Documentation ¶
Overview ¶
Package merkle computes a deterministic minimal height Merkle tree hash. If the number of items is not a power of two, some leaves will be at different levels. Tries to keep both sides of the tree the same size, but the left may be one greater.
Use this for short deterministic trees, such as the validator list. For larger datasets, use IAVLTree.
Be aware that the current implementation by itself does not prevent second pre-image attacks. Hence, use this library with caution. Otherwise you might run into similar issues as, e.g., in early Bitcoin: https://bitcointalk.org/?topic=102395
* / \ / \ / \ / \ * * / \ / \ / \ / \ / \ / \ * * * h6 / \ / \ / \ h0 h1 h2 h3 h4 h5
TODO(ismail): add 2nd pre-image protection or clarify further on how we use this and why this secure.
Index ¶
- Constants
- Variables
- func IsProofOpReprEmpty(goor ProofOp) (empty bool)
- func IsProofReprEmpty(goor Proof) (empty bool)
- func KeyPathToKeys(path string) (keys [][]byte, err error)
- func SimpleHashFromByteSlices(items [][]byte) []byte
- func SimpleHashFromByteSlicesIterative(input [][]byte) []byte
- func SimpleHashFromMap(m map[string][]byte) []byte
- func SimpleProofsFromMap(m map[string][]byte) (rootHash []byte, proofs map[string]*SimpleProof, keys []string)
- type KVPair
- type Key
- type KeyPath
- type OpDecoder
- type Proof
- type ProofOp
- type ProofOperator
- type ProofOperators
- type ProofRuntime
- func (prt *ProofRuntime) Decode(pop ProofOp) (ProofOperator, error)
- func (prt *ProofRuntime) DecodeProof(proof *Proof) (ProofOperators, error)
- func (prt *ProofRuntime) RegisterOpDecoder(typ string, dec OpDecoder)
- func (prt *ProofRuntime) Verify(proof *Proof, root []byte, keypath string, args [][]byte) (err error)
- func (prt *ProofRuntime) VerifyAbsence(proof *Proof, root []byte, keypath string) (err error)
- func (prt *ProofRuntime) VerifyValue(proof *Proof, root []byte, keypath string, value []byte) (err error)
- type SimpleProof
- type SimpleProofNode
- type SimpleValueOp
- type Tree
Constants ¶
const ( KeyEncodingURL keyEncoding = iota KeyEncodingHex KeyEncodingMax // Number of known encodings. Used for testing )
const ProofOpSimpleValue = "simple:v"
Variables ¶
var Package = amino.RegisterPackage(amino.NewPackage( "github.com/tendermint/classic/crypto/merkle", "tm", amino.GetCallersDirname(), ).WithDependencies().WithTypes( ProofOp{}, Proof{}, ))
Functions ¶
func IsProofOpReprEmpty ¶
func IsProofReprEmpty ¶
func KeyPathToKeys ¶
Decode a path to a list of keys. Path must begin with `/`. Each key must use a known encoding.
func SimpleHashFromByteSlices ¶
SimpleHashFromByteSlices computes a Merkle tree where the leaves are the byte slice, in the provided order.
func SimpleHashFromByteSlicesIterative ¶
SimpleHashFromByteSliceIterative is an iterative alternative to SimpleHashFromByteSlice motivated by potential performance improvements. (#2611) had suggested that an iterative version of SimpleHashFromByteSlice would be faster, presumably because we can envision some overhead accumulating from stack frames and function calls. Additionally, a recursive algorithm risks hitting the stack limit and causing a stack overflow should the tree be too large.
Provided here is an iterative alternative, a simple test to assert correctness and a benchmark. On the performance side, there appears to be no overall difference:
BenchmarkSimpleHashAlternatives/recursive-4 20000 77677 ns/op BenchmarkSimpleHashAlternatives/iterative-4 20000 76802 ns/op
On the surface it might seem that the additional overhead is due to the different allocation patterns of the implementations. The recursive version uses a single [][]byte slices which it then re-slices at each level of the tree. The iterative version reproduces [][]byte once within the function and then rewrites sub-slices of that array at each level of the tree.
