// Allocate 96 bytes on the stack.
	MOVD	$-96(R15), R15

	// Save F8-F15 in our stack frame.
	FMOVD	F8, 32(R15)
	FMOVD	F9, 40(R15)
	FMOVD	F10, 48(R15)
	FMOVD	F11, 56(R15)
	FMOVD	F12, 64(R15)
	FMOVD	F13, 72(R15)
	FMOVD	F14, 80(R15)
	FMOVD	F15, 88(R15)

	// Initialize Go ABI environment.
	BL	runtime·load_g(SB)

	MOVD	R2, 8(R15)	// fn unsafe.Pointer
	MOVD	R3, 16(R15)	// a unsafe.Pointer
	// Skip R4 = n uint32

	FMOVD	(R1), F0               // 68001000
	FMOVD	4095(R2), F13          // 68d02fff
	FMOVD	4096(R2), F15          // edf020000165
	FMOVS	4095(R2)(R3), F13      // 78d32fff
	FMOVS	4096(R2)(R4), F15      // edf420000164
	FMOVD	F0, 4095(R1)           // 60001fff
	FMOVD	F0, 4096(R1)           // ed0010000167
	FMOVS	F13, 4095(R2)(R3)      // 70d32fff
	FMOVS	F13, 4096(R2)(R3)      // edd320000166

			inputs: []inputInfo{
				{0, 4294901760}, // F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15
				{1, 4294901760}, // F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15
				{2, 4294901760}, // F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15
			},
			outputs: []outputInfo{
				{0, 4294901760}, // F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15
			},
		},
	},
	{
		name:         "MULAD",

	std     %f12, 32(%r15)
	std     %f13, 40(%r15)
	std     %f14, 48(%r15)
	std     %f15, 56(%r15)

	/* restore g pointer */
	lgr     %r13, %r3

	/* call fn */
	basr    %r14, %r2

	/* restore floating point registers */
	ld      %f8, 0(%r15)
	ld      %f9, 8(%r15)
	ld      %f10, 16(%r15)
	ld      %f11, 24(%r15)
	ld      %f12, 32(%r15)
	ld      %f13, 40(%r15)
	ld      %f14, 48(%r15)

	MOVD F3, 240(R3)
	MOVD F4, 248(R3)
	MOVD F5, 256(R3)
	MOVD F6, 264(R3)
	MOVD F7, 272(R3)
	MOVD F8, 280(R3)
	MOVD F9, 288(R3)
	MOVD F10, 296(R3)
	MOVD F11, 304(R3)
	MOVD F12, 312(R3)
	MOVD F13, 320(R3)
	MOVD F14, 328(R3)
	MOVD F15, 336(R3)
	MOVD F16, 344(R3)
	MOVD F17, 352(R3)
	MOVD F18, 360(R3)
	MOVD F19, 368(R3)
	MOVD F20, 376(R3)
	MOVD F21, 384(R3)
	MOVD F22, 392(R3)
	MOVD F23, 400(R3)

	MOVD F8, 124(R13)
	MOVD F9, 132(R13)
	MOVD F10, 140(R13)
	MOVD F11, 148(R13)
	MOVD F12, 156(R13)
	MOVD F13, 164(R13)
	MOVD F14, 172(R13)
	MOVD F15, 180(R13)
nofp:
	CALL ·asyncPreempt2(SB)
	MOVB ·goarmsoftfp(SB), R0
	CMP $0, R0
	BNE nofp2
	MOVD 180(R13), F15
	MOVD 172(R13), F14
	MOVD 164(R13), F13
	MOVD 156(R13), F12
	MOVD 148(R13), F11
	MOVD 140(R13), F10
	MOVD 132(R13), F9
	MOVD 124(R13), F8

	MOVB    runtime·goarmsoftfp(SB), R11
	CMP     $0, R11
	BNE     skipfpsave
	MOVD	F8, (13*4+8*1)(R13)
	MOVD	F9, (13*4+8*2)(R13)
	MOVD	F10, (13*4+8*3)(R13)
	MOVD	F11, (13*4+8*4)(R13)
	MOVD	F12, (13*4+8*5)(R13)
	MOVD	F13, (13*4+8*6)(R13)
	MOVD	F14, (13*4+8*7)(R13)
	MOVD	F15, (13*4+8*8)(R13)

skipfpsave:
	BL	runtime·load_g(SB)
	// We set up the arguments to cgocallback when saving registers above.
	BL	runtime·cgocallback(SB)

	MOVD F3, 232(X2)
	MOVD F4, 240(X2)
	MOVD F5, 248(X2)
	MOVD F6, 256(X2)
	MOVD F7, 264(X2)
	MOVD F8, 272(X2)
	MOVD F9, 280(X2)
	MOVD F10, 288(X2)
	MOVD F11, 296(X2)
	MOVD F12, 304(X2)
	MOVD F13, 312(X2)
	MOVD F14, 320(X2)
	MOVD F15, 328(X2)
	MOVD F16, 336(X2)
	MOVD F17, 344(X2)
	MOVD F18, 352(X2)
	MOVD F19, 360(X2)
	MOVD F20, 368(X2)
	MOVD F21, 376(X2)
	MOVD F22, 384(X2)
	MOVD F23, 392(X2)

	for i := 0; i < 12; i++ {
		if i > 7 {
			dst.gp_regs[i+6] = uint64(src.Ints[i])
		} else {
			dst.gp_regs[i+3] = uint64(src.Ints[i])
		}
	}
	// Fprs F1..F13 are used to pass float arguments in registers on PPC64
	for i := 0; i < 12; i++ {
		dst.fp_regs[i+1] = math.Float64frombits(src.Floats[i])
	}

}

func loadRegArgs(dst *abi.RegArgs, src *sigcontext) {

	FADDD F24, F14, F12                        // cc29781e
	//TODO VFADDP V4.D2, F13                   // 8dd8707e
	//TODO VFADDP V30.S4, V3.S4, V11.S4        // 6bd43e6e
	FCCMPS LE, F17, F12, $14                   // 8ed5311e
	FCCMPD HI, F11, F15, $15                   // ef856b1e
	FCCMPES HS, F28, F13, $13                  // bd253c1e
	FCCMPED LT, F20, F4, $9                    // 99b4741e