go.crypto/sha3: update to sync with draft FIPS-202
1. API:
This exposes a minimal API: the SHA-3 functions implement hash.Hash. The
SHAKE functions implement a new "ShakeHash" interface that implements
io.Reader, io.Writer, and Reset().
(The previous Barrier() function has been removed.)
(Alternative proposal: Don't implement io.Reader, but instead provide a
"Digest(d []byte) error" function that performs a hash.Hash style copy.
Somewhat more minimal, but very easy to use incorrectly.)
2. Tests
Added the complete set of ShortMsgKATs from
https://github.com/gvanas/KeccakCodePackage
3. Correctness
In sync with draft FIPS-202.
4. Documentation
A summary of the security properties of the SHA-3 and SHAKE functions is
provided in doc.go; some concrete recommendations as well.
Fixes 8563.
R=golang-codereviews, agl
CC=golang-codereviews
https://golang.org/cl/130950043
diff --git a/sha3/sha3.go b/sha3/sha3.go
index e1f9aa8..8d77568 100644
--- a/sha3/sha3.go
+++ b/sha3/sha3.go
@@ -1,213 +1,226 @@
-// Copyright 2013 The Go Authors. All rights reserved.
+// Copyright 2014 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 sha3 implements the SHA3 hash algorithm (formerly called Keccak) chosen by NIST in 2012.
-// This file provides a SHA3 implementation which implements the standard hash.Hash interface.
-// Writing input data, including padding, and reading output data are computed in this file.
-// Note that the current implementation can compute the hash of an integral number of bytes only.
-// This is a consequence of the hash interface in which a buffer of bytes is passed in.
-// The internals of the Keccak-f function are computed in keccakf.go.
-// For the detailed specification, refer to the Keccak web site (http://keccak.noekeon.org/).
package sha3
import (
"encoding/binary"
- "hash"
)
-// laneSize is the size in bytes of each "lane" of the internal state of SHA3 (5 * 5 * 8).
-// Note that changing this size would requires using a type other than uint64 to store each lane.
-const laneSize = 8
+// spongeDirection indicates the direction bytes are flowing through the sponge.
+type spongeDirection int
-// sliceSize represents the dimensions of the internal state, a square matrix of
-// sliceSize ** 2 lanes. This is the size of both the "rows" and "columns" dimensions in the
-// terminology of the SHA3 specification.
-const sliceSize = 5
+const (
+ // spongeAbsorbing indicates that the sponge is absorbing input.
+ spongeAbsorbing spongeDirection = iota
+ // spongeSqueezing indicates that the sponge is being squeezed.
+ spongeSqueezing
+)
-// numLanes represents the total number of lanes in the state.
-const numLanes = sliceSize * sliceSize
+const (
+ // maxRate is the maximum size of the internal buffer. SHAKE-256
+ // currently needs the largest buffer.
+ maxRate = 168
+)
-// stateSize is the size in bytes of the internal state of SHA3 (5 * 5 * WSize).
-const stateSize = laneSize * numLanes
+type state struct {
+ // Generic sponge components.
+ a [25]uint64 // main state of the hash
+ buf []byte // points into storage
+ rate int // the number of bytes of state to use
-// digest represents the partial evaluation of a checksum.
-// Note that capacity, and not outputSize, is the critical security parameter, as SHA3 can output
-// an arbitrary number of bytes for any given capacity. The Keccak proposal recommends that
-// capacity = 2*outputSize to ensure that finding a collision of size outputSize requires
-// O(2^{outputSize/2}) computations (the birthday lower bound). Future standards may modify the
-// capacity/outputSize ratio to allow for more output with lower cryptographic security.
-type digest struct {
- a [numLanes]uint64 // main state of the hash
- outputSize int // desired output size in bytes
- capacity int // number of bytes to leave untouched during squeeze/absorb
- absorbed int // number of bytes absorbed thus far
+ // dsbyte contains the "domain separation" value and the first bit of
+ // the padding. In sections 6.1 and 6.2 of [1], the SHA-3 and SHAKE
+ // functions are defined with bits appended to the message: SHA-3
+ // functions have 01 and SHAKE functions have 1111. Because of the way
+ // that bits are numbered from the LSB upwards, that ends up as
+ // 00000010b and 00001111b, respectively. Then the padding rule from
+ // section 5.1 is applied to pad to a multiple of the rate, which
+ // involves adding a 1 bit, zero or more zero bits and then a final one
+ // bit. The first one bit from the padding is merged into the dsbyte
+ // value giving 00000110b (0x06) and 00011111b (0x1f), respectively.
