src/test/test_sequencer_node.cpp

/*
    Copyright (c) 2005-2020 Intel Corporation

    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.
    You may obtain a copy of the License at

        http://www.apache.org/licenses/LICENSE-2.0

    Unless required by applicable law or agreed to in writing, software
    distributed under the License is distributed on an "AS IS" BASIS,
    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    See the License for the specific language governing permissions and
    limitations under the License.
*/

#define TBB_DEPRECATED_FLOW_NODE_EXTRACTION __TBB_CPF_BUILD
#define TBB_DEPRECATED_FLOW_NODE_ALLOCATOR __TBB_CPF_BUILD

#include "harness.h"
#include "harness_graph.h"

#include "tbb/flow_graph.h"
#include "tbb/task_scheduler_init.h"
#include "tbb/tick_count.h"
#include "tbb/atomic.h"
#include "test_follows_and_precedes_api.h"

#include <cstdio>

#define N 1000
#define C 10

template< typename T >
struct seq_inspector {
    size_t operator()(const T &v) const { return size_t(v); }
};

template< typename T >
bool wait_try_get( tbb::flow::graph &g, tbb::flow::sequencer_node<T> &q, T &value ) {
    g.wait_for_all();
    return q.try_get(value);
}

template< typename T >
void spin_try_get( tbb::flow::queue_node<T> &q, T &value ) {
    while ( q.try_get(value) != true ) ;
}

template< typename T >
struct parallel_puts : NoAssign {

    tbb::flow::sequencer_node<T> &my_q;
    int my_num_threads;

    parallel_puts( tbb::flow::sequencer_node<T> &q, int num_threads ) : my_q(q), my_num_threads(num_threads) {}

    void operator()(int tid) const {
        for (int j = tid; j < N; j+=my_num_threads) {
            bool msg = my_q.try_put( T(j) );
            ASSERT( msg == true, NULL );
        }
    }

};

template< typename T >
struct touches {

    bool **my_touches;
    T *my_last_touch;
    int my_num_threads;

    touches( int num_threads ) : my_num_threads(num_threads) {
        my_last_touch = new T[my_num_threads];
        my_touches = new bool* [my_num_threads];
        for ( int p = 0; p < my_num_threads; ++p) {
            my_last_touch[p] = T(-1);
            my_touches[p] = new bool[N];
            for ( int n = 0; n < N; ++n)
                my_touches[p][n] = false;
        }
    }

    ~touches() {
        for ( int p = 0; p < my_num_threads; ++p) {
            delete [] my_touches[p];
        }
        delete [] my_touches;
        delete [] my_last_touch;
    }

    bool check( int tid, T v ) {
        if ( my_touches[tid][v] != false ) {
            printf("Error: value seen twice by local thread\n");
            return false;
        }
        if ( v <= my_last_touch[tid] ) {
            printf("Error: value seen in wrong order by local thread\n");
            return false;
        }
        my_last_touch[tid] = v;
        my_touches[tid][v] = true;
        return true;
    }

    bool validate_touches() {
        bool *all_touches = new bool[N];
        for ( int n = 0; n < N; ++n)
            all_touches[n] = false;

        for ( int p = 0; p < my_num_threads; ++p) {
            for ( int n = 0; n < N; ++n) {
                if ( my_touches[p][n] == true ) {
                    ASSERT( all_touches[n] == false, "value see by more than one thread\n" );
                    all_touches[n] = true;
                }
            }
        }
        for ( int n = 0; n < N; ++n) {
            if ( !all_touches[n] )
                printf("No touch at %d, my_num_threads = %d\n", n, my_num_threads);
            //ASSERT( all_touches[n] == true, "value not seen by any thread\n" );
        }
        delete [] all_touches;
        return true;
    }

};

template< typename T >
struct parallel_gets : NoAssign {

    tbb::flow::sequencer_node<T> &my_q;
    int my_num_threads;
    touches<T> &my_touches;

    parallel_gets( tbb::flow::sequencer_node<T> &q, int num_threads, touches<T> &t ) : my_q(q), my_num_threads(num_threads), my_touches(t) {}

    void operator()(int tid) const {
        for (int j = tid; j < N; j+=my_num_threads) {
            T v;
            spin_try_get( my_q, v );
            my_touches.check( tid, v );
        }
    }

};

template< typename T >
struct parallel_put_get : NoAssign {

    tbb::flow::sequencer_node<T> &my_s1;
    tbb::flow::sequencer_node<T> &my_s2;
    int my_num_threads;
    tbb::atomic< int > &my_counter;
    touches<T> &my_touches;

    parallel_put_get( tbb::flow::sequencer_node<T> &s1, tbb::flow::sequencer_node<T> &s2, int num_threads,
                      tbb::atomic<int> &counter, touches<T> &t ) : my_s1(s1), my_s2(s2), my_num_threads(num_threads), my_counter(counter), my_touches(t) {}

    void operator()(int tid) const {
        int i_start = 0;

        while ( (i_start = my_counter.fetch_and_add(C)) < N ) {
            int i_end = ( N < i_start + C ) ? N : i_start + C;
            for (int i = i_start; i < i_end; ++i) {
                bool msg = my_s1.try_put( T(i) );
                ASSERT( msg == true, NULL );
            }

            for (int i = i_start; i < i_end; ++i) {
                T v;
                spin_try_get( my_s2, v );
                my_touches.check( tid, v );
            }
        }
    }

