;
        };
        \endcode
        Note, the iterator object returned by \ref end, \p cend member functions points to \p nullptr and should not be dereferenced.
    */
    template <
        typename RCU,
        typename T,
#ifdef CDS_DOXYGEN_INVOKED
        typename Traits = skip_list::traits
#else
        typename Traits
#endif
    >
    class SkipListSet< cds::urcu::gc< RCU >, T, Traits >:
#ifdef CDS_DOXYGEN_INVOKED
        protected intrusive::SkipListSet< cds::urcu::gc< RCU >, T, Traits >
#else
        protected details::make_skip_list_set< cds::urcu::gc< RCU >, T, Traits >::type
#endif
    {
        //@cond
        typedef details::make_skip_list_set< cds::urcu::gc< RCU >, T, Traits >    maker;
        typedef typename maker::type base_class;
        //@endcond
    public:
        typedef typename base_class::gc gc  ; ///< Garbage collector used
        typedef T       value_type  ;   ///< Value type stored in the set
        typedef Traits  traits     ;   ///< Options specified

        typedef typename base_class::back_off       back_off        ;   ///< Back-off strategy used
        typedef typename traits::allocator         allocator_type  ;   ///< Allocator type used for allocate/deallocate the skip-list nodes
        typedef typename base_class::item_counter   item_counter    ;   ///< Item counting policy used
        typedef typename maker::key_comparator      key_comparator  ;   ///< key compare functor
        typedef typename base_class::memory_model   memory_model    ;   ///< Memory ordering. See cds::opt::memory_model option
        typedef typename traits::random_level_generator random_level_generator ; ///< random level generator
        typedef typename traits::stat              stat            ;   ///< internal statistics type
        typedef typename traits::rcu_check_deadlock    rcu_check_deadlock ; ///< Deadlock checking policy

        //@cond
        typedef cds::container::skip_list::implementation_tag implementation_tag;
        //@endcond

    protected:
        //@cond
        typedef typename maker::node_type           node_type;
        typedef typename maker::node_allocator      node_allocator;

        typedef std::unique_ptr< node_type, typename maker::node_deallocator >    scoped_node_ptr;
        //@endcond

    public:
        typedef typename base_class::rcu_lock  rcu_lock;   ///< RCU scoped lock
        /// Group of \p extract_xxx functions do not require external locking
        static CDS_CONSTEXPR const bool c_bExtractLockExternal = base_class::c_bExtractLockExternal;

        /// pointer to extracted node
        using exempt_ptr = cds::urcu::exempt_ptr< gc, node_type, value_type, typename maker::intrusive_traits::disposer >;

    private:
        //@cond
        struct raw_ptr_converter
        {
            value_type * operator()( node_type * p ) const
            {
               return p ? &p->m_Value : nullptr;
            }

            value_type& operator()( node_type& n ) const
            {
                return n.m_Value;
            }

            value_type const& operator()( node_type const& n ) const
            {
                return n.m_Value;
            }
        };
        //@endcond

    public:
        /// Result of \p get(), \p get_with() functions - pointer to the node found
        typedef cds::urcu::raw_ptr_adaptor< value_type, typename base_class::raw_ptr, raw_ptr_converter > raw_ptr;

    protected:
        //@cond
        unsigned int random_level()
        {
            return base_class::random_level();
        }
        //@endcond

    public:
        /// Default ctor
        SkipListSet()
            : base_class()
        {}

        /// Destructor destroys the set object
        ~SkipListSet()
        {}

    public:
        /// Iterator type
        typedef skip_list::details::iterator< typename base_class::iterator >  iterator;

        /// Const iterator type
        typedef skip_list::details::iterator< typename base_class::const_iterator >   const_iterator;

        /// Returns a forward iterator addressing the first element in a set
        iterator begin()
        {
            return iterator( base_class::begin() );
        }

