VKT state migration optimized methods & lower LeafNode memory pressure #338
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,196 @@ | ||
| package verkle | ||
|
|
||
| import ( | ||
| "bytes" | ||
| "sort" | ||
| ) | ||
|
|
||
| // BatchNewLeafNodeData is a struct that contains the data needed to create a new leaf node. | ||
| type BatchNewLeafNodeData struct { | ||
| Stem []byte | ||
| Values map[byte][]byte | ||
| } | ||
|
|
||
| // BatchNewLeafNode creates a new leaf node from the given data. It optimizes LeafNode creation | ||
| // by batching expensive cryptography operations. It returns the LeafNodes sorted by stem. | ||
| func BatchNewLeafNode(nodesValues []BatchNewLeafNodeData) []LeafNode { | ||
| cfg := GetConfig() | ||
|
|
||
| ret := make([]LeafNode, len(nodesValues)) | ||
| c1c2points := make([]*Point, 2*len(nodesValues)) | ||
| c1c2frs := make([]*Fr, 2*len(nodesValues)) | ||
| for i, nv := range nodesValues { | ||
| ret[i] = LeafNode{ | ||
| values: nv.Values, | ||
| stem: nv.Stem, | ||
| c1: Generator(), | ||
| c2: Generator(), | ||
| } | ||
|
|
||
| var c1poly, c2poly [NodeWidth]Fr | ||
|
|
||
| valsslice := make([][]byte, NodeWidth) | ||
| for idx := range nv.Values { | ||
| valsslice[idx] = nv.Values[idx] | ||
| } | ||
|
|
||
| fillSuffixTreePoly(c1poly[:], valsslice[:NodeWidth/2]) | ||
| ret[i].c1 = cfg.CommitToPoly(c1poly[:], 0) | ||
| fillSuffixTreePoly(c2poly[:], valsslice[NodeWidth/2:]) | ||
| ret[i].c2 = cfg.CommitToPoly(c2poly[:], 0) | ||
|
|
||
| c1c2points[2*i], c1c2points[2*i+1] = ret[i].c1, ret[i].c2 | ||
| c1c2frs[2*i], c1c2frs[2*i+1] = new(Fr), new(Fr) | ||
| } | ||
|
|
||
| toFrMultiple(c1c2frs, c1c2points) | ||
|
|
||
| var poly [NodeWidth]Fr | ||
| poly[0].SetUint64(1) | ||
| for i, nv := range nodesValues { | ||
| StemFromBytes(&poly[1], nv.Stem) | ||
| poly[2] = *c1c2frs[2*i] | ||
| poly[3] = *c1c2frs[2*i+1] | ||
|
|
||
| ret[i].commitment = cfg.CommitToPoly(poly[:], 252) | ||
| } | ||
|
|
||
| sort.Slice(ret, func(i, j int) bool { | ||
| return bytes.Compare(ret[i].stem, ret[j].stem) < 0 | ||
| }) | ||
|
|
||
| return ret | ||
| } | ||
|
|
||
| // BatchInsertOrderedLeaves creates a tree under from an ordered and deduplicated list of leaves. | ||
| // There's weak assumption that each subtree of the first stem-byte has more than 1 leaf node. | ||
| // If the whole tree has more than 2000 leaves the chance of that not being true is 0.033~=0. | ||
| func BatchInsertOrderedLeaves(leaves []LeafNode) *InternalNode { | ||
| // currentBranch is a representation of the current branch we're in. | ||
| // The length of the branch is at most StemSize, and it might only | ||
| // have non-nil values in the first N levels. | ||
| var currentBranch [StemSize]*InternalNode | ||
|
|
||
| // Initial state is a branch with only a root node at the top, pointing to | ||
| // the first leaf. | ||
| currentBranch[1] = newInternalNode(1).(*InternalNode) | ||
| currentBranch[1].cowChild(leaves[0].stem[1]) | ||
| currentBranch[1].children[leaves[0].stem[1]] = &leaves[0] | ||
|
|
||
| currentBranch[0] = New().(*InternalNode) | ||
| currentBranch[0].cowChild(leaves[0].stem[0]) | ||
| currentBranch[0].children[leaves[0].stem[0]] = currentBranch[1] | ||
|
|
||
| prevLeaf := &leaves[0] | ||
| leaves = leaves[1:] | ||
| // The idea is that we compare the newLeaf with the previousLeaf, and | ||
| // depending on how their stems differ, we adjust our currentBranch structure. | ||
| for i := range leaves { | ||
| newLeaf := &leaves[i] | ||
|
|
||
| // We get the first index in their stems that is different. | ||
| idx := firstDiffByteIdx(prevLeaf.stem, newLeaf.stem) | ||
|
|
||
| // If the currentBranch has a node at that index, we simply set the children | ||
| // to the newLeaf. | ||
| if currentBranch[idx] != nil { | ||
| currentBranch[idx].cowChild(newLeaf.stem[idx]) | ||
| currentBranch[idx].children[newLeaf.stem[idx]] = newLeaf | ||
| newLeaf.setDepth(currentBranch[idx].depth + 1) | ||
| for i := idx + 1; i < len(currentBranch); i++ { | ||
| currentBranch[i] = nil | ||
