go-ethereum/trie/trie.go

511 lines
16 KiB
Go
Raw Normal View History

2015-07-07 03:54:22 +03:00
// Copyright 2014 The go-ethereum Authors
// This file is part of the go-ethereum library.
2015-07-07 03:54:22 +03:00
//
// The go-ethereum library is free software: you can redistribute it and/or modify
2015-07-07 03:54:22 +03:00
// it under the terms of the GNU Lesser General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// The go-ethereum library is distributed in the hope that it will be useful,
2015-07-07 03:54:22 +03:00
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2015-07-07 03:54:22 +03:00
// GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
2015-07-07 03:54:22 +03:00
2015-07-07 06:08:16 +03:00
// Package trie implements Merkle Patricia Tries.
2014-10-31 15:45:03 +02:00
package trie
2014-02-15 00:56:09 +02:00
import (
2014-07-02 18:47:18 +03:00
"bytes"
2014-02-15 00:56:09 +02:00
"fmt"
2014-08-04 11:38:18 +03:00
2015-03-16 12:27:38 +02:00
"github.com/ethereum/go-ethereum/common"
"github.com/ethereum/go-ethereum/crypto/sha3"
"github.com/ethereum/go-ethereum/log"
"github.com/rcrowley/go-metrics"
2014-02-15 00:56:09 +02:00
)
2015-07-06 02:19:48 +03:00
var (
// This is the known root hash of an empty trie.
emptyRoot = common.HexToHash("56e81f171bcc55a6ff8345e692c0f86e5b48e01b996cadc001622fb5e363b421")
// This is the known hash of an empty state trie entry.
emptyState common.Hash
2015-07-06 02:19:48 +03:00
)
2014-02-15 00:56:09 +02:00
var (
cacheMissCounter = metrics.NewRegisteredCounter("trie/cachemiss", nil)
cacheUnloadCounter = metrics.NewRegisteredCounter("trie/cacheunload", nil)
)
// CacheMisses retrieves a global counter measuring the number of cache misses
// the trie had since process startup. This isn't useful for anything apart from
// trie debugging purposes.
func CacheMisses() int64 {
return cacheMissCounter.Count()
}
// CacheUnloads retrieves a global counter measuring the number of cache unloads
// the trie did since process startup. This isn't useful for anything apart from
// trie debugging purposes.
func CacheUnloads() int64 {
return cacheUnloadCounter.Count()
}
func init() {
sha3.NewKeccak256().Sum(emptyState[:0])
}
2015-07-06 02:19:48 +03:00
// Database must be implemented by backing stores for the trie.
type Database interface {
DatabaseReader
2015-07-06 02:19:48 +03:00
DatabaseWriter
}
// DatabaseReader wraps the Get method of a backing store for the trie.
type DatabaseReader interface {
2015-07-06 02:19:48 +03:00
Get(key []byte) (value []byte, err error)
2014-06-30 14:08:00 +03:00
}
2015-07-06 02:19:48 +03:00
// DatabaseWriter wraps the Put method of a backing store for the trie.
type DatabaseWriter interface {
// Put stores the mapping key->value in the database.
// Implementations must not hold onto the value bytes, the trie
// will reuse the slice across calls to Put.
Put(key, value []byte) error
2014-02-15 00:56:09 +02:00
}
2015-07-06 02:19:48 +03:00
// Trie is a Merkle Patricia Trie.
// The zero value is an empty trie with no database.
// Use New to create a trie that sits on top of a database.
//
// Trie is not safe for concurrent use.
type Trie struct {
root node
db Database
originalRoot common.Hash
// Cache generation values.
// cachegen increases by one with each commit operation.
// new nodes are tagged with the current generation and unloaded
// when their generation is older than than cachegen-cachelimit.
cachegen, cachelimit uint16
}
// SetCacheLimit sets the number of 'cache generations' to keep.
// A cache generation is created by a call to Commit.
func (t *Trie) SetCacheLimit(l uint16) {
t.cachelimit = l
}
// newFlag returns the cache flag value for a newly created node.
func (t *Trie) newFlag() nodeFlag {
return nodeFlag{dirty: true, gen: t.cachegen}
2014-02-24 13:11:00 +02:00
}
2015-07-06 02:19:48 +03:00
// New creates a trie with an existing root node from db.
