nimbus-eth2/beacon_chain/ssz.nim

413 lines
14 KiB
Nim

# beacon_chain
# Copyright (c) 2018 Status Research & Development GmbH
# Licensed and distributed under either of
# * MIT license (license terms in the root directory or at http://opensource.org/licenses/MIT).
# * Apache v2 license (license terms in the root directory or at http://www.apache.org/licenses/LICENSE-2.0).
# at your option. This file may not be copied, modified, or distributed except according to those terms.
# SSZ Serialization (simple serialize)
# See https://github.com/ethereum/eth2.0-specs/blob/master/specs/simple-serialize.md
import
endians, typetraits, options, algorithm, math,
faststreams/input_stream, serialization, eth/common, nimcrypto/keccak,
./spec/[bitfield, crypto, datatypes, digest]
# ################### Helper functions ###################################
export
serialization
type
SszReader* = object
stream: ByteStreamVar
SszWriter* = object
stream: OutputStreamVar
SszError* = object of SerializationError
CorruptedDataError* = object of SszError
RecordWritingMemo = object
initialStreamPos: int
sizePrefixCursor: DelayedWriteCursor
serializationFormat SSZ,
Reader = SszReader,
Writer = SszWriter,
PreferedOutput = seq[byte]
proc init*(T: type SszReader, stream: ByteStreamVar): T =
result.stream = stream
func toSSZType(x: Slot|Epoch): auto = x.uint64
func toSSZType(x: auto): auto = x
# toBytesSSZ convert simple fixed-length types to their SSZ wire representation
func toBytesSSZ(x: SomeInteger): array[sizeof(x), byte] =
## Convert directly to bytes the size of the int. (e.g. ``uint16 = 2 bytes``)
## All integers are serialized as **little endian**.
when x.sizeof == 8: littleEndian64(result.addr, x.unsafeAddr)
elif x.sizeof == 4: littleEndian32(result.addr, x.unsafeAddr)
elif x.sizeof == 2: littleEndian16(result.addr, x.unsafeAddr)
elif x.sizeof == 1: copyMem(result.addr, x.unsafeAddr, sizeof(result))
else: {.fatal: "Unsupported type serialization: " & $(type(x)).name.}
func toBytesSSZ(x: ValidatorIndex): array[3, byte] =
## Integers are all encoded as little endian and not padded
let v = x.uint32
result[0] = byte(v and 0xff)
result[1] = byte((v shr 8) and 0xff)
result[2] = byte((v shr 16) and 0xff)
func toBytesSSZ(x: bool): array[1, byte] =
[if x: 1'u8 else: 0'u8]
func toBytesSSZ(x: EthAddress): array[sizeof(x), byte] = x
func toBytesSSZ(x: Eth2Digest): array[32, byte] = x.data
# TODO these two are still being debated:
# https://github.com/ethereum/eth2.0-specs/issues/308#issuecomment-447026815
func toBytesSSZ(x: ValidatorPubKey|ValidatorSig): auto = x.getBytes()
type
BasicType =
# Types that serialize down to a fixed-length array - most importantly,
# these values don't carry a length prefix in the final encoding. toBytesSSZ
# provides the actual nim-type-to-bytes conversion.
# TODO think about this for a bit - depends where the serialization of
# validator keys ends up going..
# TODO can't put ranges like ValidatorIndex in here:
# https://github.com/nim-lang/Nim/issues/10027
SomeInteger | EthAddress | Eth2Digest | ValidatorPubKey | ValidatorSig |
bool | Slot | Epoch
func sszLen(v: BasicType): int = toBytesSSZ(v.toSSZType()).len
func sszLen(v: ValidatorIndex): int = toBytesSSZ(v).len
func sszLen(v: object | tuple): int =
result = 4 # Length
for field in v.fields:
result += sszLen(type field)
func sszLen(v: seq | array): int =
result = 4 # Length
for i in v:
result += sszLen(i)
func sszLen(v: BitField): int =
sszLen(v.bits)
# fromBytesSSZ copies the wire representation to a Nim variable,
# assuming there's enough data in the buffer
func fromBytesSSZ(T: type SomeInteger, data: openarray[byte]): T =
## Convert directly to bytes the size of the int. (e.g. ``uint16 = 2 bytes``)
## All integers are serialized as **little endian**.
## TODO: Assumes data points to a sufficiently large buffer
doAssert data.len == sizeof(result)
