CVE-2023-46132: Github.com/hyperledger/fabric Fabric Vulnerable To Crosslinking Transaction Vulnerability

HIGH
Advisory Identifier: GHSA-v9w2-543f-h69m
Share
CVSS 7.1
CWE-362
Published
Advisory Summary

CVE-2023-46132 is a HIGH (CVSS 7.1) fabric vulnerable to crosslinking transaction vulnerability (CWE-362: Concurrent Execution using Shared Resource with Improper Synchronization ('Race Condition')) affecting Github.com/hyperledger/fabric libraries in the Go ecosystem. The issue involves fabric vulnerable to crosslinking transaction attack. Remediation is available by upgrading affected packages to version 2.2.14, 2.5.5.

CVE-2023-46132 at a Glance

CVE Identifier CVE-2023-46132
GHSA Identifier GHSA-v9w2-543f-h69m
Severity HIGH
CVSS v3.1 Base Score 7.1
CWE Classification CWE-362: Concurrent Execution using Shared Resource with Improper Synchronization ('Race Condition')
Affected Product Github.com/hyperledger/fabric
Package Ecosystem Go
Published Date
Last Updated

Security Impact

CVSS: 7.1

The security impact metrics below are parsed directly from the CVSS v3.1 vector string (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:H/A:L).

Attack Vector Network (AV:N)
Attack Complexity Low (AC:L)
Privileges Required Low (PR:L)
User Interaction None (UI:N)
Scope Unchanged (S:U)
Confidentiality None (C:N)
Integrity High (I:H)
Availability Low (A:L)

How to Fix CVE-2023-46132

To remediate this vulnerability, upgrade all affected Github.com/hyperledger/fabric dependencies to a patched release. Both direct and transitive dependencies should be checked, as frameworks or security libraries may pull in vulnerable versions transitively.

github.com/hyperledger/fabric fixed: 2.2.14 
Upgrade github.com/hyperledger/fabric to 2.2.14
github.com/hyperledger/fabric fixed: 2.5.5 
Upgrade github.com/hyperledger/fabric to 2.5.5
Verify Direct and Transitive Dependencies

Run the following command in your project directory to verify that all resolved dependencies use the patched release: 

inspect dependencies for github.com/hyperledger/fabric

How to Check if You Are Affected

To determine whether your application is affected by CVE-2023-46132, inspect your project dependencies for Github.com/hyperledger/fabric:

inspect dependencies for github.com/hyperledger/fabric

Analyze the terminal output to locate any version of Github.com/hyperledger/fabric in your dependency tree. If any resolved version matches the vulnerable ranges listed in the table below (for example, versions earlier than 2.2.14), your application is affected.

Note: Verify declared build manifests and deployed runtime container environments.

Technical Details

Vulnerability Mechanism & Technical Analysis

This vulnerability is classified under CWE-362: Concurrent Execution using Shared Resource with Improper Synchronization ('Race Condition'), where the product contains a concurrent code sequence that requires temporary, exclusive access to a shared resource, but a timing window exists in which the shared resource can be modified by another code sequence operating concurrently. In Github.com/hyperledger/fabric, software state validation or input boundaries are circumvented during processing. Review the official source advisory statement below for specific affected components, classes, and reproduction details.