//
// If root is the zero hash or the sha3 hash of an empty string, the
// trie is initially empty and does not require a database. Otherwise,
// New will panic if db is nil and returns a MissingNodeError if root does
// not exist in the database. Accessing the trie loads nodes from db on demand.
2015-07-06 02:19:48 +03:00
func New(root common.Hash, db Database) (*Trie, error) {
trie := &Trie{db: db, originalRoot: root}
2015-07-06 02:19:48 +03:00
if (root != common.Hash{}) && root != emptyRoot {
if db == nil {
panic("trie.New: cannot use existing root without a database")
2014-02-15 00:56:09 +02:00
}
rootnode, err := trie.resolveHash(root[:], nil, nil)
if err != nil {
return nil, err
2015-07-06 02:19:48 +03:00
}
trie.root = rootnode
2014-02-15 00:56:09 +02:00
}
2015-07-06 02:19:48 +03:00
return trie, nil
2014-02-15 00:56:09 +02:00
}
// NodeIterator returns an iterator that returns nodes of the trie. Iteration starts at
// the key after the given start key.
func (t *Trie) NodeIterator(start []byte) NodeIterator {
return newNodeIterator(t, start)
2014-04-29 13:36:27 +03:00
}
2015-07-06 02:19:48 +03:00
// Get returns the value for key stored in the trie.
// The value bytes must not be modified by the caller.
func (t *Trie) Get(key []byte) []byte {
res, err := t.TryGet(key)
if err != nil {
log.Error(fmt.Sprintf("Unhandled trie error: %v", err))
}
return res
}
// TryGet returns the value for key stored in the trie.
// The value bytes must not be modified by the caller.
// If a node was not found in the database, a MissingNodeError is returned.
func (t *Trie) TryGet(key []byte) ([]byte, error) {
key = keybytesToHex(key)
value, newroot, didResolve, err := t.tryGet(t.root, key, 0)
if err == nil && didResolve {
t.root = newroot
}
return value, err
}
func (t *Trie) tryGet(origNode node, key []byte, pos int) (value []byte, newnode node, didResolve bool, err error) {
switch n := (origNode).(type) {
case nil:
return nil, nil, false, nil
case valueNode:
return n, n, false, nil
case *shortNode:
if len(key)-pos < len(n.Key) || !bytes.Equal(n.Key, key[pos:pos+len(n.Key)]) {
// key not found in trie
return nil, n, false, nil
}
value, newnode, didResolve, err = t.tryGet(n.Val, key, pos+len(n.Key))
if err == nil && didResolve {
n = n.copy()
n.Val = newnode
n.flags.gen = t.cachegen
}
return value, n, didResolve, err
case *fullNode:
value, newnode, didResolve, err = t.tryGet(n.Children[key[pos]], key, pos+1)
if err == nil && didResolve {
n = n.copy()
n.flags.gen = t.cachegen
n.Children[key[pos]] = newnode
}
return value, n, didResolve, err
case hashNode:
child, err := t.resolveHash(n, key[:pos], key[pos:])
if err != nil {
return nil, n, true, err
2015-07-06 02:19:48 +03:00
}
value, newnode, _, err := t.tryGet(child, key, pos)
return value, newnode, true, err
default:
panic(fmt.Sprintf("%T: invalid node: %v", origNode, origNode))
2014-10-29 15:20:42 +02:00
}
2014-02-15 00:56:09 +02:00
}
2015-07-06 02:19:48 +03:00
// Update associates key with value in the trie. Subsequent calls to
// Get will return value. If value has length zero, any existing value
// is deleted from the trie and calls to Get will return nil.
//
// The value bytes must not be modified by the caller while they are
// stored in the trie.
func (t *Trie) Update(key, value []byte) {
if err := t.TryUpdate(key, value); err != nil {
log.Error(fmt.Sprintf("Unhandled trie error: %v", err))
}
}
// TryUpdate associates key with value in the trie. Subsequent calls to
// Get will return value. If value has length zero, any existing value
// is deleted from the trie and calls to Get will return nil.
//
// The value bytes must not be modified by the caller while they are
// stored in the trie.