        /// Returns a forward const iterator addressing the first element in a set
        //@{
        const_iterator begin() const
        {
            return const_iterator( base_class::begin() );
        }
        const_iterator cbegin() const
        {
            return const_iterator( base_class::cbegin() );
        }
        //@}

        /// Returns a forward iterator that addresses the location succeeding the last element in a set.
        iterator end()
        {
            return iterator( base_class::end() );
        }

        /// Returns a forward const iterator that addresses the location succeeding the last element in a set.
        //@{
        const_iterator end() const
        {
            return const_iterator( base_class::end() );
        }
        const_iterator cend() const
        {
            return const_iterator( base_class::cend() );
        }
        //@}

    public:
        /// Inserts new node
        /**
            The function creates a node with copy of \p val value
            and then inserts the node created into the set.

            The type \p Q should contain as minimum the complete key for the node.
            The object of \ref value_type should be constructible from a value of type \p Q.
            In trivial case, \p Q is equal to \ref value_type.

            RCU \p synchronize method can be called. RCU should not be locked.

            Returns \p true if \p val is inserted into the set, \p false otherwise.
        */
        template <typename Q>
        bool insert( Q const& val )
        {
            scoped_node_ptr sp( node_allocator().New( random_level(), val ));
            if ( base_class::insert( *sp.get() )) {
                sp.release();
                return true;
            }
            return false;
        }

        /// Inserts new node
        /**
            The function allows to split creating of new item into two part:
            - create item with key only
            - insert new item into the set
            - if inserting is success, calls  \p f functor to initialize value-fields of \p val.

            The functor signature is:
            \code
                void func( value_type& val );
            \endcode
            where \p val is the item inserted. User-defined functor \p f should guarantee that during changing
            \p val no any other changes could be made on this set's item by concurrent threads.
            The user-defined functor is called only if the inserting is success.

            RCU \p synchronize method can be called. RCU should not be locked.
        */
        template <typename Q, typename Func>
        bool insert( Q const& val, Func f )
        {
            scoped_node_ptr sp( node_allocator().New( random_level(), val ));
            if ( base_class::insert( *sp.get(), [&f]( node_type& val ) { f( val.m_Value ); } )) {
                sp.release();
                return true;
            }
            return false;
        }

        /// Ensures that the item exists in the set
        /**
            The operation performs inserting or changing data with lock-free manner.

            If the \p val key not found in the set, then the new item created from \p val
            is inserted into the set. Otherwise, the functor \p func is called with the item found.
            The functor \p Func should be a function with signature:
            \code
                void func( bool bNew, value_type& item, const Q& val );
            \endcode
            or a functor:
            \code
                struct my_functor {
                    void operator()( bool bNew, value_type& item, const Q& val );
                };
            \endcode

            with arguments:
            - \p bNew - \p true if the item has been inserted, \p false otherwise
            - \p item - item of the set
            - \p val - argument \p key passed into the \p ensure function

            The functor may change non-key fields of the \p item; however, \p func must guarantee
            that during changing no any other modifications could be made on this item by concurrent threads.

            RCU \p synchronize method can be called. RCU should not be locked.

            Returns <tt> std::pair<bool, bool> </tt> where \p first is true if operation is successfull,
            \p second is true if new item has been added or \p false if the item with \p key
            already is in the set.
        */
        template <typename Q, typename Func>
        std::pair<bool, bool> ensure( const Q& val, Func func )
        {
            scoped_node_ptr sp( node_allocator().New( random_level(), val ));
            std::pair<bool, bool> bRes = base_class::ensure( *sp,
                [&func, &val](bool bNew, node_type& node, node_type&){ func( bNew, node.m_Value, val ); });
            if ( bRes.first && bRes.second )
                sp.release();
            return bRes;
        }

        /// Inserts data of type \ref value_type constructed with <tt>std::forward<Args>(args)...</tt>
        /**
            Returns \p true if inserting successful, \p false otherwise.

