Caplin: remove merkle tree disk-based caching (#13671)

This is a simplification. Instead keep the state at `head-1` in memory
for reorgs. overall less I/O and less RAM + faster reorg speed

cl/beacon/synced_data/interface.go

@@ -30,6 +30,7 @@ type SyncedData interface {
 	OnHeadState(newState *state.CachingBeaconState) error
 	UnsetHeadState()
 	ViewHeadState(fn ViewHeadStateFn) error
+	ViewPreviousHeadState(fn ViewHeadStateFn) error
 	Syncing() bool
 	HeadSlot() uint64
 	HeadRoot() common.Hash

cl/beacon/synced_data/synced_data.go

@@ -30,7 +30,10 @@ import (
 	"github.com/erigontech/erigon/cl/phase1/core/state"
 )
 
-var ErrNotSynced = errors.New("not synced")
+var (
+	ErrNotSynced                 = errors.New("not synced")
+	ErrPreviousStateNotAvailable = errors.New("previous state not available")
+)
 
 var _ SyncedData = (*SyncedDataManager)(nil)
 
@@ -43,11 +46,11 @@ type SyncedDataManager struct {
 	headRoot atomic.Value
 	headSlot atomic.Uint64
 
-	headState *state.CachingBeaconState
+	headState         *state.CachingBeaconState
+	previousHeadState *state.CachingBeaconState
 
 	accessLock sync.RWMutex // lock used for accessing atomic methods
-
-	mu sync.RWMutex
+	mu         sync.RWMutex
 }
 
 func NewSyncedDataManager(cfg *clparams.BeaconChainConfig, enabled bool) *SyncedDataManager {
@@ -57,6 +60,7 @@ func NewSyncedDataManager(cfg *clparams.BeaconChainConfig, enabled bool) *Synced
 	}
 }
 
+// OnHeadState updates the current head state and tracks the previous state.
 func (s *SyncedDataManager) OnHeadState(newState *state.CachingBeaconState) (err error) {
 	if !s.enabled {
 		return
@@ -67,8 +71,21 @@ func (s *SyncedDataManager) OnHeadState(newState *state.CachingBeaconState) (err
 	s.accessLock.Lock()
 	defer s.accessLock.Unlock()
 
+	// Save current state as previous state, if available.
+	if s.headState != nil {
+		if s.previousHeadState != nil {
+			err = s.headState.CopyInto(s.previousHeadState)
+		} else {
+			s.previousHeadState, err = s.headState.Copy()
+		}
+		if err != nil {
+			return err
+		}
+	}
+
 	var blkRoot common.Hash
 
+	// Update headState with the new state.
 	if s.headState == nil {
 		s.headState, err = newState.Copy()
 	} else {
@@ -83,9 +100,10 @@ func (s *SyncedDataManager) OnHeadState(newState *state.CachingBeaconState) (err
 	}
 	s.headSlot.Store(newState.Slot())
 	s.headRoot.Store(blkRoot)
-	return err
+	return nil
 }
 
+// ViewHeadState allows safe, read-only access to the current head state.
 func (s *SyncedDataManager) ViewHeadState(fn ViewHeadStateFn) error {
 	_, synced := s.headRoot.Load().(common.Hash)
 	if !s.enabled || !synced {
@@ -112,6 +130,16 @@ func (s *SyncedDataManager) ViewHeadState(fn ViewHeadStateFn) error {
 	return nil
 }
 
+// ViewPreviousHeadState allows safe, read-only access to the previous head state.
+func (s *SyncedDataManager) ViewPreviousHeadState(fn ViewHeadStateFn) error {
+	s.mu.RLock()
+	defer s.mu.RUnlock()
+	if s.previousHeadState == nil {
+		return ErrPreviousStateNotAvailable
+	}
+	return fn(s.previousHeadState)
+}
+
 func (s *SyncedDataManager) Syncing() bool {
 	_, synced := s.headRoot.Load().(common.Hash)
 	return !synced
@@ -147,6 +175,7 @@ func (s *SyncedDataManager) UnsetHeadState() {
 	s.headRoot = atomic.Value{}
 	s.headSlot.Store(uint64(0))
 	s.headState = nil
+	s.previousHeadState = nil
 }
 
