cat >umod-amd64.s <<'EOF'
# tu_int __umodti3(tu_int x, tu_int y)
# x is rsi:rdi, y is rcx:rdx, return result is rdx:rax.
.globl __umodti3
__umodti3:
	# specialized to u128 % u64, so verify that
	test %rcx,%rcx
	jne 1f

	# save divisor
	movq %rdx, %r8

	# reduce top 64 bits mod divisor
	movq %rsi, %rax
	xorl %edx, %edx
	divq %r8

	# reduce full 128-bit mod divisor

	MOVD    sinhxinit<>+0(SB), R1
	FMOVD   F0, F4
	MOVD    R1, R3
	BLTU    L19
	FMOVD   F0, F2
L2:
	WORD    $0xED205010     //cdb %f2,.L22-.L21(%r5)
	BYTE    $0x00
	BYTE    $0x19
	BGE     L15     //jnl   .L15
	BVS     L15
	WFCEDBS V2, V2, V0
	BEQ     L20
L12:
	FMOVD   F4, F0
	FMOVD   F0, ret+8(FP)
	RET

L15:
	WFCEDBS V2, V2, V0
	BVS     L12
	MOVD    $sinhxlim1<>+0(SB), R2
	FMOVD   0(R2), F0

	// n > 0
	MOVL z+0(FP), DI
	MOVL x+12(FP), SI
	MOVL s+24(FP), CX
	MOVL (SI)(BX*4), AX	// w1 = x[n-1]
	MOVL $0, DX
	SHLL CX, AX, DX		// w1>>ŝ
	MOVL DX, c+28(FP)

	CMPL BX, $0
	JLE X8a			// i <= 0

	// i > 0
L8:	MOVL AX, DX		// w = w1
	MOVL -4(SI)(BX*4), AX	// w1 = x[i-1]
	SHLL CX, AX, DX		// w<<s | w1>>ŝ
	MOVL DX, (DI)(BX*4)	// z[i] = w<<s | w1>>ŝ
	SUBL $1, BX		// i--