            RCU \p synchronize method can be called. RCU should not be locked.
        */
        template <typename... Args>
        bool emplace( Args&&... args )
        {
            scoped_node_ptr sp( node_allocator().New( random_level(), std::forward<Args>(args)... ));
            if ( base_class::insert( *sp.get() )) {
                sp.release();
                return true;
            }
            return false;
        }

        /// Delete \p key from the set
        /** \anchor cds_nonintrusive_SkipListSet_rcu_erase_val

            The item comparator should be able to compare the type \p value_type
            and the type \p Q.

            RCU \p synchronize method can be called. RCU should not be locked.

            Return \p true if key is found and deleted, \p false otherwise
        */
        template <typename Q>
        bool erase( Q const& key )
        {
            return base_class::erase( key );
        }

        /// Deletes the item from the set using \p pred predicate for searching
        /**
            The function is an analog of \ref cds_nonintrusive_SkipListSet_rcu_erase_val "erase(Q const&)"
            but \p pred is used for key comparing.
            \p Less functor has the interface like \p std::less.
            \p Less must imply the same element order as the comparator used for building the set.
        */
        template <typename Q, typename Less>
        bool erase_with( Q const& key, Less pred )
        {
            CDS_UNUSED( pred );
            return base_class::erase_with( key, cds::details::predicate_wrapper< node_type, Less, typename maker::value_accessor >());
        }

        /// Delete \p key from the set
        /** \anchor cds_nonintrusive_SkipListSet_rcu_erase_func

            The function searches an item with key \p key, calls \p f functor
            and deletes the item. If \p key is not found, the functor is not called.

            The functor \p Func interface:
            \code
            struct extractor {
                void operator()(value_type const& val);
            };
            \endcode

            Since the key of MichaelHashSet's \p value_type is not explicitly specified,
            template parameter \p Q defines the key type searching in the list.
            The list item comparator should be able to compare the type \p T of list item
            and the type \p Q.

            RCU \p synchronize method can be called. RCU should not be locked.

            Return \p true if key is found and deleted, \p false otherwise

            See also: \ref erase
        */
        template <typename Q, typename Func>
        bool erase( Q const& key, Func f )
        {
            return base_class::erase( key, [&f]( node_type const& node) { f( node.m_Value ); } );
        }

        /// Deletes the item from the set using \p pred predicate for searching
        /**
            The function is an analog of \ref cds_nonintrusive_SkipListSet_rcu_erase_func "erase(Q const&, Func)"
            but \p pred is used for key comparing.
            \p Less functor has the interface like \p std::less.
            \p Less must imply the same element order as the comparator used for building the set.
        */
        template <typename Q, typename Less, typename Func>
        bool erase_with( Q const& key, Less pred, Func f )
        {
            CDS_UNUSED( pred );
            return base_class::erase_with( key, cds::details::predicate_wrapper< node_type, Less, typename maker::value_accessor >(),
                [&f]( node_type const& node) { f( node.m_Value ); } );
        }

        /// Extracts the item from the set with specified \p key
        /** \anchor cds_nonintrusive_SkipListSet_rcu_extract
            The function searches an item with key equal to \p key in the set,
            unlinks it from the set, and returns \ref cds::urcu::exempt_ptr "exempt_ptr" pointer to the item found.
            If the item is not found the function returns an empty \p exempt_ptr

            Note the compare functor from \p Traits class' template argument
            should accept a parameter of type \p Q that can be not the same as \p value_type.

            RCU \p synchronize method can be called. RCU should NOT be locked.