| } | ||
| } else { | ||
| // In this case there's no InternalNode in the current branch at the index. | ||
| // We need to "fill the gap" between the previous non-nil internal node up to | ||
| // the idx with new internal nodes. Then we set the last created internal node | ||
| // to the previous and new leaf. | ||
| prevNonNilIdx := 0 | ||
| for i := idx - 1; i >= 0; i-- { | ||
| if currentBranch[i] != nil { | ||
| prevNonNilIdx = i | ||
| break | ||
| } | ||
| } | ||
| for k := prevNonNilIdx + 1; k <= idx; k++ { | ||
| currentBranch[k] = newInternalNode(currentBranch[k-1].depth + 1).(*InternalNode) | ||
| currentBranch[k-1].cowChild(newLeaf.stem[k-1]) | ||
| currentBranch[k-1].children[newLeaf.stem[k-1]] = currentBranch[k] | ||
| } | ||
|
|
||
| currentBranch[idx].cowChild(prevLeaf.stem[idx]) | ||
| currentBranch[idx].children[prevLeaf.stem[idx]] = prevLeaf | ||
| prevLeaf.setDepth(currentBranch[idx].depth + 1) | ||
| currentBranch[idx].cowChild(newLeaf.stem[idx]) | ||
| currentBranch[idx].children[newLeaf.stem[idx]] = newLeaf | ||
|
|
||
| for i := idx + 1; i < len(currentBranch); i++ { | ||
| currentBranch[i] = nil | ||
| } | ||
| } | ||
|
|
||
| prevLeaf = newLeaf | ||
| } | ||
|
|
||
| return currentBranch[0] | ||
| } | ||
|
|
||
| // firstDiffByteIdx will return the first index in which the two stems differ. | ||
| // Both stems *must* be different. | ||
| func firstDiffByteIdx(stem1 []byte, stem2 []byte) int { | ||
| for i := range stem1 { | ||
| if stem1[i] != stem2[i] { | ||
| return i | ||
| } | ||
| } | ||
| panic("stems are equal") | ||
| } | ||
|
|
||
| // GetInternalNodeCommitment returns the commitment of the internal node at | ||
| // the partialStem. e.g: if partialStem is [a, b] it will walk to the a-th | ||
| // children of the node, and then to the b-th children of that node, returning | ||
| // its commitment.. | ||
| func GetInternalNodeCommitment(node *InternalNode, partialStem []byte) *Point { | ||
| for i := range partialStem { | ||
| nextNode, ok := node.children[partialStem[i]].(*InternalNode) | ||
| if !ok { | ||
| return node.children[partialStem[i]].(*LeafNode).commitment | ||
| } | ||
| node = nextNode | ||
| } | ||
|
|
||
| return node.commitment | ||
| } | ||
|
|
||
| // BuildFirstTwoLayers builds the first two layers of the tree from all the precalculated | ||
| // commitments of the children of the second level. This method is generally used if tree | ||
| // construction was done in partitions, and you want to glue them together without having | ||
| // the whole tree in memory. | ||
| func BuildFirstTwoLayers(commitments [NodeWidth][NodeWidth][32]byte) *InternalNode { | ||
| var secondLevelInternalNodes [NodeWidth]*InternalNode | ||
|
|
||
| for stemFirstByte := range commitments { | ||
| for stemSecondByte := range commitments[stemFirstByte] { | ||
| if commitments[stemFirstByte][stemSecondByte] == [32]byte{} { | ||
| continue | ||
| } | ||
| if secondLevelInternalNodes[stemFirstByte] == nil { | ||
| secondLevelInternalNodes[stemFirstByte] = newInternalNode(1).(*InternalNode) | ||
| } | ||
| hashedNode := HashedNode{commitment: commitments[stemFirstByte][stemSecondByte][:]} | ||
| secondLevelInternalNodes[stemFirstByte].cowChild(byte(stemSecondByte)) | ||
| secondLevelInternalNodes[stemFirstByte].SetChild(stemSecondByte, &hashedNode) | ||
| } | ||
| } | ||
|
|
||
| root := newInternalNode(0).(*InternalNode) | ||
| for i, node := range secondLevelInternalNodes { | ||
| if node == nil { | ||
| continue | ||
| } | ||
| root.cowChild(byte(i)) | ||
| root.SetChild(i, node) | ||
| } | ||
|
|
||
| return root | ||
| } | ||
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Now that we merged the other emptyCode cached PR, we can exploit the same here. Maybe I can extract this code section to be shared between both.
For now, it won't make a big difference in the conversion since we're bottleneck by compactions.
I prefer to merge this PR as is with our correctness guarantees, and I'll do a PR right after with only that refactor.