What is the execution status of LongAdder in Java concurrent programming?

longAccumulate method

final void longAccumulate(long x, LongBinaryOperator fn,
                          boolean wasUncontended) {
    int h;
    if ((h = getProbe()) == 0) {
        ThreadLocalRandom.current(); // force initialization
        h = getProbe();
        wasUncontended = true;
    }
    boolean collide = false;                // True if last slot nonempty
    for (;;) {
        Cell[] as; Cell a; int n; long v;
        if ((as = cells) != null && (n = as.length) > 0) {
            if ((a = as[(n - 1) & h]) == null) {
                if (cellsBusy == 0) {       // Try to attach new Cell
                    Cell r = new Cell(x);   // Optimistically create
                    if (cellsBusy == 0 && casCellsBusy()) {
                        boolean created = false;
                        try {               // Recheck under lock
                            Cell[] rs; int m, j;
                            if ((rs = cells) != null &&
                                (m = rs.length) > 0 &&
                                rs[j = (m - 1) & h] == null) {
                                rs[j] = r;
                                created = true;
                            }
                        } finally {
                            cellsBusy = 0;
                        }
                        if (created)
                            break;
                        continue;           // Slot is now non-empty
                    }
                }
                collide = false;
            }
            else if (!wasUncontended)       // CAS already known to fail
                wasUncontended = true;      // Continue after rehash
            else if (a.cas(v = a.value, ((fn == null) ? v + x :
                                         fn.applyAsLong(v, x))))
                break;
            else if (n >= NCPU || cells != as)
                collide = false;            // At max size or stale
            else if (!collide)
                collide = true;
            else if (cellsBusy == 0 && casCellsBusy()) {
                try {
                    if (cells == as) {      // Expand table unless stale
                        Cell[] rs = new Cell[n << 1];
                        for (int i = 0; i < n; ++i)
                            rs[i] = as[i];
                        cells = rs;
                    }
                } finally {
                    cellsBusy = 0;
                }
                collide = false;
                continue;                   // Retry with expanded table
            }
            h = advanceProbe(h);
        }
        else if (cellsBusy == 0 && cells == as && casCellsBusy()) {
            boolean init = false;
            try {                           // Initialize table
                if (cells == as) {
                    Cell[] rs = new Cell[2];
                    rs[h & 1] = new Cell(x);
                    cells = rs;
                    init = true;
                }
            } finally {
                cellsBusy = 0;
            }
            if (init)
                break;
        }
        else if (casBase(v = base, ((fn == null) ? v + x :
                                    fn.applyAsLong(v, x))))
            break;                          // Fall back on using base
    }
}

The code is long, let’s analyze it in sections, first introduce the content of each part

• Part one:forThe code before the loop is mainly to get the hash value of the thread. If it is 0, it will be forced to initialize

• The second part: forThe first # in the loop ##if Statement, accumulate and expand the Cell array

• Part 3:

  forThe first in the loop else if statement, the function of this part is to create the Cell array and initialize the

• Part 4:

  for loop In the second else if statement, when the Cell array is in fierce competition, try to accumulate on base

Thread hash value

int h; 
if ((h = getProbe()) == 0) { 
    ThreadLocalRandom.current(); // force initialization 
    h = getProbe(); 
    wasUncontended = true;   // true表示没有竞争
} 
boolean collide = false; // True if last slot nonempty 可以理解为是否需要扩容

The core code of this part is the

getProbe method. The function of this method is to obtain the hash value of the thread, which is convenient for locating # through bit operations later. ##CellA certain position in the array, if it is 0, it will be forced initializationInitialize the Cell array

final void longAccumulate(long x, LongBinaryOperator fn,
                          boolean wasUncontended) {
    // 省略...
    for (;;) {
        Cell[] as; Cell a; int n; long v;
        if ((as = cells) != null && (n = as.length) > 0) {
            // 省略...
        }
        else if (cellsBusy == 0 && cells == as && casCellsBusy()) {  // 获取锁
            boolean init = false;  // 初始化标志
            try {                           // Initialize table
                if (cells == as) {
                    Cell[] rs = new Cell[2];  // 创建Cell数组
                    rs[h & 1] = new Cell(x);  // 索引1位置创建Cell元素，值为x=1
                    cells = rs;   // cells指向新数组
                    init = true;  // 初始化完成
                }
            } finally {
                cellsBusy = 0;  // 释放锁
            }
            if (init)
                break;  // 跳出循环
        }
        else if (casBase(v = base, ((fn == null) ? v + x :
                                    fn.applyAsLong(v, x))))
            break;                          // Fall back on using base
    }
}

