src/parsec/disk-image/parsec/parsec-benchmark/pkgs/libs/tbblib/src/src/test/test_mutex.cpp

/*
    Copyright 2005-2010 Intel Corporation.  All Rights Reserved.

    This file is part of Threading Building Blocks.

    Threading Building Blocks is free software; you can redistribute it
    and/or modify it under the terms of the GNU General Public License
    version 2 as published by the Free Software Foundation.

    Threading Building Blocks is distributed in the hope that it will be
    useful, but WITHOUT ANY WARRANTY; without even the implied warranty
    of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.

    You should have received a copy of the GNU General Public License
    along with Threading Building Blocks; if not, write to the Free Software
    Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA

    As a special exception, you may use this file as part of a free software
    library without restriction.  Specifically, if other files instantiate
    templates or use macros or inline functions from this file, or you compile
    this file and link it with other files to produce an executable, this
    file does not by itself cause the resulting executable to be covered by
    the GNU General Public License.  This exception does not however
    invalidate any other reasons why the executable file might be covered by
    the GNU General Public License.
*/

//------------------------------------------------------------------------
// Test TBB mutexes when used with parallel_for.h
//
// Usage: test_Mutex.exe [-v] nthread
//
// The -v option causes timing information to be printed.
//
// Compile with _OPENMP and -openmp
//------------------------------------------------------------------------
#include "tbb/spin_mutex.h"
#include "tbb/critical_section.h"
#include "tbb/spin_rw_mutex.h"
#include "tbb/queuing_rw_mutex.h"
#include "tbb/queuing_mutex.h"
#include "tbb/mutex.h"
#include "tbb/recursive_mutex.h"
#include "tbb/null_mutex.h"
#include "tbb/null_rw_mutex.h"
#include "tbb/parallel_for.h"
#include "tbb/blocked_range.h"
#include "tbb/tick_count.h"
#include "tbb/atomic.h"
#include "harness.h"
#include <cstdlib>
#include <cstdio>
#if _OPENMP
#include "test/OpenMP_Mutex.h"
#endif /* _OPENMP */
#include "tbb/tbb_profiling.h"

#ifndef TBB_TEST_LOW_WORKLOAD
    #define TBB_TEST_LOW_WORKLOAD TBB_USE_THREADING_TOOLS
#endif

// This test deliberately avoids a "using tbb" statement,
// so that the error of putting types in the wrong namespace will be caught.

template<typename M>
struct Counter {
    typedef M mutex_type;
    M mutex;
    volatile long value;
};

//! Function object for use with parallel_for.h.
template<typename C>
struct AddOne: NoAssign {
    C& counter;
    /** Increments counter once for each iteration in the iteration space. */
    void operator()( tbb::blocked_range<size_t>& range ) const {
        for( size_t i=range.begin(); i!=range.end(); ++i ) {
            if( i&1 ) {
                // Try implicit acquire and explicit release
                typename C::mutex_type::scoped_lock lock(counter.mutex);
                counter.value = counter.value+1;
                lock.release();
            } else {
                // Try explicit acquire and implicit release
                typename C::mutex_type::scoped_lock lock;
                lock.acquire(counter.mutex);
                counter.value = counter.value+1;
            }
        }
    }
    AddOne( C& counter_ ) : counter(counter_) {}
};

//! Adaptor for using ISO C++0x style mutex as a TBB-style mutex.
template<typename M>
class TBB_MutexFromISO_Mutex {
    M my_iso_mutex;
public:
    typedef TBB_MutexFromISO_Mutex mutex_type;

    class scoped_lock;
    friend class scoped_lock;

    class scoped_lock {
        mutex_type* my_mutex;
    public:
        scoped_lock() : my_mutex(NULL) {}
        scoped_lock( mutex_type& m ) : my_mutex(NULL) {
            acquire(m);
        }
        scoped_lock( mutex_type& m, bool is_writer ) : my_mutex(NULL) {
            acquire(m,is_writer);
        }
        void acquire( mutex_type& m ) {
            m.my_iso_mutex.lock();
            my_mutex = &m;
        }
        bool try_acquire( mutex_type& m ) {
            if( m.my_iso_mutex.try_lock() ) {
                my_mutex = &m;
                return true;
            } else {
                return false;
            }
        }
        void release() {
            my_mutex->my_iso_mutex.unlock();
            my_mutex = NULL;
        }