//
// If a node was not found in the database, a MissingNodeError is returned.
func (t *Trie) TryUpdate(key, value []byte) error {
k := keybytesToHex(key)
2015-01-08 12:47:04 +02:00
if len(value) != 0 {
_, n, err := t.insert(t.root, nil, k, valueNode(value))
if err != nil {
return err
}
t.root = n
} else {
_, n, err := t.delete(t.root, nil, k)
if err != nil {
return err
}
t.root = n
}
return nil
2014-07-02 14:40:02 +03:00
}
func (t *Trie) insert(n node, prefix, key []byte, value node) (bool, node, error) {
2015-01-08 12:47:04 +02:00
if len(key) == 0 {
if v, ok := n.(valueNode); ok {
return !bytes.Equal(v, value.(valueNode)), value, nil
}
return true, value, nil
2014-02-15 00:56:09 +02:00
}
2015-07-06 02:19:48 +03:00
switch n := n.(type) {
case *shortNode:
2015-07-06 02:19:48 +03:00
matchlen := prefixLen(key, n.Key)
// If the whole key matches, keep this short node as is
// and only update the value.
if matchlen == len(n.Key) {
dirty, nn, err := t.insert(n.Val, append(prefix, key[:matchlen]...), key[matchlen:], value)
if !dirty || err != nil {
return false, n, err
}
return true, &shortNode{n.Key, nn, t.newFlag()}, nil
2015-01-08 12:47:04 +02:00
}
2015-07-06 02:19:48 +03:00
// Otherwise branch out at the index where they differ.
branch := &fullNode{flags: t.newFlag()}
var err error
_, branch.Children[n.Key[matchlen]], err = t.insert(nil, append(prefix, n.Key[:matchlen+1]...), n.Key[matchlen+1:], n.Val)
if err != nil {
return false, nil, err
}
_, branch.Children[key[matchlen]], err = t.insert(nil, append(prefix, key[:matchlen+1]...), key[matchlen+1:], value)
if err != nil {
return false, nil, err
}
2015-07-06 02:19:48 +03:00
// Replace this shortNode with the branch if it occurs at index 0.
if matchlen == 0 {
return true, branch, nil
2014-02-15 00:56:09 +02:00
}
2015-07-06 02:19:48 +03:00
// Otherwise, replace it with a short node leading up to the branch.
return true, &shortNode{key[:matchlen], branch, t.newFlag()}, nil
2014-02-15 00:56:09 +02:00
case *fullNode:
dirty, nn, err := t.insert(n.Children[key[0]], append(prefix, key[0]), key[1:], value)
if !dirty || err != nil {
return false, n, err
}
n = n.copy()
n.flags = t.newFlag()
n.Children[key[0]] = nn
return true, n, nil
2014-02-15 00:56:09 +02:00
2015-07-06 02:19:48 +03:00
case nil:
return true, &shortNode{key, value, t.newFlag()}, nil
2014-02-15 00:56:09 +02:00
2015-07-06 02:19:48 +03:00
case hashNode:
// We've hit a part of the trie that isn't loaded yet. Load
// the node and insert into it. This leaves all child nodes on
// the path to the value in the trie.
rn, err := t.resolveHash(n, prefix, key)
if err != nil {
return false, nil, err
}
dirty, nn, err := t.insert(rn, prefix, key, value)
if !dirty || err != nil {
return false, rn, err
}
return true, nn, nil
2014-02-15 00:56:09 +02:00
2015-01-08 12:47:04 +02:00
default:
2015-07-06 02:19:48 +03:00
panic(fmt.Sprintf("%T: invalid node: %v", n, n))
2014-02-15 00:56:09 +02:00
}
}
2015-07-06 02:19:48 +03:00
// Delete removes any existing value for key from the trie.
func (t *Trie) Delete(key []byte) {
if err := t.TryDelete(key); err != nil {
log.Error(fmt.Sprintf("Unhandled trie error: %v", err))
}
}
// TryDelete removes any existing value for key from the trie.
// If a node was not found in the database, a MissingNodeError is returned.
func (t *Trie) TryDelete(key []byte) error {
k := keybytesToHex(key)
_, n, err := t.delete(t.root, nil, k)
if err != nil {
return err
}
t.root = n
return nil
2014-02-15 00:56:09 +02:00
}
2015-07-06 02:19:48 +03:00
// delete returns the new root of the trie with key deleted.
