Re: hook question

[Ralph Hempel <rhempel@...>]

Glenn Maynard wrote:

> Just a different idea:
>
>  do
>    A = read hook
>  while A != read hook
>
> If "hook" is volatile, the compiler should actually do the necessary two
> reads here.  This doesn't require sig_atomic_t, which is expensive on
> platforms without native atomic opcodes (may require disabling interrupts,
> etc).

Yes, but if the thread with this code is reading in between
the non-atomic write of the other thread, are we not back where we
started? :-)

Ralph