Update dependencies and go1.18 (#1873)

* Update dependencies and go1.18

* Exclude unnecessary linters and update build to go1.18

vendor/filippo.io/edwards25519/README.md

@@ -7,7 +7,7 @@ import "filippo.io/edwards25519"
 This library implements the edwards25519 elliptic curve, exposing the necessary APIs to build a wide array of higher-level primitives.
 Read the docs at [pkg.go.dev/filippo.io/edwards25519](https://pkg.go.dev/filippo.io/edwards25519).
 
-The code is originally derived from Adam Langley's internal implementation in the Go standard library, and includes George Tankersley's [performance improvements](https://golang.org/cl/71950). It was then further developed by Henry de Valence for use in ristretto255.
+The code is originally derived from Adam Langley's internal implementation in the Go standard library, and includes George Tankersley's [performance improvements](https://golang.org/cl/71950). It was then further developed by Henry de Valence for use in ristretto255, and was finally [merged back into the Go standard library](https://golang.org/cl/276272) as of Go 1.17. It now tracks the upstream codebase and extends it with additional functionality.
 
 Most users don't need this package, and should instead use `crypto/ed25519` for signatures, `golang.org/x/crypto/curve25519` for Diffie-Hellman, or `github.com/gtank/ristretto255` for prime order group logic. However, for anyone currently using a fork of `crypto/ed25519/internal/edwards25519` or `github.com/agl/edwards25519`, this package should be a safer, faster, and more powerful alternative.
 

vendor/filippo.io/edwards25519/field/fe.go

@@ -188,12 +188,13 @@ func (v *Element) Set(a *Element) *Element {
 }
 
 // SetBytes sets v to x, where x is a 32-byte little-endian encoding. If x is
-// not of the right length, SetUniformBytes returns nil and an error, and the
+// not of the right length, SetBytes returns nil and an error, and the
 // receiver is unchanged.
 //
 // Consistent with RFC 7748, the most significant bit (the high bit of the
 // last byte) is ignored, and non-canonical values (2^255-19 through 2^255-1)
-// are accepted. Note that this is laxer than specified by RFC 8032.
+// are accepted. Note that this is laxer than specified by RFC 8032, but
+// consistent with most Ed25519 implementations.
 func (v *Element) SetBytes(x []byte) (*Element, error) {
 	if len(x) != 32 {
 		return nil, errors.New("edwards25519: invalid field element input size")
@@ -211,7 +212,7 @@ func (v *Element) SetBytes(x []byte) (*Element, error) {
 	// Bits 153:204 (bytes 19:27, bits 152:216, shift 1, mask 51).
 	v.l3 = binary.LittleEndian.Uint64(x[19:27]) >> 1
 	v.l3 &= maskLow51Bits
-	// Bits 204:251 (bytes 24:32, bits 192:256, shift 12, mask 51).
+	// Bits 204:255 (bytes 24:32, bits 192:256, shift 12, mask 51).
 	// Note: not bytes 25:33, shift 4, to avoid overread.
 	v.l4 = binary.LittleEndian.Uint64(x[24:32]) >> 12
 	v.l4 &= maskLow51Bits
@@ -394,26 +395,26 @@ var sqrtM1 = &Element{1718705420411056, 234908883556509,
 // If u/v is square, SqrtRatio returns r and 1. If u/v is not square, SqrtRatio
 // sets r according to Section 4.3 of draft-irtf-cfrg-ristretto255-decaf448-00,
 // and returns r and 0.
-func (r *Element) SqrtRatio(u, v *Element) (rr *Element, wasSquare int) {
-	var a, b Element
+func (r *Element) SqrtRatio(u, v *Element) (R *Element, wasSquare int) {
+	t0 := new(Element)
 
 	// r = (u * v3) * (u * v7)^((p-5)/8)
-	v2 := a.Square(v)
-	uv3 := b.Multiply(u, b.Multiply(v2, v))
-	uv7 := a.Multiply(uv3, a.Square(v2))
-	r.Multiply(uv3, r.Pow22523(uv7))
+	v2 := new(Element).Square(v)
+	uv3 := new(Element).Multiply(u, t0.Multiply(v2, v))
+	uv7 := new(Element).Multiply(uv3, t0.Square(v2))
+	rr := new(Element).Multiply(uv3, t0.Pow22523(uv7))
 
-	check := a.Multiply(v, a.Square(r)) // check = v * r^2
+	check := new(Element).Multiply(v, t0.Square(rr)) // check = v * r^2
 
-	uNeg := b.Negate(u)
+	uNeg := new(Element).Negate(u)
 	correctSignSqrt := check.Equal(u)
 	flippedSignSqrt := check.Equal(uNeg)
-	flippedSignSqrtI := check.Equal(uNeg.Multiply(uNeg, sqrtM1))
+	flippedSignSqrtI := check.Equal(t0.Multiply(uNeg, sqrtM1))
 