+ //
+ // [1] http://csrc.nist.gov/publications/drafts/fips-202/fips_202_draft.pdf,
+ dsbyte byte
+ storage [maxRate]byte
+
+ // Specific to SHA-3 and SHAKE.
+ fixedOutput bool // whether this is a fixed-ouput-length instance
+ outputLen int // the default output size in bytes
+ state spongeDirection // current direction of the sponge
}
-// minInt returns the lesser of two integer arguments, to simplify the absorption routine.
-func minInt(v1, v2 int) int {
- if v1 <= v2 {
- return v1
- }
- return v2
-}
+// BlockSize returns the rate of sponge underlying this hash function.
+func (d *state) BlockSize() int { return d.rate }
-// rate returns the number of bytes of the internal state which can be absorbed or squeezed
-// in between calls to the permutation function.
-func (d *digest) rate() int {
- return stateSize - d.capacity
-}
+// Size returns the output size of the hash function in bytes.
+func (d *state) Size() int { return d.outputLen }
-// Reset clears the internal state by zeroing bytes in the state buffer.
-// This can be skipped for a newly-created hash state; the default zero-allocated state is correct.
-func (d *digest) Reset() {
- d.absorbed = 0
+// Reset clears the internal state by zeroing the sponge state and
+// the byte buffer, and setting Sponge.state to absorbing.
+func (d *state) Reset() {
+ // Zero the permutation's state.
for i := range d.a {
d.a[i] = 0
}
+ d.state = spongeAbsorbing
+ d.buf = d.storage[:0]
}
-// BlockSize, required by the hash.Hash interface, does not have a standard intepretation
-// for a sponge-based construction like SHA3. We return the data rate: the number of bytes which
-// can be absorbed per invocation of the permutation function. For Merkle-Damgård based hashes
-// (ie SHA1, SHA2, MD5) the output size of the internal compression function is returned.
-// We consider this to be roughly equivalent because it represents the number of bytes of output
-// produced per cryptographic operation.
-func (d *digest) BlockSize() int { return d.rate() }
-
-// Size returns the output size of the hash function in bytes.
-func (d *digest) Size() int {
- return d.outputSize
-}
-
-// unalignedAbsorb is a helper function for Write, which absorbs data that isn't aligned with an
-// 8-byte lane. This requires shifting the individual bytes into position in a uint64.
-func (d *digest) unalignedAbsorb(p []byte) {
- var t uint64
- for i := len(p) - 1; i >= 0; i-- {
- t <<= 8
- t |= uint64(p[i])
+func (d *state) clone() *state {
+ ret := *d
+ if ret.state == spongeAbsorbing {
+ ret.buf = ret.storage[:len(ret.buf)]
+ } else {
+ ret.buf = ret.storage[d.rate-cap(d.buf) : d.rate]
}
- offset := (d.absorbed) % d.rate()
- t <<= 8 * uint(offset%laneSize)
- d.a[offset/laneSize] ^= t
- d.absorbed += len(p)
+
+ return &ret
}
-// Write "absorbs" bytes into the state of the SHA3 hash, updating as needed when the sponge
-// "fills up" with rate() bytes. Since lanes are stored internally as type uint64, this requires
-// converting the incoming bytes into uint64s using a little endian interpretation. This
-// implementation is optimized for large, aligned writes of multiples of 8 bytes (laneSize).
-// Non-aligned or uneven numbers of bytes require shifting and are slower.
-func (d *digest) Write(p []byte) (int, error) {
- // An initial offset is needed if the we aren't absorbing to the first lane initially.
- offset := d.absorbed % d.rate()
- toWrite := len(p)
+// xorIn xors a buffer into the state, byte-swapping to
+// little-endian as necessary; it returns the number of bytes
+// copied, including any zeros appended to the bytestring.
+func (d *state) xorIn(buf []byte) {
+ n := len(buf) / 8
- // The first lane may need to absorb unaligned and/or incomplete data.
- if (offset%laneSize != 0 || len(p) < 8) && len(p) > 0 {
- toAbsorb := minInt(laneSize-(offset%laneSize), len(p))
- d.unalignedAbsorb(p[:toAbsorb])
- p = p[toAbsorb:]
- offset = (d.absorbed) % d.rate()
+ for i := 0; i < n; i++ {
+ a := binary.LittleEndian.Uint64(buf)
+ d.a[i] ^= a
+ buf = buf[8:]
+ }
+ if len(buf) != 0 {
+ // XOR in the last partial ulint64.
