🚀 go-pugleaf

I'm getting very confused trying to implement atan2(x,y) with an
asm() statement on a bsd386 system (using gcc-2.2.2).

The 387 instruction "fpatan" requires input operands x and y to be in
on the top of the floating point register stack in %st and %st(1)
respectivly.  The output operand will be placed in %st after the stack
has been poped (i.e. %st(2) will end up in %st(1) afterwards.

From what I can gleam from the documentation, this should look
_something_ like this:

	#define atan2(x,y) \
	({ \
		double _x = (x), _y = (y), _z; \
		asm const ( \
			"fpatan" \
			: "=f" (_z) \
			: "0" (_x), "f" (_y) \
			: "2" \
		); \
		_z; \
	})

But isn't quite correct.  Do I need to specify registers in the
declaration of _x and _y?  How do I tell it that the stack will be
poped?  Do I need to add something to the last field in the asm() to
tell it that the floating point status word will be clobbered?


very worried,
brad
	huntting@glarp.com

   < <  zippy is so confused  * *