Source Advisory Statement

# Short summary Combining two molecules to one another, called "cross-linking" results in a molecule with a chemical formula that is composed of all atoms of the original two molecules. In Fabric, one can take a block of transactions and cross-link the transactions in a way that alters the way the peers parse the transactions. If a first peer receives a block `B` and a second peer receives a block identical to `B` but with the transactions being cross-linked, the second peer will parse transactions in a different way and thus its world state will deviate from the first peer. Orderers or peers cannot detect that a block has its transactions cross-linked, because there is a vulnerability in the way Fabric hashes the transactions of blocks. It simply and naively concatenates them, which is insecure and lets an adversary craft a "cross-linked block" (block with cross-linked transactions) which alters the way peers process transactions. For example, it is possible to select a transaction and manipulate a peer to completely avoid processing it, without changing the computed hash of the block. Additional validations have been added in v2.2.14 and v2.5.5 to detect potential cross-linking issues before processing blocks. ## Impact In V1 and V2, we only have a crash fault tolerant orderer and as such, the security model Fabric operates in is that the orderer is honest, but peers may be malicious. As such, a peer that replicates a block from a malicious peer can have a state fork. In V3 which we did not a release a GA yet (only a preview), we have a byzantine fault tolerant orderering service, so the security model that Fabric operates in such a case includes malicious orderers. If the orderer is malicious, it can cause state forks for peers, and can infect non-malicious orderers with cross-linked blocks. # Long summary In order to create a signature on a big chunk of data such as a block, the data needs to be "compressed" first to the input size of the signature algorithm. In Fabric's case, we use a hash function which compressed a Fabric block from arbitrary size to a 32 byte string. In order to understand the problem we need to be more specific: The block structure has three parts to it: (1) Header, (2) Transactions, and (3) Metadata. When hashing the block, the header and metadata are stitched together and then hashed, and this hash of the header and the metadata is what signed (it's a simplification but let's not get into details) However, the transactions of the block are not part of the above hash. Instead, the header contains a hash, called the "Data hash" and despite the fact that in the comments it is said: "// The hash of the BlockData, by MerkleTree", actually it is far from being the case, and that is where our problem lies. The problem is that the way the transactions are hashed gives an attacker some freedom in manipulating the data. To create the Data Hash, the transactions in the block are concatenated to one another, creating a big long byte array and then this big long byte array is hashed, and this is essentially the Data Hash. The transactions in the block are a list of raw byte arrays, and when they are concatenated they look like this: `|$$$$$$$$$$$$|*************|@@@@@@@@@@@@|%%%%%%%%%|` (The vertical lines " | " represent how transactions are separated in a block.) When the transactions are concatenated in order to be hashed, the payload that is hashed is: `$$$$$$$$$$$$*************@@@@@@@@@@@@%%%%%%%%%` An adversary can't change the bytes of the concatenation, however what it can do, is to modify how transactions are encoded in the block: For example, consider an adversary wants to manipulate a peer to skip the second transaction (******). It can then create a block with the transactions as follows: `|$$$$$$$$$$$$*************|@@@@@@@@@@@@|%%%%%%%%%| ` Notice that a block with the above transactions has the same concatenation of bytes as the original block, but the block has one less transaction - the first transaction is a concatenation of the first and second transactions in the original block. When the peer receives this block, it looks at the first transaction and when it parses it, it completely ignores the ***** bytes, (we will see why soon), and so, an adversary can create a block with the same hash but different transactions and this would create a fork in the network. I made a small PoC where I created a block with 2 transactions (by invoking two chaincodes at the same time) with a Raft orderer: ``` [e][OrdererOrg.orderer] 2023-10-14 23:07:34.076 CEST 0079 INFO [orderer.consensus.etcdraft] propose -> Created block [10] with 2 transactions, there are 0 blocks in flight channel=testchannel node=1 ``` But right after creating the block, I just modified only its