Experimenting by modifying the code to simply calculate the hash and not store the result show little to no difference in performance.
These preliminary results suggest:
- The performance of the SimpleHashFromByteSlice is pretty good
- Go has low overhead for recursive functions
- The performance of the SimpleHashFromByteSlice routine is dominated by the actual hashing of data
Although this work is in no way exhaustive, point #3 suggests that optimization of this routine would need to take an alternative approach to make significant improvements on the current performance.
Finally, considering that the recursive implementation is easier to read, it might not be worthwhile to switch to a less intuitive implementation for so little benefit.
func SimpleHashFromMap ¶
SimpleHashFromMap computes a Merkle tree from sorted map. Like calling SimpleHashFromHashers with `item = []byte(Hash(key) | Hash(value))`, sorted by `item`.
func SimpleProofsFromMap ¶
func SimpleProofsFromMap(m map[string][]byte) (rootHash []byte, proofs map[string]*SimpleProof, keys []string)
SimpleProofsFromMap generates proofs from a map. The keys/values of the map will be used as the keys/values in the underlying key-value pairs. The keys are sorted before the proofs are computed.
Types ¶
type KVPair ¶
A local extension to KVPair that can be hashed. Key and value are length prefixed and concatenated, then hashed.
type OpDecoder ¶
type OpDecoder func(ProofOp) (ProofOperator, error)
type Proof ¶
type Proof struct {
Ops []ProofOp `json:"ops"`
}
func (*Proof) FromPBMessage ¶
func (Proof) GetTypeURL ¶
type ProofOp ¶
func (ProofOp) EmptyPBMessage ¶
func (*ProofOp) FromPBMessage ¶
func (ProofOp) GetTypeURL ¶
type ProofOperator ¶
ProofOperator is a layer for calculating intermediate Merkle roots when a series of Merkle trees are chained together. Run() takes leaf values from a tree and returns the Merkle root for the corresponding tree. It takes and returns a list of bytes to allow multiple leaves to be part of a single proof, for instance in a range proof. ProofOp() encodes the ProofOperator in a generic way so it can later be decoded with OpDecoder.
func SimpleValueOpDecoder ¶
func SimpleValueOpDecoder(pop ProofOp) (ProofOperator, error)
type ProofOperators ¶
type ProofOperators []ProofOperator
ProofOperators is a slice of ProofOperator(s). Each operator will be applied to the input value sequentially and the last Merkle root will be verified with already known data
func (ProofOperators) Verify ¶
func (poz ProofOperators) Verify(root []byte, keypath string, args [][]byte) (err error)
func (ProofOperators) VerifyValue ¶
func (poz ProofOperators) VerifyValue(root []byte, keypath string, value []byte) (err error)
type ProofRuntime ¶
type ProofRuntime struct {
// contains filtered or unexported fields
}
func DefaultProofRuntime ¶
func DefaultProofRuntime() (prt *ProofRuntime)
DefaultProofRuntime only knows about Simple value proofs. To use e.g. IAVL proofs, register op-decoders as defined in the IAVL package.
func NewProofRuntime ¶
func NewProofRuntime() *ProofRuntime
func (*ProofRuntime) Decode ¶
func (prt *ProofRuntime) Decode(pop ProofOp) (ProofOperator, error)
func (*ProofRuntime) DecodeProof ¶
func (prt *ProofRuntime) DecodeProof(proof *Proof) (ProofOperators, error)
func (*ProofRuntime) RegisterOpDecoder ¶
func (prt *ProofRuntime) RegisterOpDecoder(typ string, dec OpDecoder)
func (*ProofRuntime) VerifyAbsence ¶
func (prt *ProofRuntime) VerifyAbsence(proof *Proof, root []byte, keypath string) (err error)
TODO In the long run we'll need a method of classifcation of ops, whether existence or absence or perhaps a third?