+ a := uint64(0)
+ for i, v := range buf {
+ a |= uint64(v) << uint64(8*i)
+ }
+ d.a[n] ^= a
+ }
+}
- // For every rate() bytes absorbed, the state must be permuted via the F Function.
- if (d.absorbed)%d.rate() == 0 {
- keccakF(&d.a)
+// copyOut copies ulint64s to a byte buffer.
+func (d *state) copyOut(b []byte) {
+ for i := 0; len(b) >= 8; i++ {
+ binary.LittleEndian.PutUint64(b, d.a[i])
+ b = b[8:]
+ }
+}
+
+// permute applies the KeccakF-1600 permutation. It handles
+// any input-output buffering.
+func (d *state) permute() {
+ switch d.state {
+ case spongeAbsorbing:
+ // If we're absorbing, we need to xor the input into the state
+ // before applying the permutation.
+ d.xorIn(d.buf)
+ d.buf = d.storage[:0]
+ keccakF1600(&d.a)
+ case spongeSqueezing:
+ // If we're squeezing, we need to apply the permutatin before
+ // copying more output.
+ keccakF1600(&d.a)
+ d.buf = d.storage[:d.rate]
+ d.copyOut(d.buf)
+ }
+}
+
+// pads appends the domain separation bits in dsbyte, applies
+// the multi-bitrate 10..1 padding rule, and permutes the state.
+func (d *state) padAndPermute(dsbyte byte) {
+ if d.buf == nil {
+ d.buf = d.storage[:0]
+ }
+ // Pad with this instance's domain-separator bits. We know that there's
+ // at least one byte of space in d.buf because, if it were full,
+ // permute would have been called to empty it. dsbyte also contains the
+ // first one bit for the padding. See the comment in the state struct.
+ d.buf = append(d.buf, dsbyte)
+ zerosStart := len(d.buf)
+ d.buf = d.storage[:d.rate]
+ for i := zerosStart; i < d.rate; i++ {
+ d.buf[i] = 0
+ }
+ // This adds the final one bit for the padding. Because of the way that
+ // bits are numbered from the LSB upwards, the final bit is the MSB of
+ // the last byte.
+ d.buf[d.rate-1] ^= 0x80
+ // Apply the permutation
+ d.permute()
+ d.state = spongeSqueezing
+ d.buf = d.storage[:d.rate]
+ d.copyOut(d.buf)
+}
+
+// Write absorbs more data into the hash's state. It produces an error
+// if more data is written to the ShakeHash after writing
+func (d *state) Write(p []byte) (written int, err error) {
+ if d.state != spongeAbsorbing {
+ panic("sha3: write to sponge after read")
+ }
+ if d.buf == nil {
+ d.buf = d.storage[:0]
+ }
+ written = len(p)
+
+ for len(p) > 0 {
+ if len(d.buf) == 0 && len(p) >= d.rate {
+ // The fast path; absorb a full "rate" bytes of input and apply the permutation.
+ d.xorIn(p[:d.rate])
+ p = p[d.rate:]
+ keccakF1600(&d.a)
+ } else {
+ // The slow path; buffer the input until we can fill the sponge, and then xor it in.
+ todo := d.rate - len(d.buf)
+ if todo > len(p) {
+ todo = len(p)
+ }
+ d.buf = append(d.buf, p[:todo]...)
+ p = p[todo:]
+
+ // If the sponge is full, apply the permutation.
+ if len(d.buf) == d.rate {
+ d.permute()
+ }
}
}
- // This loop should absorb the bulk of the data into full, aligned lanes.
- // It will call the update function as necessary.
- for len(p) > 7 {
- firstLane := offset / laneSize
- lastLane := minInt(d.rate()/laneSize, firstLane+len(p)/laneSize)
-
- // This inner loop absorbs input bytes into the state in groups of 8, converted to uint64s.
- for lane := firstLane; lane < lastLane; lane++ {
- d.a[lane] ^= binary.LittleEndian.Uint64(p[:laneSize])
- p = p[laneSize:]
- }
- d.absorbed += (lastLane - firstLane) * laneSize
- // For every rate() bytes absorbed, the state must be permuted via the F Function.
- if (d.absorbed)%d.rate() == 0 {
- keccakF(&d.a)
- }
-
- offset = 0
- }
-
- // If there are insufficient bytes to fill the final lane, an unaligned absorption.