In the first case## The #Cell

array is

null, so it will enter the first else if statement, and no other thread will operate, so cellsBusy==0, cells==as is also true, casCellsBusy()try to perform cas operation on cellsBusy and change it to 1 is also true. First creates a Cell

array with two elements, and then uses the bit operation of thread

h & 1 at the position of index 1 Set a Cell whose value is 1, and then reassign it to cells, mark the initialization successfully, and modify cellsBusy 0 means releasing the lock, finally jumping out of the loop, and the initialization operation is completed. Accumulate base

final void longAccumulate(long x, LongBinaryOperator fn,
                          boolean wasUncontended) {
    // 省略...
    for (;;) {
        Cell[] as; Cell a; int n; long v;
        if ((as = cells) != null && (n = as.length) > 0) {
            // 省略...
        }
        else if (cellsBusy == 0 && cells == as && casCellsBusy()) {
            // 省略...
        }
        else if (casBase(v = base, ((fn == null) ? v + x :
                                    fn.applyAsLong(v, x))))
            break;                          // Fall back on using base
    }
}

The second

else if

statement means that when the competition for all positions in the

Cell array is fierce, Just try to accumulate on base, which can be understood as the final guaranteeAfter the Cell array is initialized

final void longAccumulate(long x, LongBinaryOperator fn,
                          boolean wasUncontended) {
    // 省略...
    for (;;) {
        Cell[] as; Cell a; int n; long v;
        if ((as = cells) != null && (n = as.length) > 0) {  // as初始化之后满足条件
            if ((a = as[(n - 1) & h]) == null) {  // as中某个位置的值为null
                if (cellsBusy == 0) {       // Try to attach new Cell 是否加锁
                    Cell r = new Cell(x);   // Optimistically create 创建新Cell
                    if (cellsBusy == 0 && casCellsBusy()) { // 双重检查是否有锁，并尝试加锁
                        boolean created = false;  // 
                        try {               // Recheck under lock
                            Cell[] rs; int m, j;
                            if ((rs = cells) != null &&
                                (m = rs.length) > 0 &&
                                rs[j = (m - 1) & h] == null) {  // 重新检查该位置是否为null
                                rs[j] = r;  // 该位置添加Cell元素
                                created = true;  // 新Cell创建成功
                            }
                        } finally {
                            cellsBusy = 0;  // 释放锁
                        }
                        if (created)
                            break;  // 创建成功，跳出循环
                        continue;           // Slot is now non-empty
                    }
                }
                collide = false;  // 扩容标志
            }
            else if (!wasUncontended)       // 上面定位到的索引位置的值不为null
                wasUncontended = true;      // 重新计算hash，重新定位其他索引位置重试
            else if (a.cas(v = a.value, ((fn == null) ? v + x :
                                         fn.applyAsLong(v, x))))  // 尝试在该索引位置进行累加
                break;
            else if (n >= NCPU || cells != as)  // 如果数组长度大于等于CPU核心数，就不能在扩容
                collide = false;            // At max size or stale
            else if (!collide)  // 数组长度没有达到最大值，修改扩容标志可以扩容
                collide = true;
            else if (cellsBusy == 0 && casCellsBusy()) { // 尝试加锁
                try {
                    if (cells == as) {      // Expand table unless stale
                        Cell[] rs = new Cell[n << 1];  // 创建一个原来长度2倍的数组
                        for (int i = 0; i < n; ++i)
                            rs[i] = as[i];  // 把原来的元素拷贝到新数组中
                        cells = rs;  // cells指向新数组
                    }
                } finally {
                    cellsBusy = 0;  // 释放锁
                }
                collide = false;  // 已经扩容完成，修改标志不用再扩容
                continue;                   // Retry with expanded table
            }
            h = advanceProbe(h);  // 重新获取hash值
        }
        // 省略...
}

Analyze the overall logic according to the comments in the code

First of all, if the value found at a certain position in the array is

null
• , it means that operations can be performed at this position, create a new

  Cell and initialize the value Place 1 at this position. If it fails, recalculate the hash value and try again.

  The positioned position already has a value. , indicating that there is competition between threads. If
wasUncontended
• is

  false, change it to true and recalculate hashtry again

  The positioned position has a value and
wasUncontended
• is already

  true, so try to accumulate at that position

  When the accumulation fails, determine whether the array capacity has reached the maximum. If so, it cannot be expanded and can only
rehash
• and try again
  If the previous conditions are not met Satisfied, and if the expansion flag
collide
• is marked as

  false, change it to true, indicating that expansion can be performed, and then rehashtry again

  First try to lock. If successful, the expansion operation will be performed. Each expansion length is
2
• times the previous one, and then copy the original array content to the new array. The expansion operation is finished.

The above is the detailed content of What is the execution status of LongAdder in Java concurrent programming?. For more information, please follow other related articles on the PHP Chinese website!