};

//
// Tests
//
// multiple parallel senders, multiple receivers, properly sequenced (relative to receiver) at output
// chained sequencers, multiple parallel senders, multiple receivers, properly sequenced (relative to receiver) at output
//

template< typename T >
int test_parallel(int num_threads) {
    tbb::flow::graph g;

    tbb::flow::sequencer_node<T> s(g, seq_inspector<T>());
    NativeParallelFor( num_threads, parallel_puts<T>(s, num_threads) );
    {
        touches<T> t( num_threads );
        NativeParallelFor( num_threads, parallel_gets<T>(s, num_threads, t) );
        g.wait_for_all();
        ASSERT( t.validate_touches(), NULL );
    }
    T bogus_value(-1);
    T j = bogus_value;
    ASSERT( s.try_get( j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );
    g.wait_for_all();

    tbb::flow::sequencer_node<T> s1(g, seq_inspector<T>());
    tbb::flow::sequencer_node<T> s2(g, seq_inspector<T>());
    tbb::flow::sequencer_node<T> s3(g, seq_inspector<T>());
    tbb::flow::make_edge( s1, s2 );
    tbb::flow::make_edge( s2, s3 );

    {
        touches<T> t( num_threads );
        tbb::atomic<int> counter;
        counter = 0;
        NativeParallelFor( num_threads, parallel_put_get<T>(s1, s3, num_threads, counter, t) );
        g.wait_for_all();
        t.validate_touches();
    }
    g.wait_for_all();
    ASSERT( s1.try_get( j ) == false, NULL );
    g.wait_for_all();
    ASSERT( s2.try_get( j ) == false, NULL );
    g.wait_for_all();
    ASSERT( s3.try_get( j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );

    // test copy constructor
    tbb::flow::sequencer_node<T> s_copy(s);
    NativeParallelFor( num_threads, parallel_puts<T>(s_copy, num_threads) );
    for (int i = 0; i < N; ++i) {
        j = bogus_value;
        spin_try_get( s_copy, j );
        ASSERT( i == j, NULL );
    }
    j = bogus_value;
    g.wait_for_all();
    ASSERT( s_copy.try_get( j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );

    return 0;
}


//
// Tests
//
// No predecessors can be registered
// Request from empty buffer fails
// In-order puts, single sender, single receiver, properly sequenced at output
// Reverse-order puts, single sender, single receiver, properly sequenced at output
// Chained sequencers (3), in-order and reverse-order tests, properly sequenced at output
//

template< typename T >
int test_serial() {
    tbb::flow::graph g;
    T bogus_value(-1);

    tbb::flow::sequencer_node<T> s(g, seq_inspector<T>());
    tbb::flow::sequencer_node<T> s2(g, seq_inspector<T>());
    T j = bogus_value;

    //
    // Rejects attempts to add / remove predecessor
    // Rejects request from empty Q
    //
    ASSERT( s.register_predecessor( s2 ) == false, NULL );
    ASSERT( s.remove_predecessor( s2 ) == false, NULL );
    ASSERT( s.try_get( j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );

    //
    // In-order simple puts and gets
    //

    for (int i = 0; i < N; ++i) {
        bool msg = s.try_put( T(i) );
        ASSERT( msg == true, NULL );
        ASSERT(!s.try_put( T(i) ), NULL);  // second attempt to put should reject
    }


    for (int i = 0; i < N; ++i) {
        j = bogus_value;
        ASSERT(wait_try_get( g, s, j ) == true, NULL);
        ASSERT( i == j, NULL );
        ASSERT(!s.try_put( T(i) ),NULL );  // after retrieving value, subsequent put should fail
    }
    j = bogus_value;
    g.wait_for_all();
    ASSERT( s.try_get( j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );

    //
    // Reverse-order simple puts and gets
    //

    for (int i = N-1; i >= 0; --i) {
        bool msg = s2.try_put( T(i) );
        ASSERT( msg == true, NULL );
    }

    for (int i = 0; i < N; ++i) {
        j = bogus_value;
        ASSERT(wait_try_get( g, s2, j ) == true, NULL);
        ASSERT( i == j, NULL );
    }
    j = bogus_value;
    g.wait_for_all();
    ASSERT( s2.try_get( j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );

    //
    // Chained in-order simple puts and gets
    //

    tbb::flow::sequencer_node<T> s3(g, seq_inspector<T>());
    tbb::flow::sequencer_node<T> s4(g, seq_inspector<T>());
    tbb::flow::sequencer_node<T> s5(g, seq_inspector<T>());
    tbb::flow::make_edge( s3, s4 );
    tbb::flow::make_edge( s4, s5 );

    for (int i = 0; i < N; ++i) {
        bool msg = s3.try_put( T(i) );
        ASSERT( msg == true, NULL );
    }

    for (int i = 0; i < N; ++i) {
        j = bogus_value;
        ASSERT(wait_try_get( g, s5, j ) == true, NULL);
        ASSERT( i == j, NULL );
    }
    j = bogus_value;
    ASSERT( wait_try_get( g, s3, j ) == false, NULL );
    ASSERT( wait_try_get( g, s4, j ) == false, NULL );
    ASSERT( wait_try_get( g, s5, j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );

    g.wait_for_all();
    tbb::flow::remove_edge( s3, s4 );
    ASSERT( s3.try_put( N ) == true, NULL );
    ASSERT( wait_try_get( g, s4, j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );
    ASSERT( wait_try_get( g, s5, j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );
    ASSERT( wait_try_get( g, s3, j ) == true, NULL );
    ASSERT( j == N, NULL );

    //
    // Chained reverse-order simple puts and gets
    //

    tbb::flow::sequencer_node<T> s6(g, seq_inspector<T>());
    tbb::flow::sequencer_node<T> s7(g, seq_inspector<T>());
    tbb::flow::sequencer_node<T> s8(g, seq_inspector<T>());
    tbb::flow::make_edge( s6, s7 );
    tbb::flow::make_edge( s7, s8 );

    for (int i = N-1; i >= 0; --i) {
        bool msg = s6.try_put( T(i) );
        ASSERT( msg == true, NULL );
    }

    for (int i = 0; i < N; ++i) {
        j = bogus_value;
        ASSERT( wait_try_get( g, s8, j ) == true, NULL );
        ASSERT( i == j, NULL );
    }
    j = bogus_value;
    ASSERT( wait_try_get( g, s6, j ) == false, NULL );
    ASSERT( wait_try_get( g, s7, j ) == false, NULL );
    ASSERT( wait_try_get( g, s8, j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );

    g.wait_for_all();
    tbb::flow::remove_edge( s6, s7 );
    ASSERT( s6.try_put( N ) == true, NULL );
    ASSERT( wait_try_get( g, s7, j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );
    ASSERT( wait_try_get( g, s8, j ) == false, NULL );
    ASSERT( j == bogus_value, NULL );
    ASSERT( wait_try_get( g, s6, j ) == true, NULL );
    ASSERT( j == N, NULL );

    return 0;
}

#if __TBB_PREVIEW_FLOW_GRAPH_NODE_SET
#include <array>
#include <vector>
void test_follows_and_precedes_api() {
    std::array<int, 3> messages_for_follows = { {0, 1, 2} };
    std::vector<int> messages_for_precedes = {0, 1, 2};

    follows_and_precedes_testing::test_follows
        <int, tbb::flow::sequencer_node<int>>
        (messages_for_follows, [](const int& i) { return i; });

    follows_and_precedes_testing::test_precedes
        <int, tbb::flow::sequencer_node<int>>
        (messages_for_precedes, [](const int& i) { return i; });
}
#endif

#if __TBB_CPP17_DEDUCTION_GUIDES_PRESENT
template <typename Body>
void test_deduction_guides_common(Body body) {
    using namespace tbb::flow;
    graph g;
    broadcast_node<int> br(g);

    sequencer_node s1(g, body);
    static_assert(std::is_same_v<decltype(s1), sequencer_node<int>>);

#if __TBB_PREVIEW_FLOW_GRAPH_NODE_SET
    sequencer_node s2(follows(br), body);
    static_assert(std::is_same_v<decltype(s2), sequencer_node<int>>);
#endif

    sequencer_node s3(s1);
    static_assert(std::is_same_v<decltype(s3), sequencer_node<int>>);
}

int sequencer_body_f(const int&) { return 1; }

void test_deduction_guides() {
    test_deduction_guides_common([](const int&)->int { return 1; });
    test_deduction_guides_common([](const int&) mutable ->int { return 1; });
    test_deduction_guides_common(sequencer_body_f);
}
#endif

#if TBB_DEPRECATED_FLOW_NODE_ALLOCATOR
void test_node_allocator() {
    tbb::flow::graph g;
    tbb::flow::sequencer_node< int, std::allocator<int> > tmp(g, seq_inspector<int>());
}
#endif

int TestMain() {
    tbb::tick_count start = tbb::tick_count::now(), stop;
    for (int p = 2; p <= 4; ++p) {
        tbb::task_scheduler_init init(p);
        test_serial<int>();
        test_parallel<int>(p);
    }
#if __TBB_PREVIEW_FLOW_GRAPH_NODE_SET
    test_follows_and_precedes_api();
#endif
#if __TBB_CPP17_DEDUCTION_GUIDES_PRESENT
    test_deduction_guides();
#endif
#if TBB_DEPRECATED_FLOW_NODE_EXTRACTION
    test_buffer_extract<tbb::flow::sequencer_node<int> >().run_tests();
#endif
#if TBB_DEPRECATED_FLOW_NODE_ALLOCATOR
    test_node_allocator();
#endif
    stop = tbb::tick_count::now();
    REMARK("Sequencer_Node Time=%6.6f\n", (stop-start).seconds());
    return Harness::Done;
}