        // Methods for reader-writer mutex
        // These methods can be instantiated only if M supports lock_read() and try_lock_read().
        
        void acquire( mutex_type& m, bool is_writer ) {
            if( is_writer ) m.my_iso_mutex.lock();
            else m.my_iso_mutex.lock_read();
            my_mutex = &m;
        } 
        bool try_acquire( mutex_type& m, bool is_writer ) {
            if( is_writer ? m.my_iso_mutex.try_lock() : m.my_iso_mutex.try_lock_read() ) {
                my_mutex = &m;
                return true;
            } else {
                return false;
            }
        }
        bool upgrade_to_writer() {
            my_mutex->my_iso_mutex.unlock();
            my_mutex->my_iso_mutex.lock(); 
            return false;
        }
        bool downgrade_to_reader() {
            my_mutex->my_iso_mutex.unlock();
            my_mutex->my_iso_mutex.lock_read(); 
            return false;
        }
        ~scoped_lock() {
            if( my_mutex ) 
                release();
        }
    };    
  
    static const bool is_recursive_mutex = M::is_recursive_mutex;
    static const bool is_rw_mutex = M::is_rw_mutex;
};

namespace tbb {
    namespace profiling {
        template<typename M>
        void set_name( const TBB_MutexFromISO_Mutex<M>&, const char* ) {}  
    }
}

//! Generic test of a TBB mutex type M.
/** Does not test features specific to reader-writer locks. */
template<typename M>
void Test( const char * name ) {
    REMARK("%s time = ",name);
    Counter<M> counter;
    counter.value = 0;
    tbb::profiling::set_name(counter.mutex, name);
#if TBB_TEST_LOW_WORKLOAD
    const int n = 10000;
#else
    const int n = 100000;
#endif /* TBB_TEST_LOW_WORKLOAD */
    tbb::tick_count t0 = tbb::tick_count::now();
    tbb::parallel_for(tbb::blocked_range<size_t>(0,n,n/10),AddOne<Counter<M> >(counter));
    tbb::tick_count t1 = tbb::tick_count::now();
    REMARK("%g usec\n",(t1-t0).seconds());
    if( counter.value!=n )
        REPORT("ERROR for %s: counter.value=%ld\n",name,counter.value);
}

template<typename M, size_t N>
struct Invariant {
    typedef M mutex_type;
    M mutex;
    const char* mutex_name;
    volatile long value[N];
    volatile long single_value;
    Invariant( const char* mutex_name_ ) :
        mutex_name(mutex_name_)
    {
        single_value = 0;
        for( size_t k=0; k<N; ++k )
            value[k] = 0;
        tbb::profiling::set_name(mutex, mutex_name_);
    }
    void update() {
        for( size_t k=0; k<N; ++k )
            ++value[k];
    }
    bool value_is( long expected_value ) const {
        long tmp;
        for( size_t k=0; k<N; ++k )
            if( (tmp=value[k])!=expected_value ) {
                REPORT("ERROR: %ld!=%ld\n", tmp, expected_value);
                return false;
            }
        return true;
    }
    bool is_okay() {
        return value_is( value[0] );
    }
};