 func (s *SyncedDataManager) ValidatorPublicKeyByIndex(index int) (common.Bytes48, error) {

cl/clparams/config.go

@@ -124,9 +124,8 @@ const (
 
 var (
 	MainnetBootstrapNodes = []string{
-		// Teku team's bootnode
-		"enr:-KG4QOtcP9X1FbIMOe17QNMKqDxCpm14jcX5tiOE4_TyMrFqbmhPZHK_ZPG2Gxb1GE2xdtodOfx9-cgvNtxnRyHEmC0ghGV0aDKQ9aX9QgAAAAD__________4JpZIJ2NIJpcIQDE8KdiXNlY3AyNTZrMaEDhpehBDbZjM_L9ek699Y7vhUJ-eAdMyQW_Fil522Y0fODdGNwgiMog3VkcIIjKA",
-		"enr:-KG4QL-eqFoHy0cI31THvtZjpYUu_Jdw_MO7skQRJxY1g5HTN1A0epPCU6vi0gLGUgrzpU-ygeMSS8ewVxDpKfYmxMMGhGV0aDKQtTA_KgAAAAD__________4JpZIJ2NIJpcIQ2_DUbiXNlY3AyNTZrMaED8GJ2vzUqgL6-KD1xalo1CsmY4X1HaDnyl6Y_WayCo9GDdGNwgiMog3VkcIIjKA",
+		"enr:-KG4QNTx85fjxABbSq_Rta9wy56nQ1fHK0PewJbGjLm1M4bMGx5-3Qq4ZX2-iFJ0pys_O90sVXNNOxp2E7afBsGsBrgDhGV0aDKQu6TalgMAAAD__________4JpZIJ2NIJpcIQEnfA2iXNlY3AyNTZrMaECGXWQ-rQ2KZKRH1aOW4IlPDBkY4XDphxg9pxKytFCkayDdGNwgiMog3VkcIIjKA",
+		"enr:-KG4QF4B5WrlFcRhUU6dZETwY5ZzAXnA0vGC__L1Kdw602nDZwXSTs5RFXFIFUnbQJmhNGVU6OIX7KVrCSTODsz1tK4DhGV0aDKQu6TalgMAAAD__________4JpZIJ2NIJpcIQExNYEiXNlY3AyNTZrMaECQmM9vp7KhaXhI-nqL_R0ovULLCFSFTa9CPPSdb1zPX6DdGNwgiMog3VkcIIjKA",
 		// Prylab team's bootnodes
 		"enr:-Ku4QImhMc1z8yCiNJ1TyUxdcfNucje3BGwEHzodEZUan8PherEo4sF7pPHPSIB1NNuSg5fZy7qFsjmUKs2ea1Whi0EBh2F0dG5ldHOIAAAAAAAAAACEZXRoMpD1pf1CAAAAAP__________gmlkgnY0gmlwhBLf22SJc2VjcDI1NmsxoQOVphkDqal4QzPMksc5wnpuC3gvSC8AfbFOnZY_On34wIN1ZHCCIyg",
 		"enr:-Ku4QP2xDnEtUXIjzJ_DhlCRN9SN99RYQPJL92TMlSv7U5C1YnYLjwOQHgZIUXw6c-BvRg2Yc2QsZxxoS_pPRVe0yK8Bh2F0dG5ldHOIAAAAAAAAAACEZXRoMpD1pf1CAAAAAP__________gmlkgnY0gmlwhBLf22SJc2VjcDI1NmsxoQMeFF5GrS7UZpAH2Ly84aLK-TyvH-dRo0JM1i8yygH50YN1ZHCCJxA",

cl/cltypes/solid/hash_list.go

@@ -18,7 +18,6 @@ package solid
 
 import (
 	"encoding/json"
-	"io"
 