transaction content (without modifying the block hash) and then the peers only detect a single transaction inside that block: ``` [e][Org2.peer0] 2023-10-14 23:07:34.079 CEST 0099 INFO [kvledger] commit -> [testchannel] Committed block [10] with 1 transaction(s) in 0ms (state_validation=0ms block_and_pvtdata_commit=0ms state_commit=0ms) commitHash=[c5ecca818da9319af2f276dd01cd1337938f20c3535dd23f95a33933a114fe84] ``` The important takeaway from this experiment is that the peer does not detect any tempering was done to the block. If an attacker performs this attack, the network can be forked silently and no one will notice the network was forked until it's too late. # Patches Here is the patch I propose (the explanation is further below): ``` diff --git a/internal/peer/gossip/mcs.go b/internal/peer/gossip/mcs.go index b46df8b6a..9c3b5c8fd 100644 --- a/internal/peer/gossip/mcs.go +++ b/internal/peer/gossip/mcs.go @@ -150,6 +150,10 @@ func (s *MSPMessageCryptoService) VerifyBlock(chainID common.ChannelID, seqNum u return fmt.Errorf("Block with id [%d] on channel [%s] does not have metadata. Block not valid.", block.Header.Number, chainID) } + if err := protoutil.VerifyTransactionsAreWellFormed(block); err != nil { + return err + } + // - Verify that Header.DataHash is equal to the hash of block.Data // This is to ensure that the header is consistent with the data carried by this block if !bytes.Equal(protoutil.BlockDataHash(block.Data), block.Header.DataHash) { diff --git a/orderer/common/cluster/util.go b/orderer/common/cluster/util.go index e229bebfc..05b1bfaa9 100644 --- a/orderer/common/cluster/util.go +++ b/orderer/common/cluster/util.go @@ -260,6 +260,9 @@ func VerifyBlockHash(indexInBuffer int, blockBuff []*common.Block) error { if block.Header == nil { return errors.New("missing block header") } + if err := protoutil.VerifyTransactionsAreWellFormed(block); err != nil { + return err + } seq := block.Header.Number dataHash := protoutil.BlockDataHash(block.Data) // Verify data hash matches the hash in the header diff --git a/orderer/consensus/smartbft/verifier.go b/orderer/consensus/smartbft/verifier.go index 2b9fdfc4c..f232a1eae 100644 --- a/orderer/consensus/smartbft/verifier.go +++ b/orderer/consensus/smartbft/verifier.go @@ -237,6 +237,10 @@ func verifyHashChain(block *cb.Block, prevHeaderHash string) error { return errors.Errorf("previous header hash is %s but expected %s", thisHdrHashOfPrevHdr, prevHeaderHash) } + if err := protoutil.VerifyTransactionsAreWellFormed(block); err != nil { + return err + } + dataHash := hex.EncodeToString(block.Header.DataHash) actualHashOfData := hex.EncodeToString(protoutil.BlockDataHash(block.Data)) if dataHash != actualHashOfData { diff --git a/protoutil/blockutils.go b/protoutil/blockutils.go index 8527869e4..fca3c386f 100644 --- a/protoutil/blockutils.go +++ b/protoutil/blockutils.go @@ -10,6 +10,7 @@ import ( "bytes" "crypto/sha256" "encoding/asn1" + "encoding/base64" "fmt" "math/big" @@ -298,3 +299,35 @@ func searchConsenterIdentityByID(consenters []*cb.Consenter, identifier uint32) } return nil } + +func VerifyTransactionsAreWellFormed(block *cb.Block) error { + if block == nil || block.Data == nil || len(block.Data.Data) == 0 { + return nil + } + + for i, rawTx := range block.Data.Data { + env := &cb.Envelope{} + if err := proto.Unmarshal(rawTx, env); err != nil { + return fmt.Errorf("transaction %d is invalid: %v", i, err) + } + + if len(env.Payload) == 0 { + return fmt.Errorf("transaction %d has no payload", i) + } + + if len(env.Signature) == 0 { + return fmt.Errorf("transaction %d has no signature", i) + } + + expected := MarshalOrPanic(env) + if len(expected) < len(rawTx) { + return fmt.Errorf("transaction %d has %d trailing bytes", i, len(rawTx)-len(expected)) + } + if !bytes.Equal(expected, rawTx) { + return fmt.Errorf("transaction %d (%s) does not match its raw form (%s)", i, + base64.StdEncoding.EncodeToString(expected), base64.StdEncoding.EncodeToString(rawTx)) + } + } + + return nil +} diff --git a/protoutil/blockutils_test.go b/protoutil/blockutils_test.go index b2159da9f..2871483f1 100644 --- a/protoutil/blockutils_test.go +++ b/protoutil/blockutils_test.go @@ -489,3 +489,109 @@ func TestBlockSignatureVerifierByCreator(t *testing.T) { require.Len(t, signatureSet, 1) require.Equal(t, []byte("creator1"), signatureSet[0].Identity) } + +func TestVerifyTransactionsAreWellFormed(t *testing.T) { + originalBlock := &cb.Block{ + Data: &cb.BlockData{ + Data: [][]byte{ + marshalOrPanic(&cb.Envelope{ + Payload: []byte{1, 2, 3}, + Signature: []byte{4, 5, 6}, + }), + marshalOrPanic(&cb.Envelope{ + Payload: []byte{7, 8, 9}, + Signature: []byte{10, 11, 12}, + }), + }, + }, + } + + forgedBlock := proto.Clone(originalBlock).