//! Function object for use with parallel_for.h.
template<typename I>
struct TwiddleInvariant: NoAssign {
    I& invariant;
    /** Increments counter once for each iteration in the iteration space. */
    void operator()( tbb::blocked_range<size_t>& range ) const {
        for( size_t i=range.begin(); i!=range.end(); ++i ) {
            //! Every 8th access is a write access
            bool write = (i%8)==7;
            bool okay = true;
            bool lock_kept = true;
            if( (i/8)&1 ) {
                // Try implicit acquire and explicit release
                typename I::mutex_type::scoped_lock lock(invariant.mutex,write);
                if( write ) {
                    long my_value = invariant.value[0];
                    invariant.update();
                    if( i%16==7 ) {
                        lock_kept = lock.downgrade_to_reader();
                        if( !lock_kept )
                            my_value = invariant.value[0] - 1;
                        okay = invariant.value_is(my_value+1);
                    }
                } else {
                    okay = invariant.is_okay();
                    if( i%8==3 ) {
                        long my_value = invariant.value[0];
                        lock_kept = lock.upgrade_to_writer();
                        if( !lock_kept )
                            my_value = invariant.value[0];
                        invariant.update();
                        okay = invariant.value_is(my_value+1);
                    }
                }
                lock.release();
            } else {
                // Try explicit acquire and implicit release
                typename I::mutex_type::scoped_lock lock;
                lock.acquire(invariant.mutex,write);
                if( write ) {
                    long my_value = invariant.value[0];
                    invariant.update();
                    if( i%16==7 ) {
                        lock_kept = lock.downgrade_to_reader();
                        if( !lock_kept )
                            my_value = invariant.value[0] - 1;
                        okay = invariant.value_is(my_value+1);
                    }
                } else {
                    okay = invariant.is_okay();
                    if( i%8==3 ) {
                        long my_value = invariant.value[0];
                        lock_kept = lock.upgrade_to_writer();
                        if( !lock_kept )
                            my_value = invariant.value[0];
                        invariant.update();
                        okay = invariant.value_is(my_value+1);
                    }
                }
            }
            if( !okay ) {
                REPORT( "ERROR for %s at %ld: %s %s %s %s\n",invariant.mutex_name, long(i),
                             write?"write,":"read,", write?(i%16==7?"downgrade,":""):(i%8==3?"upgrade,":""),
                             lock_kept?"lock kept,":"lock not kept,", (i/8)&1?"imp/exp":"exp/imp" );
            }
        }
    }
    TwiddleInvariant( I& invariant_ ) : invariant(invariant_) {}
};

/** This test is generic so that we can test any other kinds of ReaderWriter locks we write later. */
template<typename M>
void TestReaderWriterLock( const char * mutex_name ) {
    REMARK( "%s readers & writers time = ", mutex_name );
    Invariant<M,8> invariant(mutex_name);
#if TBB_TEST_LOW_WORKLOAD
    const size_t n = 10000;
#else
    const size_t n = 500000;
#endif /* TBB_TEST_LOW_WORKLOAD */
    tbb::tick_count t0 = tbb::tick_count::now();
    tbb::parallel_for(tbb::blocked_range<size_t>(0,n,n/100),TwiddleInvariant<Invariant<M,8> >(invariant));
    tbb::tick_count t1 = tbb::tick_count::now();
    // There is either a writer or a reader upgraded to a writer for each 4th iteration
    long expected_value = n/4;
    if( !invariant.value_is(expected_value) )
        REPORT("ERROR for %s: final invariant value is wrong\n",mutex_name);
    REMARK( "%g usec\n", (t1-t0).seconds() );
}

#if _MSC_VER && !defined(__INTEL_COMPILER)
    // Suppress "conditional expression is constant" warning.
    #pragma warning( push )
    #pragma warning( disable: 4127 )
#endif

/** Test try_acquire_reader functionality of a non-reenterable reader-writer mutex */
template<typename M>
void TestTryAcquireReader_OneThread( const char * mutex_name ) {
    M tested_mutex;
    typename M::scoped_lock lock1;
    if( M::is_rw_mutex ) {
        if( lock1.try_acquire(tested_mutex, false) )
            lock1.release();
        else
            REPORT("ERROR for %s: try_acquire failed though it should not\n", mutex_name);
        {
            typename M::scoped_lock lock2(tested_mutex, false);
            if( lock1.try_acquire(tested_mutex) )
                REPORT("ERROR for %s: try_acquire succeeded though it should not\n", mutex_name);
            lock2.release();
            lock2.acquire(tested_mutex, true);
            if( lock1.try_acquire(tested_mutex, false) )
                REPORT("ERROR for %s: try_acquire succeeded though it should not\n", mutex_name);
        }
        if( lock1.try_acquire(tested_mutex, false) )
            lock1.release();
        else
            REPORT("ERROR for %s: try_acquire failed though it should not\n", mutex_name);
    }
}