# TODO: any better way to get a suitably aligned buffer in nim???
# see also: https://github.com/nim-lang/Nim/issues/9206
var tmp: uint64
var alignedBuf = cast[ptr byte](tmp.addr)
copyMem(alignedBuf, unsafeAddr data[0], result.sizeof)
when result.sizeof == 8: littleEndian64(result.addr, alignedBuf)
elif result.sizeof == 4: littleEndian32(result.addr, alignedBuf)
elif result.sizeof == 2: littleEndian16(result.addr, alignedBuf)
elif result.sizeof == 1: copyMem(result.addr, alignedBuf, sizeof(result))
else: {.fatal: "Unsupported type deserialization: " & $(type(result)).name.}
func fromBytesSSZ(T: type bool, data: openarray[byte]): T =
# TODO: spec doesn't say what to do if the value is >1 - we'll use the C
# definition for now, but maybe this should be a parse error instead?
fromBytesSSZ(uint8, data) != 0
func fromBytesSSZ(T: type ValidatorIndex, data: openarray[byte]): T =
## Integers are all encoded as littleendian and not padded
doAssert data.len == 3
var tmp: uint32
tmp = tmp or uint32(data[0])
tmp = tmp or uint32(data[1]) shl 8
tmp = tmp or uint32(data[2]) shl 16
result = tmp.ValidatorIndex
func fromBytesSSZ(T: type EthAddress, data: openarray[byte]): T =
doAssert data.len == sizeof(result)
copyMem(result.addr, unsafeAddr data[0], sizeof(result))
func fromBytesSSZ(T: type Eth2Digest, data: openarray[byte]): T =
doAssert data.len == sizeof(result.data)
copyMem(result.data.addr, unsafeAddr data[0], sizeof(result.data))
proc init*(T: type SszWriter, stream: OutputStreamVar): T =
result.stream = stream
proc writeValue*(w: var SszWriter, obj: auto)
# This is an alternative lower-level API useful for RPC
# frameworks that can simulate the serialization of an
# object without constructing an actual instance:
proc beginRecord*(w: var SszWriter, T: type): RecordWritingMemo =
result.initialStreamPos = w.stream.pos
result.sizePrefixCursor = w.stream.delayFixedSizeWrite sizeof(uint32)
template writeField*(w: var SszWriter, name: string, value: auto) =
w.writeValue(value)
proc endRecord*(w: var SszWriter, memo: RecordWritingMemo) =
let finalSize = uint32(w.stream.pos - memo.initialStreamPos - 4)
memo.sizePrefixCursor.endWrite(finalSize.toBytesSSZ)
proc writeValue*(w: var SszWriter, obj: auto) =
# We are not using overloads here, because this leads to
# slightly better error messages when the user provides
# additional overloads for `writeValue`.
mixin writeValue
when obj is ValidatorIndex|BasicType:
w.stream.append obj.toSSZType().toBytesSSZ
elif obj is enum:
w.stream.append uint64(obj).toBytesSSZ
else:
let memo = w.beginRecord(obj.type)
when obj is seq|array|openarray:
# If you get an error here that looks like:
# type mismatch: got <type range 0..8191(uint64)>
# you just used an unsigned int for an array index thinking you'd get
# away with it (surprise, surprise: you can't, uints are crippled!)
# https://github.com/nim-lang/Nim/issues/9984
for elem in obj:
w.writeValue elem
elif obj is BitField:
for elem in obj.bits:
w.writeValue elem
else:
obj.serializeFields(fieldName, field):
# for research/serialized_sizes, remove when appropriate
when defined(debugFieldSizes) and obj is (BeaconState|BeaconBlock):
let start = w.stream.pos
w.writeValue field.toSSZType
debugEcho fieldName, ": ", w.stream.pos - start
else:
w.writeValue field.toSSZType
w.endRecord(memo)
proc readValue*(r: var SszReader, result: var auto) =
# We are not using overloads here, because this leads to
# slightly better error messages when the user provides
# additional overloads for `readValue`.
type T = result.type
mixin readValue
template checkEof(n: int) =
if not r.stream[].ensureBytes(n):
raise newException(UnexpectedEofError, "SSZ has insufficient number of bytes")
when result is ValidatorIndex|BasicType:
let bytesToRead = result.sszLen;
checkEof bytesToRead
when result is ValidatorPubKey|ValidatorSig:
if not result.init(r.stream.readBytes(bytesToRead)):
raise newException(CorruptedDataError, "Failed to load a BLS key or signature")
else:
result = T.fromBytesSSZ(r.stream.readBytes(bytesToRead))
elif result is enum:
# TODO what to do with out-of-range values?? rejecting means breaking
# forwards compatibility..
result = cast[T](r.readValue(uint64))
elif result is string:
{.error: "The SSZ format doesn't support the string type yet".}
else:
let totalLen = int r.readValue(uint32)
checkEof totalLen
let endPos = r.stream[].pos + totalLen
when T is seq:
type ElemType = type(result[0])
# Items are of homogenous type, but not necessarily homogenous length,
# cannot pre-allocate item list generically
while r.stream[].pos < endPos:
result.add r.readValue(ElemType)
elif T is BitField:
type ElemType = type(result.bits[0])
while r.stream[].pos < endPos:
result.bits.add r.readValue(ElemType)
elif T is array:
type ElemType = type(result[0])
var i = 0
while r.stream[].pos < endPos:
if i > result.len:
raise newException(CorruptedDataError, "SSZ includes unexpected bytes past the end of an array")
result[i] = r.readValue(ElemType)
i += 1
else:
result.deserializeFields(fieldName, field):