(*cb.Block) + tmp := make([]byte, len(forgedBlock.Data.Data[0])+len(forgedBlock.Data.Data[1])) + copy(tmp, forgedBlock.Data.Data[0]) + copy(tmp[len(forgedBlock.Data.Data[0]):], forgedBlock.Data.Data[1]) + forgedBlock.Data.Data = [][]byte{tmp} // Replace transactions {0,1} with transaction {0 || 1} + + for _, tst := range []struct { + name string + expectedError string + block *cb.Block + }{ + { + name: "empty block", + }, + { + name: "no block data", + block: &cb.Block{}, + }, + { + name: "no transactions", + block: &cb.Block{Data: &cb.BlockData{}}, + }, + { + name: "single transaction", + block: &cb.Block{Data: &cb.BlockData{Data: [][]byte{marshalOrPanic(&cb.Envelope{ + Payload: []byte{1, 2, 3}, + Signature: []byte{4, 5, 6}, + })}}}, + }, + { + name: "good block", + block: originalBlock, + }, + { + name: "forged block", + block: forgedBlock, + expectedError: "transaction 0 has 10 trailing bytes", + }, + { + name: "no signature", + expectedError: "transaction 0 has no signature", + block: &cb.Block{ + Data: &cb.BlockData{ + Data: [][]byte{ + marshalOrPanic(&cb.Envelope{ + Payload: []byte{1, 2, 3}, + }), + }, + }, + }, + }, + { + name: "no payload", + expectedError: "transaction 0 has no payload", + block: &cb.Block{ + Data: &cb.BlockData{ + Data: [][]byte{ + marshalOrPanic(&cb.Envelope{ + Signature: []byte{4, 5, 6}, + }), + }, + }, + }, + }, + { + name: "transaction invalid", + expectedError: "transaction 0 is invalid: proto: cannot parse invalid wire-format data", + block: &cb.Block{ + Data: &cb.BlockData{ + Data: [][]byte{ + marshalOrPanic(&cb.Envelope{ + Payload: []byte{1, 2, 3}, + Signature: []byte{4, 5, 6}, + })[9:], + }, + }, + }, + }, + } { + t.Run(tst.name, func(t *testing.T) { + err := protoutil.VerifyTransactionsAreWellFormed(tst.block) + if tst.expectedError == "" { + require.NoError(t, err) + } else { + require.EqualError(t, err, tst.expectedError) + } + }) + } +} ``` The idea is as follows: When we validate that the block's transactions match the hash in the header, we not only hash the transactions are earlier, but also ensure that if the transactions in the block are encoded into bytes, they re-create the exact split in the original block: `|$$$$$$$$$$$$|***********|@@@@@@@@@|%%%%%%%%%%%|` More specifically, each transaction in the block is parsed and then re-encoded to bytes, and we check that the original encoding of a transaction is as the second encoding after parsing the original bytes of the transaction. This fix keeps the legacy way of hashing transactions to create the block data hash, but also aims to check if some manipulation was done. To understand why the fix works, we need to understand how protobuf, the wire protocol that Fabric uses to encode transactions (and almost anything it sends over the wire or writes to disk) encodes a transaction. A transaction is a protobuf message with two fields of bytes: (1) Payload and (2) Signature. When encoding a field of bytes, protobuf first writes a tag for the field (a byte) and then writes the length of the field in variable-length encoding, and then the bytes themselves. For example, to encode a transaction, protobuf writes 10 (the tag for payload), then two bytes for the length of the payload, then the payload, and then 18, the tag for the signature, and then a single byte for the length of the signature, and finally the signature. Now, we can understand a proof sketch of why my solution works: Assume in contradiction that an adversary takes a block of transactions and changes the split of the concatenation in a way that changes the transactions for a peer: From `|$$$$$$$$$$$$|************|@@@@@@@@@@@|...|%%%%%%%|` to (for example): From `|$$$$$$$$$$$$************|@@@@@@@@@@@|...|%%%%%%%|` Since this split is not identical to the original split, there exists at least one transaction index of different size between the two splits. Let's look at the first transaction that is of different size. For example, for the split: `|$$$$$$$$$$$$|************|@@@@@@@@@@@|...|%%%%%%%|` we have two options: 1. The first transaction of different size is smaller in the new split: `|$$$$$$$$$$$$|*****|*******|@@@@@@@@@@@|...|%%%%%%%|` In such a case, it must contain both a payload and a signature, so it needs two fields (we can say we will return an error if one of the two is missing). If the protobuf parser detects it lacks bytes to parse a payload, it will fail with an error. Else, it has enough bytes to parse the payload, and then the signature is parsed. If the signature field is too short then we also error similarly. 2. The first transaction of different size is bigger in the new split: `|$$$$$$$$$$$$|************@@@@|@@@@@@@|...|%%%%%%%|` In that case, once this transaction is parsed, the extra bytes are skipped, so encoding the transaction to bytes yields a shorter byte array, and we detect a tempering.