-	rPrime := b.Multiply(r, sqrtM1) // r_prime = SQRT_M1 * r
+	rPrime := new(Element).Multiply(rr, sqrtM1) // r_prime = SQRT_M1 * r
 	// r = CT_SELECT(r_prime IF flipped_sign_sqrt | flipped_sign_sqrt_i ELSE r)
-	r.Select(rPrime, r, flippedSignSqrt|flippedSignSqrtI)
+	rr.Select(rPrime, rr, flippedSignSqrt|flippedSignSqrtI)
 
-	r.Absolute(r) // Choose the nonnegative square root.
+	r.Absolute(rr) // Choose the nonnegative square root.
 	return r, correctSignSqrt | flippedSignSqrt
 }

vendor/filippo.io/edwards25519/field/fe_extra.go

@@ -0,0 +1,50 @@
+// Copyright (c) 2021 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package field
+
+import "errors"
+
+// This file contains additional functionality that is not included in the
+// upstream crypto/ed25519/internal/edwards25519/field package.
+
+// SetWideBytes sets v to x, where x is a 64-byte little-endian encoding, which
+// is reduced modulo the field order. If x is not of the right length,
+// SetWideBytes returns nil and an error, and the receiver is unchanged.
+//
+// SetWideBytes is not necessary to select a uniformly distributed value, and is
+// only provided for compatibility: SetBytes can be used instead as the chance
+// of bias is less than 2⁻²⁵⁰.
+func (v *Element) SetWideBytes(x []byte) (*Element, error) {
+	if len(x) != 64 {
+		return nil, errors.New("edwards25519: invalid SetWideBytes input size")
+	}
+
+	// Split the 64 bytes into two elements, and extract the most significant
+	// bit of each, which is ignored by SetBytes.
+	lo, _ := new(Element).SetBytes(x[:32])
+	loMSB := uint64(x[31] >> 7)
+	hi, _ := new(Element).SetBytes(x[32:])
+	hiMSB := uint64(x[63] >> 7)
+
+	// The output we want is
+	//
+	//   v = lo + loMSB * 2²⁵⁵ + hi * 2²⁵⁶ + hiMSB * 2⁵¹¹
+	//
+	// which applying the reduction identity comes out to
+	//
+	//   v = lo + loMSB * 19 + hi * 2 * 19 + hiMSB * 2 * 19²
+	//
+	// l0 will be the sum of a 52 bits value (lo.l0), plus a 5 bits value
+	// (loMSB * 19), a 6 bits value (hi.l0 * 2 * 19), and a 10 bits value
+	// (hiMSB * 2 * 19²), so it fits in a uint64.
+
+	v.l0 = lo.l0 + loMSB*19 + hi.l0*2*19 + hiMSB*2*19*19
+	v.l1 = lo.l1 + hi.l1*2*19
+	v.l2 = lo.l2 + hi.l2*2*19
+	v.l3 = lo.l3 + hi.l3*2*19
+	v.l4 = lo.l4 + hi.l4*2*19
+
+	return v.carryPropagate(), nil
+}

vendor/filippo.io/edwards25519/field/fe_generic.go

@@ -254,6 +254,8 @@ func (v *Element) carryPropagateGeneric() *Element {
 	c3 := v.l3 >> 51
 	c4 := v.l4 >> 51
 
+	// c4 is at most 64 - 51 = 13 bits, so c4*19 is at most 18 bits, and
+	// the final l0 will be at most 52 bits. Similarly for the rest.
 	v.l0 = v.l0&maskLow51Bits + c4*19
 	v.l1 = v.l1&maskLow51Bits + c0
 	v.l2 = v.l2&maskLow51Bits + c1

vendor/filippo.io/edwards25519/scalar.go

@@ -22,7 +22,7 @@ import (
 // The zero value is a valid zero element.
 type Scalar struct {
 	// s is the Scalar value in little-endian. The value is always reduced
-	// between operations.
+	// modulo l between operations.
 	s [32]byte
 }
 
@@ -79,9 +79,12 @@ func (s *Scalar) Set(x *Scalar) *Scalar {
 	return s
 }
 
-// SetUniformBytes sets s to an uniformly distributed value given 64 uniformly
-// distributed random bytes. If x is not of the right length, SetUniformBytes
-// returns nil and an error, and the receiver is unchanged.
+// SetUniformBytes sets s = x mod l, where x is a 64-byte little-endian integer.
+// If x is not of the right length, SetUniformBytes returns nil and an error,
+// and the receiver is unchanged.
+//
+// SetUniformBytes can be used to set s to an uniformly distributed value given
+// 64 uniformly distributed random bytes.
 func (s *Scalar) SetUniformBytes(x []byte) (*Scalar, error) {
 	if len(x) != 64 {
 		return nil, errors.New("edwards25519: invalid SetUniformBytes input length")