# TODO This hardcoding's ugly; generalize & abstract.
when field is Slot:
field = r.readValue(uint64).Slot
elif field is Epoch:
field = r.readValue(uint64).Epoch
else:
field = r.readValue(field.type)
if r.stream[].pos != endPos:
raise newException(CorruptedDataError, "SSZ includes unexpected bytes past the end of the deserialized object")
# ################### Hashing ###################################
# Sample hash_tree_root implementation based on:
# https://github.com/ethereum/eth2.0-specs/blob/v0.5.1/specs/simple-serialize.md
# https://github.com/ethereum/eth2.0-specs/blob/v0.5.1/utils/phase0/minimal_ssz.py
# TODO Probably wrong - the spec is pretty bare-bones and no test vectors yet
const
BYTES_PER_CHUNK = 32
# ################### Hashing helpers ###################################
# TODO varargs openarray, anyone?
template withHash(body: untyped): array[32, byte] =
let tmp = withEth2Hash: body
toBytesSSZ tmp
func hash(a, b: openArray[byte]): array[32, byte] =
withHash:
h.update(a)
h.update(b)
type
Chunk = array[BYTES_PER_CHUNK, byte]
# TODO: er, how is this _actually_ done?
# Mandatory bug: https://github.com/nim-lang/Nim/issues/9825
func empty(T: type): T = discard
const emptyChunk = empty(Chunk)
func mix_in_length(root: Chunk, length: int): Chunk =
var dataLen: array[32, byte]
var lstLen = uint64(length)
littleEndian64(dataLen[32-8].addr, lstLen.addr)
hash(root, dataLen)
proc pack(values: seq|array): iterator(): Chunk =
result = iterator (): Chunk =
# TODO should be trivial to avoid this seq also..
# TODO I get a feeling a copy of the array is taken to the closure, which
# also needs fixing
# TODO avoid closure iterators that involve GC
var tmp =
newSeqOfCap[byte](values.len() * sizeof(toBytesSSZ(values[0].toSSZType())))
for v in values:
tmp.add toBytesSSZ(v.toSSZType)
for v in 0..<tmp.len div sizeof(Chunk):
var c: Chunk
copyMem(addr c, addr tmp[v * sizeof(Chunk)], sizeof(Chunk))
yield c
let remains = tmp.len mod sizeof(Chunk)
if remains != 0:
var c: Chunk
copyMem(addr c, addr tmp[tmp.len - remains], remains)
yield c
proc pad(iter: iterator(): Chunk): iterator(): Chunk =
# Pad a list of chunks to the next power-of-two length with empty chunks -
# this includes ensuring there's at least one chunk return
result = iterator(): Chunk =
var count = 0
while true:
let item = iter()
if finished(iter): break
count += 1
yield item
doAssert nextPowerOfTwo(0) == 1,
"Usefully, empty lists will be padded to one empty block"
for _ in count..<nextPowerOfTwo(count):
yield emptyChunk
func merkleize(chunker: iterator(): Chunk): Chunk =
var
stack: seq[tuple[height: int, chunk: Chunk]]
paddedChunker = pad(chunker)
while true:
let chunk = paddedChunker()
if finished(paddedChunker): break
# Leaves start at height 0 - every time they move up, height is increased
# allowing us to detect two chunks at the same height ready for
# consolidation
# See also: http://szydlo.com/logspacetime03.pdf
stack.add (0, chunk)
# Consolidate items of the same height - this keeps stack size at log N
while stack.len > 1 and stack[^1].height == stack[^2].height:
# As tradition dictates - one feature, at least one nim bug:
# https://github.com/nim-lang/Nim/issues/9684
let tmp = hash(stack[^2].chunk, stack[^1].chunk)
stack[^2].height += 1
stack[^2].chunk = tmp
discard stack.pop
doAssert stack.len == 1,
"With power-of-two leaves, we should end up with a single root"
stack[0].chunk
template elementType[T, N](_: type array[N, T]): typedesc = T
template elementType[T](_: type seq[T]): typedesc = T
func hash_tree_root*[T](value: T): Eth2Digest =
# Merkle tree
Eth2Digest(data:
when T is BasicType:
merkleize(pack([value]))
elif T is array|seq:
when T.elementType() is BasicType:
mix_in_length(merkleize(pack(value)), len(value))
else:
var roots = iterator(): Chunk =
for v in value:
yield hash_tree_root(v).data
mix_in_length(merkleize(roots), len(value))
elif T is object:
var roots = iterator(): Chunk =
for v in value.fields:
yield hash_tree_root(v).data
merkleize(roots)
else:
static: doAssert false, "Unexpected type: " & T.name
)
# https://github.com/ethereum/eth2.0-specs/blob/0.4.0/specs/simple-serialize.md#signed-roots
func signed_root*[T: object](x: T): Eth2Digest =
# TODO write tests for this (check vs hash_tree_root)
var found_field_name = false
var roots = iterator(): Chunk =
for name, field in x.fieldPairs:
# TODO we should truncate the last field, regardless of its name.. this
# hack works for now - how to skip the last fieldPair though??
if name == "signature":
found_field_name = true
break
yield hash_tree_root(field).data
let root = merkleize(roots)
doAssert found_field_name
Eth2Digest(data: root)