// It reduces the trie to minimal form by simplifying
// nodes on the way up after deleting recursively.
func (t *Trie) delete(n node, prefix, key []byte) (bool, node, error) {
2015-07-06 02:19:48 +03:00
switch n := n.(type) {
case *shortNode:
2015-07-06 02:19:48 +03:00
matchlen := prefixLen(key, n.Key)
if matchlen < len(n.Key) {
return false, n, nil // don't replace n on mismatch
2015-07-06 02:19:48 +03:00
}
if matchlen == len(key) {
return true, nil, nil // remove n entirely for whole matches
2015-07-06 02:19:48 +03:00
}
// The key is longer than n.Key. Remove the remaining suffix
// from the subtrie. Child can never be nil here since the
// subtrie must contain at least two other values with keys
// longer than n.Key.
dirty, child, err := t.delete(n.Val, append(prefix, key[:len(n.Key)]...), key[len(n.Key):])
if !dirty || err != nil {
return false, n, err
}
2015-07-06 02:19:48 +03:00
switch child := child.(type) {
case *shortNode:
2015-07-06 02:19:48 +03:00
// Deleting from the subtrie reduced it to another
// short node. Merge the nodes to avoid creating a
// shortNode{..., shortNode{...}}. Use concat (which
// always creates a new slice) instead of append to
// avoid modifying n.Key since it might be shared with
// other nodes.
return true, &shortNode{concat(n.Key, child.Key...), child.Val, t.newFlag()}, nil
2015-07-06 02:19:48 +03:00
default:
return true, &shortNode{n.Key, child, t.newFlag()}, nil
2014-02-20 15:40:00 +02:00
}
case *fullNode:
dirty, nn, err := t.delete(n.Children[key[0]], append(prefix, key[0]), key[1:])
if !dirty || err != nil {
return false, n, err
}
n = n.copy()
n.flags = t.newFlag()
n.Children[key[0]] = nn
2015-07-06 02:19:48 +03:00
// Check how many non-nil entries are left after deleting and
// reduce the full node to a short node if only one entry is
// left. Since n must've contained at least two children
// before deletion (otherwise it would not be a full node) n
// can never be reduced to nil.
//
// When the loop is done, pos contains the index of the single
// value that is left in n or -2 if n contains at least two
// values.
2015-01-08 12:47:04 +02:00
pos := -1
for i, cld := range n.Children {
2015-07-06 02:19:48 +03:00
if cld != nil {
2015-01-08 12:47:04 +02:00
if pos == -1 {
pos = i
2014-02-20 15:40:00 +02:00
} else {
2015-01-08 12:47:04 +02:00
pos = -2
2015-07-06 02:19:48 +03:00
break
2014-02-20 15:40:00 +02:00
}
}
}
2015-07-06 02:19:48 +03:00
if pos >= 0 {
if pos != 16 {
// If the remaining entry is a short node, it replaces
// n and its key gets the missing nibble tacked to the
// front. This avoids creating an invalid
// shortNode{..., shortNode{...}}. Since the entry
// might not be loaded yet, resolve it just for this
// check.
cnode, err := t.resolve(n.Children[pos], prefix, []byte{byte(pos)})
if err != nil {
return false, nil, err
}
if cnode, ok := cnode.(*shortNode); ok {
2015-07-06 02:19:48 +03:00
k := append([]byte{byte(pos)}, cnode.Key...)
return true, &shortNode{k, cnode.Val, t.newFlag()}, nil
2015-07-06 02:19:48 +03:00
}
2015-01-08 12:47:04 +02:00
}
2015-07-06 02:19:48 +03:00
// Otherwise, n is replaced by a one-nibble short node
// containing the child.
return true, &shortNode{[]byte{byte(pos)}, n.Children[pos], t.newFlag()}, nil
2014-02-20 15:40:00 +02:00
}
2015-07-06 02:19:48 +03:00
// n still contains at least two values and cannot be reduced.
return true, n, nil
2014-02-20 15:40:00 +02:00
case valueNode:
return true, nil, nil
2015-01-08 12:47:04 +02:00
case nil:
return false, nil, nil
2015-07-06 02:19:48 +03:00
case hashNode:
// We've hit a part of the trie that isn't loaded yet. Load
// the node and delete from it. This leaves all child nodes on
// the path to the value in the trie.
