News for hackers. Stuff that matters.
Re:Ntimed – NTPD replacement by asveikau on Wednesday December 24, 2014 @11:49AM (#8793449)
Other than the address compare and swap takes two operands: expected value and new value. It does an equality check and (if equal) the swap atomically. Therefore the equality check is handy to atomically check if a previous potentially unsafe read is still valid, and if so, set a new value.
Re:Ntimed – NTPD replacement by asveikau on Wednesday December 24, 2014 @02:46PM (#8794251)
Too late to edit the post but maybe some code sample will help illustrate the pattern. I am typing code into a web form so I make no guarantees that it is perfect.

  // Performs some modification on the input.
  extern int f(int);

  volatile int global_var = /* ... */;

  int expected;
  int new; // assuming C and not C++, where "new" is reserved
  do
  {
     // Do a speculative read
     expected = global_var;

     // Make some (non-atomic) modification
     new = f(expected);

     // loop until the speculative read was OK. (won the race)
  } while (compare_and_swap(&global_var, expected, new) != expected);
Re:Ntimed – NTPD replacement by dbaupp on Thursday December 25, 2014 @05:22PM (#8797361)
Whoops, I got the signature of CAS slightly wrong. In any case, I already understood the pattern and understand why it is OK. However, as I said, it is the exception that a data race is OK, which is why Rust disallows them by default (it does not disallow them completely).