 	libcommon "github.com/erigontech/erigon-lib/common"
 	"github.com/erigontech/erigon-lib/common/length"
@@ -190,24 +189,3 @@ func (h *hashList) Range(fn func(int, libcommon.Hash, int) bool) {
 func (h *hashList) Pop() libcommon.Hash {
 	panic("didnt ask, dont need it, go fuck yourself")
 }
-
-func (h *hashList) ReadMerkleTree(r io.Reader) error {
-	if h.MerkleTree == nil {
-		h.MerkleTree = &merkle_tree.MerkleTree{}
-		h.MerkleTree.Initialize(h.l, merkle_tree.OptimalMaxTreeCacheDepth, func(idx int, out []byte) {
-			copy(out, h.u[idx*length.Hash:(idx+1)*length.Hash])
-		}, /*limit=*/ nil)
-	}
-	return h.MerkleTree.ReadMerkleTree(r)
-}
-
-func (h *hashList) WriteMerkleTree(w io.Writer) error {
-	if h.MerkleTree == nil {
-		cap := uint64(h.c)
-		h.MerkleTree = &merkle_tree.MerkleTree{}
-		h.MerkleTree.Initialize(h.l, merkle_tree.OptimalMaxTreeCacheDepth, func(idx int, out []byte) {
-			copy(out, h.u[idx*length.Hash:(idx+1)*length.Hash])
-		}, /*limit=*/ &cap)
-	}
-	return h.MerkleTree.WriteMerkleTree(w)
-}

cl/cltypes/solid/hash_vector.go

@@ -18,7 +18,6 @@ package solid
 
 import (
 	"encoding/json"
-	"io"
 
 	libcommon "github.com/erigontech/erigon-lib/common"
 	"github.com/erigontech/erigon-lib/common/length"
@@ -118,11 +117,3 @@ func (h *hashVector) Range(fn func(int, libcommon.Hash, int) bool) {
 func (h *hashVector) Pop() libcommon.Hash {
 	panic("didnt ask, dont need it, go fuck yourself")
 }
-
-func (h *hashVector) ReadMerkleTree(r io.Reader) error {
-	return h.u.ReadMerkleTree(r)
-}
-
-func (h *hashVector) WriteMerkleTree(w io.Writer) error {
-	return h.u.WriteMerkleTree(w)
-}

cl/cltypes/solid/uint64_list.go

@@ -18,7 +18,6 @@ package solid
 
 import (
 	"encoding/json"
-	"io"
 
 	"github.com/erigontech/erigon-lib/types/clonable"
 )
@@ -119,14 +118,6 @@ func (arr *uint64ListSSZ) Append(v uint64) {
 	arr.u.Append(v)
 }
 
-func (arr *uint64ListSSZ) ReadMerkleTree(r io.Reader) error {
-	return arr.u.ReadMerkleTree(r)
-}
-
-func (arr *uint64ListSSZ) WriteMerkleTree(w io.Writer) error {
-	return arr.u.WriteMerkleTree(w)
-}
-
 // Check if it is sorted and check if there are duplicates. O(N) complexity.
 func IsUint64SortedSet(set IterableSSZ[uint64]) bool {
 	for i := 0; i < set.Length()-1; i++ {

cl/cltypes/solid/uint64_vector.go

@@ -18,7 +18,6 @@ package solid
 
 import (
 	"encoding/json"
-	"io"
 
 	"github.com/erigontech/erigon-lib/types/clonable"
 )
@@ -110,11 +109,3 @@ func (arr *uint64VectorSSZ) Pop() uint64 {
 func (arr *uint64VectorSSZ) Append(uint64) {
 	panic("not implemented")
 }
-
-func (arr *uint64VectorSSZ) ReadMerkleTree(r io.Reader) error {
-	return arr.u.ReadMerkleTree(r)
-}
-
-func (arr *uint64VectorSSZ) WriteMerkleTree(w io.Writer) error {
-	return arr.u.WriteMerkleTree(w)
-}

cl/cltypes/solid/uint64slice_byte.go

@@ -19,7 +19,6 @@ package solid
 import (
 	"encoding/binary"
 	"encoding/json"
-	"io"
 	"strconv"
 