/** Test try_acquire functionality of a non-reenterable mutex */
template<typename M>
void TestTryAcquire_OneThread( const char * mutex_name ) {
    M tested_mutex;
    typename M::scoped_lock lock1;
    if( lock1.try_acquire(tested_mutex) )
        lock1.release();
    else
        REPORT("ERROR for %s: try_acquire failed though it should not\n", mutex_name);
    {
        if( M::is_recursive_mutex ) {
            typename M::scoped_lock lock2(tested_mutex);
            if( lock1.try_acquire(tested_mutex) )
                lock1.release();
            else
                REPORT("ERROR for %s: try_acquire on recursive lock failed though it should not\n", mutex_name);
            //windows.. -- both are recursive
        } else {
            typename M::scoped_lock lock2(tested_mutex);
            if( lock1.try_acquire(tested_mutex) )
                REPORT("ERROR for %s: try_acquire succeeded though it should not\n", mutex_name);
        }
    }
    if( lock1.try_acquire(tested_mutex) )
        lock1.release();
    else
        REPORT("ERROR for %s: try_acquire failed though it should not\n", mutex_name);
} 

#if _MSC_VER && !defined(__INTEL_COMPILER)
    #pragma warning( pop )
#endif

const int RecurN = 4;
int RecurArray[ RecurN ];
tbb::recursive_mutex RecurMutex[ RecurN ];

struct RecursiveAcquisition {
    /** x = number being decoded in base N
        max_lock = index of highest lock acquired so far
        mask = bit mask; ith bit set if lock i has been acquired. */
    void Body( size_t x, int max_lock=-1, unsigned int mask=0 ) const
    {
        int i = (int) (x % RecurN);
        bool first = (mask&1U<<i)==0;
        if( first ) {
            // first time to acquire lock
            if( i<max_lock ) 
                // out of order acquisition might lead to deadlock, so stop
                return;
            max_lock = i;
        }

        if( (i&1)!=0 ) {
            // acquire lock on location RecurArray[i] using explict acquire
            tbb::recursive_mutex::scoped_lock r_lock;
            r_lock.acquire( RecurMutex[i] );
            int a = RecurArray[i];
            ASSERT( (a==0)==first, "should be either a==0 if it is the first time to acquire the lock or a!=0 otherwise" );
            ++RecurArray[i];
            if( x ) 
                Body( x/RecurN, max_lock, mask|1U<<i );
            --RecurArray[i];
            ASSERT( a==RecurArray[i], "a is not equal to RecurArray[i]" );                        

            // release lock on location RecurArray[i] using explicit release; otherwise, use implicit one
            if( (i&2)!=0 ) r_lock.release();
        } else {
            // acquire lock on location RecurArray[i] using implicit acquire
            tbb::recursive_mutex::scoped_lock r_lock( RecurMutex[i] );
            int a = RecurArray[i];

            ASSERT( (a==0)==first, "should be either a==0 if it is the first time to acquire the lock or a!=0 otherwise" );

            ++RecurArray[i];
            if( x ) 
                Body( x/RecurN, max_lock, mask|1U<<i );
            --RecurArray[i];

            ASSERT( a==RecurArray[i], "a is not equal to RecurArray[i]" );                        

            // release lock on location RecurArray[i] using explicit release; otherwise, use implicit one
            if( (i&2)!=0 ) r_lock.release();
        }
    }

    void operator()( const tbb::blocked_range<size_t> &r ) const
    {   
        for( size_t x=r.begin(); x<r.end(); x++ ) {
            Body( x );
        }
    }
};

