Min and Max functions in the Linux kernel __oracle

Look at the code today to see an interesting dongdong, that is, the Linux kernel also has the Min function, but its implementation is very strange, first posted out:
/*
* min ()/max () macros that also do
* Strict type-checking. The
* "Unnecessary" pointer comparison.
*/
#define MIN (x,y) ({typeof (x) _x = (x); typeof (Y) _y = (y); (void) (&_x = = &_y); _x < _y? _x: _y; })
#define MAX (X,y) ({typeof (x) _x = (x); typeof (Y) _y = (y); (void) (&_x = = &_y); _x > _y? _x: _y; })

Others are common, but the middle (void) (&_x = = &_y) is strange, what is the use of this sentence?
Check the next network found:
(void) (&_x = = &_y) This sentence itself executes the procedure to say completely is Div:eq (0) > Div:eq (3) > Table:eq (2) > Tbody:eq (0) > Tr:eq (0) > td:eq (0) > Div:eq (0) > #content: eq (0) > Br:eq (+) "anchortype=" Previous "jquery1251853974125=" 6 "> sentence
Nonsense, it is in itself that we can not do such operations typeof (_x) ==typeof (_y), so deliberately judged their 2 address pointers are equal, it is obviously impossible to equal, but if the _x and _y type is not the same, its pointer type will be different, 2 different pointer types for comparison operations, a compilation warning is thrown. That is to say Char *p; int *q; And then p==q, and this judgment because one is char* and one is int*, it generates a warning at compile time. Ingenious on the ingenious here.

In addition, why to define two local variable _x,_y, the function is as follows:
If this is defined as follows:
#define MIN (A,b) (((a) < (b))? (a): (b))
Imagine:
Min (++a,++b) ==> (++a) < (++B)? (++a): (++b)
Is there a problem, the incoming parameters were added two times.
Defined as the above so there will be no problem, hehe