- // This should always start at a correct lane boundary though, or else it would be caught
- // by the uneven opening lane case above.
- if len(p) > 0 {
- d.unalignedAbsorb(p)
- }
-
- return toWrite, nil
+ return
}
-// pad computes the SHA3 padding scheme based on the number of bytes absorbed.
-// The padding is a 1 bit, followed by an arbitrary number of 0s and then a final 1 bit, such that
-// the input bits plus padding bits are a multiple of rate(). Adding the padding simply requires
-// xoring an opening and closing bit into the appropriate lanes.
-func (d *digest) pad() {
- offset := d.absorbed % d.rate()
- // The opening pad bit must be shifted into position based on the number of bytes absorbed
- padOpenLane := offset / laneSize
- d.a[padOpenLane] ^= 0x0000000000000001 << uint(8*(offset%laneSize))
- // The closing padding bit is always in the last position
- padCloseLane := (d.rate() / laneSize) - 1
- d.a[padCloseLane] ^= 0x8000000000000000
-}
-
-// finalize prepares the hash to output data by padding and one final permutation of the state.
-func (d *digest) finalize() {
- d.pad()
- keccakF(&d.a)
-}
-
-// squeeze outputs an arbitrary number of bytes from the hash state.
-// Squeezing can require multiple calls to the F function (one per rate() bytes squeezed),
-// although this is not the case for standard SHA3 parameters. This implementation only supports
-// squeezing a single time, subsequent squeezes may lose alignment. Future implementations
-// may wish to support multiple squeeze calls, for example to support use as a PRNG.
-func (d *digest) squeeze(in []byte, toSqueeze int) []byte {
- // Because we read in blocks of laneSize, we need enough room to read
- // an integral number of lanes
- needed := toSqueeze + (laneSize-toSqueeze%laneSize)%laneSize
- if cap(in)-len(in) < needed {
- newIn := make([]byte, len(in), len(in)+needed)
- copy(newIn, in)
- in = newIn
+// Read squeezes an arbitrary number of bytes from the sponge.
+func (d *state) Read(out []byte) (n int, err error) {
+ // If we're still absorbing, pad and apply the permutation.
+ if d.state == spongeAbsorbing {
+ d.padAndPermute(d.dsbyte)
}
- out := in[len(in) : len(in)+needed]
+ n = len(out)
+
+ // Now, do the squeezing.
for len(out) > 0 {
- for i := 0; i < d.rate() && len(out) > 0; i += laneSize {
- binary.LittleEndian.PutUint64(out[:], d.a[i/laneSize])
- out = out[laneSize:]
- }
- if len(out) > 0 {
- keccakF(&d.a)
+ n := copy(out, d.buf)
+ d.buf = d.buf[n:]
+ out = out[n:]
+
+ // Apply the permutation if we've squeezed the sponge dry.
+ if len(d.buf) == 0 {
+ d.permute()
}
}
- return in[:len(in)+toSqueeze] // Re-slice in case we wrote extra data.
+
+ return
}
-// Sum applies padding to the hash state and then squeezes out the desired nubmer of output bytes.
-func (d *digest) Sum(in []byte) []byte {
- // Make a copy of the original hash so that caller can keep writing and summing.
- dup := *d
- dup.finalize()
- return dup.squeeze(in, dup.outputSize)
+// Sum applies padding to the hash state and then squeezes out the desired
+// number of output bytes.
+func (d *state) Sum(in []byte) []byte {
+ // Make a copy of the original hash so that caller can keep writing
+ // and summing.
+ dup := d.clone()
+ hash := make([]byte, dup.outputLen)
+ dup.Read(hash)
+ return append(in, hash...)
}
-
-// The NewKeccakX constructors enable initializing a hash in any of the four recommend sizes
-// from the Keccak specification, all of which set capacity=2*outputSize. Note that the final
-// NIST standard for SHA3 may specify different input/output lengths.
-// The output size is indicated in bits but converted into bytes internally.
-func NewKeccak224() hash.Hash { return &digest{outputSize: 224 / 8, capacity: 2 * 224 / 8} }
-func NewKeccak256() hash.Hash { return &digest{outputSize: 256 / 8, capacity: 2 * 256 / 8} }
-func NewKeccak384() hash.Hash { return &digest{outputSize: 384 / 8, capacity: 2 * 384 / 8} }
-func NewKeccak512() hash.Hash { return &digest{outputSize: 512 / 8, capacity: 2 * 512 / 8} }