rn, err := t.resolveHash(n, prefix, key)
if err != nil {
return false, nil, err
}
dirty, nn, err := t.delete(rn, prefix, key)
if !dirty || err != nil {
return false, rn, err
}
return true, nn, nil
2015-07-06 02:19:48 +03:00
2015-01-08 12:47:04 +02:00
default:
2015-07-06 02:19:48 +03:00
panic(fmt.Sprintf("%T: invalid node: %v (%v)", n, n, key))
2014-02-20 15:40:00 +02:00
}
2014-02-24 13:11:00 +02:00
}
2015-07-06 02:19:48 +03:00
func concat(s1 []byte, s2 ...byte) []byte {
r := make([]byte, len(s1)+len(s2))
copy(r, s1)
copy(r[len(s1):], s2)
return r
}
func (t *Trie) resolve(n node, prefix, suffix []byte) (node, error) {
2015-07-06 02:19:48 +03:00
if n, ok := n.(hashNode); ok {
return t.resolveHash(n, prefix, suffix)
2015-07-06 02:19:48 +03:00
}
return n, nil
2015-07-06 02:19:48 +03:00
}
func (t *Trie) resolveHash(n hashNode, prefix, suffix []byte) (node, error) {
cacheMissCounter.Inc(1)
2015-07-06 02:19:48 +03:00
enc, err := t.db.Get(n)
if err != nil || enc == nil {
return nil, &MissingNodeError{
RootHash: t.originalRoot,
NodeHash: common.BytesToHash(n),
PrefixLen: len(prefix),
SuffixLen: len(suffix),
2014-02-24 13:11:00 +02:00
}
2015-07-06 02:19:48 +03:00
}
dec := mustDecodeNode(n, enc, t.cachegen)
return dec, nil
2015-07-06 02:19:48 +03:00
}
// Root returns the root hash of the trie.
// Deprecated: use Hash instead.
func (t *Trie) Root() []byte { return t.Hash().Bytes() }
2014-02-24 13:11:00 +02:00
2015-07-06 02:19:48 +03:00
// Hash returns the root hash of the trie. It does not write to the
// database and can be used even if the trie doesn't have one.
func (t *Trie) Hash() common.Hash {
hash, cached, _ := t.hashRoot(nil)
t.root = cached
return common.BytesToHash(hash.(hashNode))
2014-02-24 13:11:00 +02:00
}
2015-07-06 02:19:48 +03:00
// Commit writes all nodes to the trie's database.
// Nodes are stored with their sha3 hash as the key.
//
// Committing flushes nodes from memory.
// Subsequent Get calls will load nodes from the database.
func (t *Trie) Commit() (root common.Hash, err error) {
if t.db == nil {
panic("Commit called on trie with nil database")
2014-02-24 13:11:00 +02:00
}
2015-07-06 02:19:48 +03:00
return t.CommitTo(t.db)
2014-02-24 13:11:00 +02:00
}
2015-07-06 02:19:48 +03:00
// CommitTo writes all nodes to the given database.
// Nodes are stored with their sha3 hash as the key.
//
// Committing flushes nodes from memory. Subsequent Get calls will
// load nodes from the trie's database. Calling code must ensure that
// the changes made to db are written back to the trie's attached
// database before using the trie.
func (t *Trie) CommitTo(db DatabaseWriter) (root common.Hash, err error) {
hash, cached, err := t.hashRoot(db)
2015-07-06 02:19:48 +03:00
if err != nil {
return (common.Hash{}), err
}
t.root = cached
t.cachegen++
return common.BytesToHash(hash.(hashNode)), nil
2015-07-06 02:19:48 +03:00
}
2014-05-27 02:08:51 +03:00
func (t *Trie) hashRoot(db DatabaseWriter) (node, node, error) {
2015-07-06 02:19:48 +03:00
if t.root == nil {
return hashNode(emptyRoot.Bytes()), nil, nil
2015-07-06 02:19:48 +03:00
}
h := newHasher(t.cachegen, t.cachelimit)
defer returnHasherToPool(h)
return h.hash(t.root, db, true)
2014-05-27 02:08:51 +03:00
}