CVE-2023-46132 Affected Github.com/hyperledger/fabric Versions

2 component(s)
Fixed Versions:
2.2.142.5.5
Product / Package Ecosystem Affected Versions Fixed Version
github.com/hyperledger/fabric
golang/github.com/hyperledger/fabric
Go
≥ 1.0.0-alpha
< 2.2.14
2.2.14
github.com/hyperledger/fabric
golang/github.com/hyperledger/fabric
Go
≥ 2.3.0
< 2.5.5
2.5.5

References

GitHub Advisory / Wiki ADVISORY
https://github.com/hyperledger/fabric/security/advisories/GHSA-v9w2-543f-h69m
NVD — CVE-2023-46132 NVD
https://nvd.nist.gov/vuln/detail/CVE-2023-46132
Github.com/hyperledger/fabric Source Repository REPOSITORY
https://github.com/hyperledger/fabric/pull/4503
Github.com/hyperledger/fabric Source Repository REPOSITORY
https://github.com/hyperledger/fabric/pull/4504
Security Fix Commit FIX COMMIT
https://github.com/hyperledger/fabric/commit/389b2e66de9a6fbc6043216d554c97bbbdf0e008
Security Fix Commit FIX COMMIT
https://github.com/hyperledger/fabric/commit/93bef10bd3ce3c54d7f3b064f765dbde61da7def
Github.com/hyperledger/fabric Source Repository REPOSITORY
https://github.com/hyperledger/fabric
Github.com/hyperledger/fabric Release RELEASE
https://github.com/hyperledger/fabric/releases/tag/v2.2.14
Github.com/hyperledger/fabric Release RELEASE
https://github.com/hyperledger/fabric/releases/tag/v2.5.5

