// left shifting, in this case by k bits. The de Bruijn (64 bit) constant
	// is such that all six bit, consecutive substrings are distinct.
	// Therefore, if we have a left shifted version of this constant we can
	// find by how many bits it was shifted by looking at which six bit
	// substring ended up at the top of the word.
	// (Knuth, volume 4, section 7.3.1)
	return int(deBruijn64tab[(x&-x)*deBruijn64>>(64-6)])
}

//go:noinline
func lshNop1(x uint64) uint64 {
	// two outer shifts should be removed
	return (((x << 5) >> 2) << 2)
}

//go:noinline
func lshNop2(x uint64) uint64 {
	return (((x << 5) >> 2) << 3)
}

//go:noinline
func lshNop3(x uint64) uint64 {
	return (((x << 5) >> 2) << 6)
}

//go:noinline
func lshNotNop(x uint64) uint64 {
	// outer shift can't be removed
	return (((x << 5) >> 2) << 1)
}

	// Unused portions of AuxInt are filled by sign-extending the used portion,
	// even if the represented value is unsigned.
	// Users of AuxInt which interpret AuxInt as unsigned (e.g. shifts) must be careful.
	// Use Value.AuxUnsigned to get the zero-extended value of AuxInt.
	AuxInt int64
	Aux    Aux

	// Arguments of this value
	Args []*Value

	// Containing basic block
	Block *Block

(And(64|32|16|8) ...) => (AND ...)
(Or(64|32|16|8) ...) => (OR ...)
(Xor(64|32|16|8) ...) => (XOR ...)

// shifts
// hardware instruction uses only the low 6 bits of the shift
// we compare to 64 to ensure Go semantics for large shifts
(Lsh64x64 <t> x y) => (AND (NEGV <t> (SGTU (MOVVconst <typ.UInt64> [64]) y)) (SLLV <t> x y))

// trailing 0 bits. x.mant is normalized if the msb of x.mant == 1 (i.e.,
// the msb is shifted all the way "to the left"). Thus, if the mantissa has
// trailing 0 bits or x.prec is not a multiple of the Word size _W,
// x.mant[0] has trailing zero bits. The msb of the mantissa corresponds
// to the value 0.5; the exponent x.exp shifts the binary point as needed.
//
// A zero or non-finite Float x ignores x.mant and x.exp.
//

		// shifts
		{name: "SLLV", argLength: 2, reg: gp21, asm: "SLLV"},                      // arg0 << arg1, shift amount is mod 64
		{name: "SLLVconst", argLength: 1, reg: gp11, asm: "SLLV", aux: "Int64"},   // arg0 << auxInt
		{name: "SRLV", argLength: 2, reg: gp21, asm: "SRLV"},                      // arg0 >> arg1, unsigned, shift amount is mod 64

		// shifts
		{name: "SLLV", argLength: 2, reg: gp21, asm: "SLLV"},                    // arg0 << arg1, shift amount is mod 64
		{name: "SLLVconst", argLength: 1, reg: gp11, asm: "SLLV", aux: "Int64"}, // arg0 << auxInt
		{name: "SRLV", argLength: 2, reg: gp21, asm: "SRLV"},                    // arg0 >> arg1, unsigned, shift amount is mod 64

(Not    x) => (XOR (MOVDconst [1]) x)

// shifts
// hardware instruction uses only the low 6 bits of the shift
// we compare to 64 to ensure Go semantics for large shifts
// Rules about rotates with non-const shift are based on the following rules,
// if the following rules change, please also modify the rules based on them.

// check shiftIsBounded first, if shift value is proved to be valid then we
// can do the shift directly.

		{name: "SHLLconst", argLength: 1, reg: gp11, asm: "SHLL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 << auxint, shift amount 0-31
		// Note: x86 is weird, the 16 and 8 byte shifts still use all 5 bits of shift amount!

	if err != nil {
		log.Fatal(err)
	}
	err = os.WriteFile(*filename, data, 0644)
	if err != nil {
		log.Fatal(err)
	}
}

type Data struct {
	a, b, c, d string
	Shift1     []int
	Shift2     []int
	Shift3     []int
	Shift4     []int
	Table1     []uint32
	Table2     []uint32
	Table3     []uint32
	Table4     []uint32
}

var funcs = template.FuncMap{
	"dup":     dup,
	"relabel": relabel,