Question

I’m reading the source code of Libevent2.0 and I find some code below which I can’t understand. Why we can access the member “ev_ncalls” without accessing “_ev” and “ev_signal”?
And I complied some codes like that which can’t pass through gcc. Here is the code:

struct event {
    union {
            /* used for io events */
            struct {
                    TAILQ_ENTRY(event) ev_io_next;
                    struct timeval ev_timeout;
            } ev_io;

            /* used by signal events */
            struct {
                    TAILQ_ENTRY(event) ev_signal_next;
                    short ev_ncalls;
                    /* Allows deletes in callback */
                    short *ev_pncalls;
            } ev_signal;
    } _ev;
};

struct event *ev;
int ncall = ev->ev_ncalls;

Answer 1

It’s because event-internal.h has stuff like:

#define ev_ncalls   _ev.ev_signal.ev_ncalls
#define ev_pncalls  _ev.ev_signal.ev_pncalls

So when you say ev->ev_ncalls the compiler sees ev->_ev.ev_signal.ev_ncalls.