Frequently Asked Questions

What is CVE-2023-46132?
CVE-2023-46132 is a security vulnerability in Github.com/hyperledger/fabric involving Fabric vulnerable to crosslinking transaction attack under CWE-362: Concurrent Execution using Shared Resource with Improper Synchronization ('Race Condition').
Which Github.com/hyperledger/fabric versions are affected?
Affected versions include releases prior to 2.2.14, 2.5.5 across affected packages: github.com/hyperledger/fabric.
What version fixes CVE-2023-46132?
Upgrading to 2.2.14, 2.5.5 (or the latest patched release for your respective branch) resolves the vulnerability.
Is github.com/hyperledger/fabric affected?
Yes, github.com/hyperledger/fabric is affected in releases prior to 2.2.14, 2.5.5. Upgrade your dependency to 2.2.14, 2.5.5 or later.
What is the CVSS score of CVE-2023-46132?
CVE-2023-46132 has a CVSS v3.1 base score of 7.1 (HIGH) with vector string CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:H/A:L.
What CWE is associated with CVE-2023-46132?
CVE-2023-46132 is classified under CWE-362: Concurrent Execution using Shared Resource with Improper Synchronization ('Race Condition') (The product contains a concurrent code sequence that requires temporary, exclusive access to a shared resource, but a timing window exists in which the shared resource can be modified by another code sequence operating concurrently.).
How can I check Go dependencies for CVE-2023-46132?
Run the dependency tree command: 'inspect dependencies for github.com/hyperledger/fabric'. This command lists direct and transitive dependencies in your build configuration so you can verify if a vulnerable version is resolved.
Machine-readable OSV JSON OSV v1.5 Schema
Normalized schema output
View Raw JSON 
{
  "schema_version" : "1.4.0",
  "id" : "GHSA-v9w2-543f-h69m",
  "modified" : "2023-11-14T21:37:09Z",
  "published" : "2023-11-14T20:28:34Z",
  "aliases" : [ "CVE-2023-46132" ],
  "summary" : "Fabric vulnerable to crosslinking transaction attack",
  "details" : "# Short summary\n\nCombining two molecules to one another, called \"cross-linking\" results in a molecule with a chemical formula that is composed of all atoms of the original two molecules. \n\nIn Fabric, one can take a block of transactions and cross-link the transactions in a way that alters the way the peers parse the transactions. If a first peer receives a block `B` and a second peer receives a block identical to `B` but with the transactions being cross-linked, the second peer will parse transactions in a different way and thus its world state will deviate from the first peer. \n\nOrderers or peers cannot detect that a block has its transactions cross-linked, because there is a vulnerability in the way Fabric hashes the transactions of blocks. It simply and naively concatenates them, which is insecure and lets an adversary craft a \"cross-linked block\" (block with cross-linked transactions) which alters the way peers process transactions. \nFor example, it is possible to select a transaction and manipulate a peer to completely avoid processing it, without changing the computed hash of the block.\n\nAdditional validations have been added in v2.2.14 and v2.5.5 to detect potential cross-linking issues before processing blocks.\n\n## Impact\nIn V1 and V2, we only have a crash fault tolerant orderer and as such, the security model Fabric operates in is that the orderer is honest,\nbut peers may be malicious. As such, a peer that replicates a block from a malicious peer can have a state fork.\n\nIn V3 which we did not a release a GA yet (only a preview), we have a byzantine fault tolerant orderering service, so the security model that Fabric operates in such a case includes malicious orderers. If the orderer is malicious, it can cause state forks for peers, and can infect non-malicious orderers with cross-linked blocks.\n\n# Long summary\n\nIn order to create a signature on a big chunk of data  such as a block, the data needs to be \"compressed\" first to the input size of the signature algorithm.\n\nIn Fabric's case, we use a hash function which compressed a Fabric block from arbitrary size to a 32 byte string.\n\nIn order to understand the problem we need to be more specific: The block structure has three parts to it: (1) Header, (2) Transactions, and (3) Metadata.