func (*ProofRuntime) VerifyValue ¶
type SimpleProof ¶
type SimpleProof struct { Total int `json:"total"` // Total number of items. Index int `json:"index"` // Index of item to prove. LeafHash []byte `json:"leaf_hash"` // Hash of item value. Aunts [][]byte `json:"aunts"` // Hashes from leaf's sibling to a root's child. }
SimpleProof represents a simple Merkle proof. NOTE: The convention for proofs is to include leaf hashes but to exclude the root hash. This convention is implemented across IAVL range proofs as well. Keep this consistent unless there's a very good reason to change everything. This also affects the generalized proof system as well.
func SimpleProofsFromByteSlices ¶
func SimpleProofsFromByteSlices(items [][]byte) (rootHash []byte, proofs []*SimpleProof)
SimpleProofsFromByteSlices computes inclusion proof for given items. proofs[0] is the proof for items[0].
func (*SimpleProof) ComputeRootHash ¶
func (sp *SimpleProof) ComputeRootHash() []byte
Compute the root hash given a leaf hash. Does not verify the result.
func (*SimpleProof) String ¶
func (sp *SimpleProof) String() string
String implements the stringer interface for SimpleProof. It is a wrapper around StringIndented.
func (*SimpleProof) StringIndented ¶
func (sp *SimpleProof) StringIndented(indent string) string
StringIndented generates a canonical string representation of a SimpleProof.
func (*SimpleProof) ValidateBasic ¶
func (sp *SimpleProof) ValidateBasic() error
ValidateBasic performs basic validation. NOTE: - it expects LeafHash and Aunts of tmhash.Size size
- it expects no more than 100 aunts
type SimpleProofNode ¶
type SimpleProofNode struct { Hash []byte Parent *SimpleProofNode Left *SimpleProofNode // Left sibling (only one of Left,Right is set) Right *SimpleProofNode // Right sibling (only one of Left,Right is set) }
SimpleProofNode is a helper structure to construct merkle proof. The node and the tree is thrown away afterwards. Exactly one of node.Left and node.Right is nil, unless node is the root, in which case both are nil. node.Parent.Hash = hash(node.Hash, node.Right.Hash) or hash(node.Left.Hash, node.Hash), depending on whether node is a left/right child.
func (*SimpleProofNode) FlattenAunts ¶
func (spn *SimpleProofNode) FlattenAunts() [][]byte
FlattenAunts will return the inner hashes for the item corresponding to the leaf, starting from a leaf SimpleProofNode.
type SimpleValueOp ¶
type SimpleValueOp struct { // To encode in ProofOp.Data Proof *SimpleProof `json:"simple_proof"` // contains filtered or unexported fields }
SimpleValueOp takes a key and a single value as argument and produces the root hash. The corresponding tree structure is the SimpleMap tree. SimpleMap takes a Hasher, and currently Tendermint uses aminoHasher. SimpleValueOp should support the hash function as used in aminoHasher. TODO support additional hash functions here as options/args to this operator.
If the produced root hash matches the expected hash, the proof is good.
func NewSimpleValueOp ¶
func NewSimpleValueOp(key []byte, proof *SimpleProof) SimpleValueOp
func (SimpleValueOp) GetKey ¶
func (op SimpleValueOp) GetKey() []byte
func (SimpleValueOp) ProofOp ¶
func (op SimpleValueOp) ProofOp() ProofOp
func (SimpleValueOp) String ¶
func (op SimpleValueOp) String() string
type Tree ¶
type Tree interface { Size() (size int) Height() (height int8) Has(key []byte) (has bool) Proof(key []byte) (value []byte, proof []byte, exists bool) // TODO make it return an index Get(key []byte) (index int, value []byte, exists bool) GetByIndex(index int) (key []byte, value []byte) Set(key []byte, value []byte) (updated bool) Remove(key []byte) (value []byte, removed bool) HashWithCount() (hash []byte, count int) Hash() (hash []byte) Save() (hash []byte) Load(hash []byte) Copy() Tree Iterate(func(key []byte, value []byte) (stop bool)) (stopped bool) IterateRange(start []byte, end []byte, ascending bool, fx func(key []byte, value []byte) (stop bool)) (stopped bool) }
Tree is a Merkle tree interface.