/** This test is generic so that we may test other kinds of recursive mutexes.*/
template<typename M>
void TestRecursiveMutex( const char * mutex_name )
{
    for ( int i = 0; i < RecurN; ++i ) {
        tbb::profiling::set_name(RecurMutex[i], mutex_name);
    }
    tbb::tick_count t0 = tbb::tick_count::now();
    tbb::parallel_for(tbb::blocked_range<size_t>(0,10000,500), RecursiveAcquisition());
    tbb::tick_count t1 = tbb::tick_count::now();
    REMARK( "%s recursive mutex time = %g usec\n", mutex_name, (t1-t0).seconds() );
}

template<typename C>
struct NullRecursive: NoAssign {
    void recurse_till( size_t i, size_t till ) const {
        if( i==till ) {
            counter.value = counter.value+1;
            return;
        }
        if( i&1 ) {
            typename C::mutex_type::scoped_lock lock2(counter.mutex);
            recurse_till( i+1, till );
            lock2.release();
        } else {
            typename C::mutex_type::scoped_lock lock2;
            lock2.acquire(counter.mutex);
            recurse_till( i+1, till );
        }
    }

    void operator()( tbb::blocked_range<size_t>& range ) const {
        typename C::mutex_type::scoped_lock lock(counter.mutex);
        recurse_till( range.begin(), range.end() );
    }
    NullRecursive( C& counter_ ) : counter(counter_) {
        ASSERT( C::mutex_type::is_recursive_mutex, "Null mutex should be a recursive mutex." );
    }
    C& counter;
};

template<typename M>
struct NullUpgradeDowngrade: NoAssign {
    void operator()( tbb::blocked_range<size_t>& range ) const {
        typename M::scoped_lock lock2;
        for( size_t i=range.begin(); i!=range.end(); ++i ) {
            if( i&1 ) {
                typename M::scoped_lock lock1(my_mutex, true) ;
                if( lock1.downgrade_to_reader()==false )
                    REPORT("ERROR for %s: downgrade should always succeed\n", name);
            } else {
                lock2.acquire( my_mutex, false );
                if( lock2.upgrade_to_writer()==false )
                    REPORT("ERROR for %s: upgrade should always succeed\n", name);
                lock2.release();
            }
        }
    }

    NullUpgradeDowngrade( M& m_, const char* n_ ) : my_mutex(m_), name(n_) {}
    M& my_mutex;
    const char* name;
} ;

template<typename M>
void TestNullMutex( const char * name ) {
    Counter<M> counter;
    counter.value = 0;
    const int n = 100;
    REMARK("%s ",name);
    {
        tbb::parallel_for(tbb::blocked_range<size_t>(0,n,10),AddOne<Counter<M> >(counter));
    }
    counter.value = 0;
    {
        tbb::parallel_for(tbb::blocked_range<size_t>(0,n,10),NullRecursive<Counter<M> >(counter));
    }

}

template<typename M>
void TestNullRWMutex( const char * name ) {
    REMARK("%s ",name);
    const int n = 100;
    M m;
    tbb::parallel_for(tbb::blocked_range<size_t>(0,n,10),NullUpgradeDowngrade<M>(m, name));
}

//! Test ISO C++0x compatibility portion of TBB mutex 
template<typename M>
void TestISO( const char * name ) {
    typedef TBB_MutexFromISO_Mutex<M> tbb_from_iso;
    Test<tbb_from_iso>( name );
}

//! Test ISO C++0x try_lock functionality of a non-reenterable mutex */
template<typename M>
void TestTryAcquire_OneThreadISO( const char * name ) {
    typedef TBB_MutexFromISO_Mutex<M> tbb_from_iso;
    TestTryAcquire_OneThread<tbb_from_iso>( name );
}

//! Test ISO-like C++0x compatibility portion of TBB reader-writer mutex 
template<typename M>
void TestReaderWriterLockISO( const char * name ) {
    typedef TBB_MutexFromISO_Mutex<M> tbb_from_iso;
    TestReaderWriterLock<tbb_from_iso>( name );
    TestTryAcquireReader_OneThread<tbb_from_iso>( name );
}

//! Test ISO C++0x compatibility portion of TBB recursive mutex 
template<typename M>
void TestRecursiveMutexISO( const char * name ) {
    typedef TBB_MutexFromISO_Mutex<M> tbb_from_iso;
    TestRecursiveMutex<tbb_from_iso>(name); 
}