\n\nWhen hashing the block, the header and metadata are stitched together and then hashed, and this hash of the header and the metadata is what signed (it's a simplification but let's not get into details)\n\nHowever, the transactions of the block are not part of the above hash. Instead, the header contains a hash, called the \"Data hash\" and despite the fact that in the comments it is said: \"// The hash of the BlockData, by MerkleTree\", actually it is far from being the case, and that is where our problem lies.\n\nThe problem is that the way the transactions are hashed gives an attacker some freedom in manipulating the data. \n\nTo create the Data Hash, the transactions in the block are concatenated to one another, creating a big long byte array and then this big long byte array is hashed, and this is essentially the Data Hash.\n\nThe transactions in the block are a list of raw byte arrays, and when they are concatenated they look like this:\n\n`|$$$$$$$$$$$$|*************|@@@@@@@@@@@@|%%%%%%%%%|`  (The vertical lines \" | \" represent how transactions are separated in a block.)\n\nWhen the transactions are concatenated in order to be hashed, the payload that is hashed is: \n`$$$$$$$$$$$$*************@@@@@@@@@@@@%%%%%%%%%`\n\nAn adversary can't change the bytes of the concatenation, however what it can do, is to modify how transactions are encoded in the block:\n\nFor example, consider an adversary wants to manipulate a peer to skip the second transaction (******).\n\nIt can then create a block with the transactions as follows:\n\n`|$$$$$$$$$$$$*************|@@@@@@@@@@@@|%%%%%%%%%| `\n\nNotice that a block with the above transactions has the same concatenation of bytes as the original block, but the block has one less transaction - the first transaction is a concatenation of the first and second transactions in the original block.\n\nWhen the peer receives this block, it looks at the first transaction and when it parses it, it completely ignores the ***** bytes, (we will see why soon), and so, an adversary can create a block with the same hash but different transactions and this would create a fork in the network.\n \nI made a small PoC where I created a block with 2 transactions (by invoking two chaincodes at the same time) with a Raft orderer:\n\n```\n    [e][OrdererOrg.orderer] 2023-10-14 23:07:34.076 CEST 0079 INFO [orderer.consensus.etcdraft] propose -> Created block [10] with 2 transactions, there are 0 blocks in flight channel=testchannel node=1\n```\n\nBut right after creating the block, I just modified only its transaction content (without modifying the block hash) and then the peers only detect a single transaction inside that block:\n \n```\n    [e][Org2.peer0] 2023-10-14 23:07:34.079 CEST 0099 INFO [kvledger] commit -> [testchannel] Committed block [10] with 1 transaction(s) in 0ms (state_validation=0ms block_and_pvtdata_commit=0ms state_commit=0ms) commitHash=[c5ecca818da9319af2f276dd01cd1337938f20c3535dd23f95a33933a114fe84]\n```\n\nThe important takeaway from this experiment is that the peer does not detect any tempering was done to the block. If an attacker performs this attack, the network can be forked silently and no one will notice the network was forked until it's too late.\n\n \n\n# Patches\nHere is the patch I propose (the explanation is further below): \n\n```\ndiff --git a/internal/peer/gossip/mcs.go b/internal/peer/gossip/mcs.go\nindex b46df8b6a..9c3b5c8fd 100644\n--- a/internal/peer/gossip/mcs.go\n+++ b/internal/peer/gossip/mcs.go\n@@ -150,6 +150,10 @@ func (s *MSPMessageCryptoService) VerifyBlock(chainID common.ChannelID, seqNum u\n \t\treturn fmt.Errorf(\"Block with id [%d] on channel [%s] does not have metadata. Block not valid.