            The function does not free the item found.
            The item will be implicitly freed when the returned object is destroyed or when
            its \p release() member function is called.
        */
        template <typename Q>
        exempt_ptr extract( Q const& key )
        {
            return exempt_ptr( base_class::do_extract( key ));
        }

        /// Extracts the item from the set with comparing functor \p pred
        /**
            The function is an analog of \p extract(Q const&) but \p pred predicate is used for key comparing.
            \p Less has the semantics like \p std::less.
            \p pred must imply the same element order as the comparator used for building the set.
        */
        template <typename Q, typename Less>
        exempt_ptr extract_with( Q const& key, Less pred )
        {
            CDS_UNUSED( pred );
            return exempt_ptr( base_class::do_extract_with( key, cds::details::predicate_wrapper< node_type, Less, typename maker::value_accessor >()));
        }

        /// Extracts an item with minimal key from the set
        /**
            The function searches an item with minimal key, unlinks it,
            and returns \ref cds::urcu::exempt_ptr "exempt_ptr" pointer to the item.
            If the skip-list is empty the function returns an empty \p exempt_ptr.

            RCU \p synchronize method can be called. RCU should NOT be locked.

            The function does not free the item found.
            The item will be implicitly freed when the returned object is destroyed or when
            its \p release() member function is called.
        */
        exempt_ptr extract_min()
        {
            return exempt_ptr( base_class::do_extract_min());
        }

        /// Extracts an item with maximal key from the set
        /**
            The function searches an item with maximal key, unlinks it from the set,
            and returns \ref cds::urcu::exempt_ptr "exempt_ptr" pointer to the item.
            If the skip-list is empty the function returns an empty \p exempt_ptr.

            RCU \p synchronize method can be called. RCU should NOT be locked.

            The function does not free the item found.
            The item will be implicitly freed when the returned object is destroyed or when
            its \p release() member function is called.
        */
        exempt_ptr extract_max()
        {
            return exempt_ptr( base_class::do_extract_max());
        }

        /// Find the key \p val
        /**
            @anchor cds_nonintrusive_SkipListSet_rcu_find_func

            The function searches the item with key equal to \p val and calls the functor \p f for item found.
            The interface of \p Func functor is:
            \code
            struct functor {
                void operator()( value_type& item, Q& val );
            };
            \endcode
            where \p item is the item found, \p val is the <tt>find</tt> function argument.

            The functor may change non-key fields of \p item. Note that the functor is only guarantee
            that \p item cannot be disposed during functor is executing.
            The functor does not serialize simultaneous access to the set's \p item. If such access is
            possible you must provide your own synchronization schema on item level to exclude unsafe item modifications.

            The \p val argument is non-const since it can be used as \p f functor destination i.e., the functor
            can modify both arguments.

            Note the hash functor specified for class \p Traits template parameter
            should accept a parameter of type \p Q that may be not the same as \p value_type.

            The function applies RCU lock internally.

            The function returns \p true if \p val is found, \p false otherwise.
        */
        template <typename Q, typename Func>
        bool find( Q& val, Func f )
        {
            return base_class::find( val, [&f]( node_type& node, Q& v ) { f( node.m_Value, v ); });
        }
        //@cond
        template <typename Q, typename Func>
        bool find( Q const& val, Func f )
        {
            return base_class::find( val, [&f]( node_type& node, Q& v ) { f( node.m_Value, v ); } );
        }
        //@endcond

        /// Finds the key \p val using \p pred predicate for searching
        /**
            The function is an analog of \ref cds_nonintrusive_SkipListSet_rcu_find_func "find(Q&, Func)"
            but \p pred is used for key comparing.
            \p Less functor has the interface like \p std::less.
            \p Less must imply the same element order as the comparator used for building the set.
        */
        template <typename Q, typename Less, typename Func>
        bool find_with( Q& val, Less pred, Func f )
        {
            CDS_UNUSED( pred );
            return base_class::find_with( val, cds::details::predicate_wrapper< node_type, Less, typename maker::value_accessor >(),
                [&f]( node_type& node, Q& v ) { f( node.m_Value, v ); } );
        }
        //@cond
        template <typename Q, typename Less, typename Func>
        bool find_with( Q const& val, Less pred, Func f )
        {
            CDS_UNUSED( pred );
            return base_class::find_with( val, cds::details::predicate_wrapper< node_type, Less, typename maker::value_accessor >(),
                                          [&f]( node_type& node, Q& v ) { f( node.m_Value, v ); } );
        }
        //@endcond

        /// Find the key \p val
        /** @anchor cds_nonintrusive_SkipListSet_rcu_find_val

            The function searches the item with key equal to \p val
            and returns \p true if it is found, and \p false otherwise.