#include "tbb/task_scheduler_init.h"

int TestMain () {
    for( int p=MinThread; p<=MaxThread; ++p ) {
        tbb::task_scheduler_init init( p );
        REMARK( "testing with %d workers\n", static_cast<int>(p) );
#if TBB_TEST_LOW_WORKLOAD
        // The amount of work is decreased in this mode to bring the length 
        // of the runs under tools into the tolerable limits.
        const int n = 1;
#else
        const int n = 3;
#endif
        // Run each test several times.
        for( int i=0; i<n; ++i ) {
            TestNullMutex<tbb::null_mutex>( "Null Mutex" );
            TestNullMutex<tbb::null_rw_mutex>( "Null RW Mutex" );
            TestNullRWMutex<tbb::null_rw_mutex>( "Null RW Mutex" );
            Test<tbb::spin_mutex>( "Spin Mutex" );
#if _OPENMP
            Test<OpenMP_Mutex>( "OpenMP_Mutex" );
#endif /* _OPENMP */
            Test<tbb::queuing_mutex>( "Queuing Mutex" );
            Test<tbb::mutex>( "Wrapper Mutex" );
            Test<tbb::recursive_mutex>( "Recursive Mutex" );
            Test<tbb::queuing_rw_mutex>( "Queuing RW Mutex" );
            Test<tbb::spin_rw_mutex>( "Spin RW Mutex" );

            TestTryAcquire_OneThread<tbb::spin_mutex>("Spin Mutex");
            TestTryAcquire_OneThread<tbb::queuing_mutex>("Queuing Mutex");
#if USE_PTHREAD 
            // under ifdef because on Windows tbb::mutex is reenterable and the test will fail
            TestTryAcquire_OneThread<tbb::mutex>("Wrapper Mutex");
#endif /* USE_PTHREAD */
            TestTryAcquire_OneThread<tbb::recursive_mutex>( "Recursive Mutex" );
            TestTryAcquire_OneThread<tbb::spin_rw_mutex>("Spin RW Mutex"); // only tests try_acquire for writers
            TestTryAcquire_OneThread<tbb::queuing_rw_mutex>("Queuing RW Mutex"); // only tests try_acquire for writers
            TestTryAcquireReader_OneThread<tbb::spin_rw_mutex>("Spin RW Mutex"); 
            TestTryAcquireReader_OneThread<tbb::queuing_rw_mutex>("Queuing RW Mutex"); 

            TestReaderWriterLock<tbb::queuing_rw_mutex>( "Queuing RW Mutex" );
            TestReaderWriterLock<tbb::spin_rw_mutex>( "Spin RW Mutex" );

            TestRecursiveMutex<tbb::recursive_mutex>( "Recursive Mutex" );

            // Test ISO C++0x interface  
            TestISO<tbb::spin_mutex>( "ISO Spin Mutex" );
            TestISO<tbb::mutex>( "ISO Mutex" );
            TestISO<tbb::spin_rw_mutex>( "ISO Spin RW Mutex" );
            TestISO<tbb::recursive_mutex>( "ISO Recursive Mutex" );
            TestISO<tbb::critical_section>( "ISO Critical Section" );
            TestTryAcquire_OneThreadISO<tbb::spin_mutex>( "ISO Spin Mutex" );
#if USE_PTHREAD 
            // under ifdef because on Windows tbb::mutex is reenterable and the test will fail
            TestTryAcquire_OneThreadISO<tbb::mutex>( "ISO Mutex" );
#endif /* USE_PTHREAD */
            TestTryAcquire_OneThreadISO<tbb::spin_rw_mutex>( "ISO Spin RW Mutex" );
            TestTryAcquire_OneThreadISO<tbb::recursive_mutex>( "ISO Recursive Mutex" );
            TestTryAcquire_OneThreadISO<tbb::critical_section>( "ISO Critical Section" );
            TestReaderWriterLockISO<tbb::spin_rw_mutex>( "ISO Spin RW Mutex" );
            TestRecursiveMutexISO<tbb::recursive_mutex>( "ISO Recursive Mutex" );
        }
        REMARK( "calling destructor for task_scheduler_init\n" );
    }
    return Harness::Done;
}