Replacing the atomic_add Instruction

This function is used to perform atomic addition on integer variables.

• Code on x86:

  static inline void atomic_add(int i, atomic_t *v) 
  {
  asm volatile(LOCK_PREFIX "addl %1,%0" : "+m" (v->counter) : "ir" (i));// Add i to the atomic variable v->counter.
  }

• Alternatives for Kunpeng processors:

  Method 1: Use the built-in atomic operations provided by the GUN compiler collection (GCC).

  __sync_add_and_fetch(&_value.counter,1)

  Method 2: Use the inline assembly.

  void atomic_add (int i) 
  { 
  unsigned int tmp; 
  int result; 
  __asm__ volatile(" prfm pstl1strm, %2\n" 
  "1: ldaxr %w0, %2\n" // Load data to the register.
  " add %w0, %w0, %w3\n" // Perform an add operation.
  " stlxr %w1, %w0, %2\n" // Write data after the add operation to the memory and check whether the data is successfully written.
  "cbnz %w1, 1b" // If the preceding operation fails, perform the add operation again.
  : "=&r"(result), "=&r"(tmp), "+Q"(_value.counter) 
  : "Ir"(i)) 
  }