Add Generate method in merkletree

A method is added to generate a merkle tree for data, and store the
generated tree in the output.

PiperOrigin-RevId: 315966571

2 files changed, 124 insertions, 0 deletions

diff --git a/pkg/merkletree/merkletree.go b/pkg/merkletree/merkletree.go
index 965f3670b..906f67943 100644
--- a/pkg/merkletree/merkletree.go
+++ b/pkg/merkletree/merkletree.go
@@ -16,6 +16,9 @@
 package merkletree
 
 import (
+	"crypto/sha256"
+	"io"
+
 	"gvisor.dev/gvisor/pkg/usermem"
 )
 
@@ -69,3 +72,64 @@ func MakeSize(dataSize int64) Size {
 	size.levelStart = append(size.levelStart, offset)
 	return size
 }
+
+// Generate constructs a Merkle tree for the contents of data. The output is
+// written to treeWriter. The treeReader should be able to read the tree after
+// it has been written. That is, treeWriter and treeReader should point to the
+// same underlying data but have separate cursors.
+func Generate(data io.Reader, dataSize int64, treeReader io.Reader, treeWriter io.Writer) ([]byte, error) {
+	size := MakeSize(dataSize)
+
+	numBlocks := (dataSize + size.blockSize - 1) / size.blockSize
+
+	var root []byte
+	for level := 0; level < len(size.levelStart); level++ {
+		for i := int64(0); i < numBlocks; i++ {
+			buf := make([]byte, size.blockSize)
+			var (
+				n   int
+				err error
+			)
+			if level == 0 {
+				// Read data block from the target file since level 0 is directly above
+				// the raw data block.
+				n, err = data.Read(buf)
+			} else {
+				// Read data block from the tree file since levels higher than 0 are
+				// hashing the lower level hashes.
+				n, err = treeReader.Read(buf)
+			}
+
+			// err is populated as long as the bytes read is smaller than the buffer
+			// size. This could be the case if we are reading the last block, and
+			// break in that case. If this is the last block, the end of the block
+			// will be zero-padded.
+			if n == 0 && err == io.EOF {
+				break
+			} else if err != nil && err != io.EOF {
+				return nil, err
+			}
+			// Hash the bytes in buf.
+			digest := sha256.Sum256(buf)
+
+			if level == len(size.levelStart)-1 {
+				root = digest[:]
+			}
+
+			// Write the generated hash to the end of the tree file.
+			if _, err = treeWriter.Write(digest[:]); err != nil {
+				return nil, err
+			}
+		}
+		// If the genereated digests do not round up to a block, zero-padding the
+		// remaining of the last block. But no need to do so for root.
+		if level != len(size.levelStart)-1 && numBlocks%size.hashesPerBlock != 0 {
+			zeroBuf := make([]byte, size.blockSize-(numBlocks%size.hashesPerBlock)*size.digestSize)
+			if _, err := treeWriter.Write(zeroBuf[:]); err != nil {
+				return nil, err
+			}
+		}
+		numBlocks = (numBlocks + size.hashesPerBlock - 1) / size.hashesPerBlock
+	}
+	return root, nil
+}
diff --git a/pkg/merkletree/merkletree_test.go b/pkg/merkletree/merkletree_test.go
index 6221eec07..7344db0b6 100644
--- a/pkg/merkletree/merkletree_test.go
+++ b/pkg/merkletree/merkletree_test.go
@@ -15,6 +15,7 @@
 package merkletree
 
 import (
+	"bytes"
 	"fmt"
 	"testing"
 
@@ -60,3 +61,62 @@ func TestSize(t *testing.T) {
 		})
 	}
 }
+
+func TestGenerate(t *testing.T) {
+	// The input data has size dataSize. It starts with the data in startWith,
+	// and all other bytes are zeroes.
+	testCases := []struct {
+		dataSize     int
+		startWith    []byte
+		expectedRoot []byte
+	}{
+		{
+			dataSize:     usermem.PageSize,
+			startWith:    nil,
+			expectedRoot: []byte{173, 127, 172, 178, 88, 111, 198, 233, 102, 192, 4, 215, 209, 209, 107, 2, 79, 88, 5, 255, 124, 180, 124, 122, 133, 218, 189, 139, 72, 137, 44, 167},
+		},
+		{
+			dataSize:     128*usermem.PageSize + 1,
+			startWith:    nil,
+			expectedRoot: []byte{62, 93, 40, 92, 161, 241, 30, 223, 202, 99, 39, 2, 132, 113, 240, 139, 117, 99, 79, 243, 54, 18, 100, 184, 141, 121, 238, 46, 149, 202, 203, 132},
+		},
+		{
+			dataSize:     1,
+			startWith:    []byte{'a'},
+			expectedRoot: []byte{52, 75, 204, 142, 172, 129, 37, 14, 145, 137, 103, 203, 11, 162, 209, 205, 30, 169, 213, 72, 20, 28, 243, 24, 242, 2, 92, 43, 169, 59, 110, 210},
+		},
+		{
+			dataSize:     1,
+			startWith:    []byte{'1'},
+			expectedRoot: []byte{74, 35, 103, 179, 176, 149, 254, 112, 42, 65, 104, 66, 119, 56, 133, 124, 228, 15, 65, 161, 150, 0, 117, 174, 242, 34, 115, 115, 218, 37, 3, 105},
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(fmt.Sprintf("%d", tc.dataSize), func(t *testing.T) {
+			var (
+				data bytes.Buffer
+				tree bytes.Buffer
+			)
+
+			startSize := len(tc.startWith)
+			_, err := data.Write(tc.startWith)
+			if err != nil {
+				t.Fatalf("Failed to write to data: %v", err)
+			}
+			_, err = data.Write(make([]byte, tc.dataSize-startSize))
+			if err != nil {
+				t.Fatalf("Failed to write to data: %v", err)
+			}
+
+			root, err := Generate(&data, int64(tc.dataSize), &tree, &tree)
+			if err != nil {
+				t.Fatalf("Generate failed: %v", err)
+			}
+
+			if !bytes.Equal(root, tc.expectedRoot) {
+				t.Errorf("Unexpected root")
+			}
+		})
+	}
+}