 	"github.com/erigontech/erigon-lib/common/length"
@@ -229,24 +228,3 @@ func (arr *byteBasedUint64Slice) DecodeSSZ(buf []byte, _ int) error {
 func (arr *byteBasedUint64Slice) EncodingSizeSSZ() int {
 	return arr.l * 8
 }
-
-func (arr *byteBasedUint64Slice) ReadMerkleTree(r io.Reader) error {
-	if arr.MerkleTree == nil {
-		arr.MerkleTree = &merkle_tree.MerkleTree{}
-		arr.MerkleTree.Initialize((arr.l+3)/4, merkle_tree.OptimalMaxTreeCacheDepth, func(idx int, out []byte) {
-			copy(out, arr.u[idx*length.Hash:])
-		}, nil)
-	}
-	return arr.MerkleTree.ReadMerkleTree(r)
-}
-
-func (arr *byteBasedUint64Slice) WriteMerkleTree(w io.Writer) error {
-	if arr.MerkleTree == nil {
-		arr.MerkleTree = &merkle_tree.MerkleTree{}
-		cap := uint64((arr.c*8 + length.Hash - 1) / length.Hash)
-		arr.MerkleTree.Initialize((arr.l+3)/4, merkle_tree.OptimalMaxTreeCacheDepth, func(idx int, out []byte) {
-			copy(out, arr.u[idx*length.Hash:])
-		}, &cap)
-	}
-	return arr.MerkleTree.WriteMerkleTree(w)
-}

cl/cltypes/solid/validator_set.go

@@ -18,7 +18,6 @@ package solid
 
 import (
 	"encoding/json"
-	"io"
 
 	libcommon "github.com/erigontech/erigon-lib/common"
 	"github.com/erigontech/erigon-lib/types/clonable"
@@ -141,14 +140,23 @@ func (v *ValidatorSet) CopyTo(t *ValidatorSet) {
 			t.MerkleTree = &merkle_tree.MerkleTree{}
 		}
 		v.MerkleTree.CopyInto(t.MerkleTree)
+
+		hashBuffer := make([]byte, 8*32)
 		t.MerkleTree.SetComputeLeafFn(func(idx int, out []byte) {
-			copy(out, t.buffer[idx*validatorSize:])
+			validator := t.Get(idx)
+			if err := validator.CopyHashBufferTo(hashBuffer); err != nil {
+				panic(err)
+			}
+			hashBuffer = hashBuffer[:(8 * 32)]
+			if err := merkle_tree.MerkleRootFromFlatLeaves(hashBuffer, out); err != nil {
+				panic(err)
+			}
 		})
 	} else {
 		t.MerkleTree = nil
 	}
 	// skip copying (unsupported for phase0)
-	t.phase0Data = make([]Phase0Data, t.l)
+	t.phase0Data = make([]Phase0Data, v.l)
 	copy(t.buffer, v.buffer)
 	copy(t.attesterBits, v.attesterBits)
 	t.attesterBits = t.attesterBits[:v.l]
@@ -377,35 +385,3 @@ func (v *ValidatorSet) UnmarshalJSON(data []byte) error {
 	}
 	return nil
 }
-
-func (v *ValidatorSet) ReadMerkleTree(r io.Reader) error {
-	if v.MerkleTree == nil {
-		v.MerkleTree = &merkle_tree.MerkleTree{}
-		hashBuffer := make([]byte, 8*32)
-		v.MerkleTree.Initialize(v.l, merkle_tree.OptimalMaxTreeCacheDepth, func(idx int, out []byte) {
-			validator := v.Get(idx)
-			if err := validator.CopyHashBufferTo(hashBuffer); err != nil {
-				panic(err)
-			}
-			hashBuffer = hashBuffer[:(8 * 32)]
-			if err := merkle_tree.MerkleRootFromFlatLeaves(hashBuffer, out); err != nil {
-				panic(err)
-			}
-		}, nil)
-	}
-	return v.MerkleTree.ReadMerkleTree(r)
-}
-
-func (arr *ValidatorSet) WriteMerkleTree(w io.Writer) error {
-	if arr.MerkleTree == nil {
-		arr.MerkleTree = &merkle_tree.MerkleTree{}
-		cap := uint64(arr.c)
-		arr.MerkleTree.Initialize(arr.l, merkle_tree.OptimalMaxTreeCacheDepth, func(idx int, out []byte) {
-			validator := arr.Get(idx)
-			if err := validator.CopyHashBufferTo(out); err != nil {
-				panic(err)
-			}
-		}, &cap)
-	}
-	return arr.MerkleTree.WriteMerkleTree(w)
-}