            Note the hash functor specified for class \p Traits template parameter
            should accept a parameter of type \p Q that may be not the same as \ref value_type.

            The function applies RCU lock internally.
        */
        template <typename Q>
        bool find( Q const & val )
        {
            return base_class::find( val );
        }

        /// Finds the key \p val using \p pred predicate for searching
        /**
            The function is an analog of \ref cds_nonintrusive_SkipListSet_rcu_find_val "find(Q const&)"
            but \p pred is used for key comparing.
            \p Less functor has the interface like \p std::less.
            \p Less must imply the same element order as the comparator used for building the set.
        */
        template <typename Q, typename Less>
        bool find_with( Q const& val, Less pred )
        {
            CDS_UNUSED( pred );
            return base_class::find_with( val, cds::details::predicate_wrapper< node_type, Less, typename maker::value_accessor >());
        }

        /// Finds \p key and return the item found
        /** \anchor cds_nonintrusive_SkipListSet_rcu_get
            The function searches the item with key equal to \p key and returns a \p raw_ptr object pointed to item found.
            If \p key is not found it returns empty \p raw_ptr.

            Note the compare functor in \p Traits class' template argument
            should accept a parameter of type \p Q that can be not the same as \p value_type.

            RCU should be locked before call of this function.
            Returned item is valid only while RCU is locked:
            \code
            typedef cds::container::SkipListSet< cds::urcu::gc< cds::urcu::general_buffered<> >, foo, my_traits > skip_list;
            skip_list theList;
            // ...
            typename skip_list::raw_ptr pVal;
            {
                // Lock RCU
                skip_list::rcu_lock lock;

                pVal = theList.get( 5 );
                if ( pVal ) {
                    // Deal with pVal
                    //...
                }
            }
            // You can manually release pVal after RCU-locked section
            pVal.release();
            \endcode
        */
        template <typename Q>
        raw_ptr get( Q const& key )
        {
            return raw_ptr( base_class::get( key ));
        }

        /// Finds the key \p val and return the item found
        /**
            The function is an analog of \ref cds_nonintrusive_SkipListSet_rcu_get "get(Q const&)"
            but \p pred is used for comparing the keys.

            \p Less functor has the semantics like \p std::less but should take arguments of type \ref value_type and \p Q
            in any order.
            \p pred must imply the same element order as the comparator used for building the set.
        */
        template <typename Q, typename Less>
        raw_ptr get_with( Q const& val, Less pred )
        {
            CDS_UNUSED( pred );
            return raw_ptr( base_class::get_with( val, cds::details::predicate_wrapper< node_type, Less, typename maker::value_accessor >() ));
        }

        /// Clears the set (non-atomic).
        /**
            The function deletes all items from the set.
            The function is not atomic, thus, in multi-threaded environment with parallel insertions
            this sequence
            \code
            set.clear();
            assert( set.empty() );
            \endcode
            the assertion could be raised.

            For each item the \ref disposer provided by \p Traits template parameter will be called.
        */
        void clear()
        {
            base_class::clear();
        }

        /// Checks if the set is empty
        bool empty() const
        {
            return base_class::empty();
        }

        /// Returns item count in the set
        /**
            The value returned depends on item counter type provided by \p Traits template parameter.
            If it is atomicity::empty_item_counter this function always returns 0.
            Therefore, the function is not suitable for checking the set emptiness, use \ref empty
            member function for this purpose.
        */
        size_t size() const
        {
            return base_class::size();
        }

        /// Returns const reference to internal statistics
        stat const& statistics() const
        {
            return base_class::statistics();
        }
    };

}} // namespace cds::container

#endif // #ifndef CDSLIB_CONTAINER_SKIP_LIST_SET_RCU_H