\", block.Header.Number, chainID)\n \t}\n \n+\tif err := protoutil.VerifyTransactionsAreWellFormed(block); err != nil {\n+\t\treturn err\n+\t}\n+\n \t// - Verify that Header.DataHash is equal to the hash of block.Data\n \t// This is to ensure that the header is consistent with the data carried by this block\n \tif !bytes.Equal(protoutil.BlockDataHash(block.Data), block.Header.DataHash) {\ndiff --git a/orderer/common/cluster/util.go b/orderer/common/cluster/util.go\nindex e229bebfc..05b1bfaa9 100644\n--- a/orderer/common/cluster/util.go\n+++ b/orderer/common/cluster/util.go\n@@ -260,6 +260,9 @@ func VerifyBlockHash(indexInBuffer int, blockBuff []*common.Block) error {\n \tif block.Header == nil {\n \t\treturn errors.New(\"missing block header\")\n \t}\n+\tif err := protoutil.VerifyTransactionsAreWellFormed(block); err != nil {\n+\t\treturn err\n+\t}\n \tseq := block.Header.Number\n \tdataHash := protoutil.BlockDataHash(block.Data)\n \t// Verify data hash matches the hash in the header\ndiff --git a/orderer/consensus/smartbft/verifier.go b/orderer/consensus/smartbft/verifier.go\nindex 2b9fdfc4c..f232a1eae 100644\n--- a/orderer/consensus/smartbft/verifier.go\n+++ b/orderer/consensus/smartbft/verifier.go\n@@ -237,6 +237,10 @@ func verifyHashChain(block *cb.Block, prevHeaderHash string) error {\n \t\treturn errors.Errorf(\"previous header hash is %s but expected %s\", thisHdrHashOfPrevHdr, prevHeaderHash)\n \t}\n \n+\tif err := protoutil.VerifyTransactionsAreWellFormed(block); err != nil {\n+\t\treturn err\n+\t}\n+\n \tdataHash := hex.EncodeToString(block.Header.DataHash)\n \tactualHashOfData := hex.EncodeToString(protoutil.BlockDataHash(block.Data))\n \tif dataHash != actualHashOfData {\ndiff --git a/protoutil/blockutils.go b/protoutil/blockutils.go\nindex 8527869e4..fca3c386f 100644\n--- a/protoutil/blockutils.go\n+++ b/protoutil/blockutils.go\n@@ -10,6 +10,7 @@ import (\n \t\"bytes\"\n \t\"crypto/sha256\"\n \t\"encoding/asn1\"\n+\t\"encoding/base64\"\n \t\"fmt\"\n \t\"math/big\"\n \n@@ -298,3 +299,35 @@ func searchConsenterIdentityByID(consenters []*cb.Consenter, identifier uint32)\n \t}\n \treturn nil\n }\n+\n+func VerifyTransactionsAreWellFormed(block *cb.Block) error {\n+\tif block == nil || block.Data == nil || len(block.Data.Data) == 0 {\n+\t\treturn nil\n+\t}\n+\n+\tfor i, rawTx := range block.Data.Data {\n+\t\tenv := &cb.Envelope{}\n+\t\tif err := proto.Unmarshal(rawTx, env); err != nil {\n+\t\t\treturn fmt.Errorf(\"transaction %d is invalid: %v\", i, err)\n+\t\t}\n+\n+\t\tif len(env.Payload) == 0 {\n+\t\t\treturn fmt.Errorf(\"transaction %d has no payload\", i)\n+\t\t}\n+\n+\t\tif len(env.Signature) == 0 {\n+\t\t\treturn fmt.Errorf(\"transaction %d has no signature\", i)\n+\t\t}\n+\n+\t\texpected := MarshalOrPanic(env)\n+\t\tif len(expected) < len(rawTx) {\n+\t\t\treturn fmt.Errorf(\"transaction %d has %d trailing bytes\", i, len(rawTx)-len(expected))\n+\t\t}\n+\t\tif !bytes.Equal(expected, rawTx) {\n+\t\t\treturn fmt.Errorf(\"transaction %d (%s) does not match its raw form (%s)\", i,\n+\t\t\t\tbase64.StdEncoding.EncodeToString(expected), base64.StdEncoding.EncodeToString(rawTx))\n+\t\t}\n+\t}\n+\n+\treturn nil\n+}\ndiff --git a/protoutil/blockutils_test.go b/protoutil/blockutils_test.go\nindex b2159da9f..2871483f1 100644\n--- a/protoutil/blockutils_test.go\n+++ b/protoutil/blockutils_test.go\n@@ -489,3 +489,109 @@ func TestBlockSignatureVerifierByCreator(t *testing.T) {\n \trequire.Len(t, signatureSet, 1)\n \trequire.Equal(t, []byte(\"creator1\"), signatureSet[0].Identity)\n }\n+\n+func TestVerifyTransactionsAreWellFormed(t *testing.T) {\n+\toriginalBlock := &cb.Block{\n+\t\tData: &cb.BlockData{\n+\t\t\tData: [][]byte{\n+\t\t\t\tmarshalOrPanic(&cb.Envelope{\n+\t\t\t\t\tPayload:   []byte{1, 2, 3},\n+\t\t\t\t\tSignature: []byte{4, 5, 6},\n+\t\t\t\t}),\n+\t\t\t\tmarshalOrPanic(&cb.Envelope{\n+\t\t\t\t\tPayload:   []byte{7, 8, 9},\n+\t\t\t\t\tSignature: []byte{10, 11, 12},\n+\t\t\t\t}),\n+\t\t\t},\n+\t\t},\n+\t}\n+\n+\tforgedBlock := proto.Clone(originalBlock).(*cb.Block)\n+\ttmp := make([]byte, len(forgedBlock.Data.Data[0])+len(forgedBlock.Data.Data[1]))\n+\tcopy(tmp, forgedBlock.Data.Data[0])\n+\tcopy(tmp[len(forgedBlock.Data.Data[0]):], forgedBlock.Data.Data[1])\n+\tforgedBlock.Data.Data = [][]byte{tmp} // Replace transactions {0,1} with transaction {0 || 1}\n+\n+\tfor _, tst := range []struct {\n+\t\tname          string\n+\t\texpectedError string\n+\t\tblock         *cb.Block\n+\t}{\n+\t\t{\n+\t\t\tname: \"empty block\",\n+\t\t},\n+\t\t{\n+\t\t\tname:  \"no block data\",\n+\t\t\tblock: &cb.Block{},\n+\t\t},\n+\t\t{\n+\t\t\tname:  \"no transactions\",\n+\t\t\tblock: &cb.Block{Data: &cb.BlockData{}},\n+\t\t},\n+\t\t{\n+\t\t\tname: \"single transaction\",\n+\t\t\tblock: &cb.Block{Data: &cb.BlockData{Data: [][]byte{marshalOrPanic(&cb.Envelope{\n+\t\t\t\tPayload:   []byte{1, 2, 3},\n+\t\t\t\tSignature: []byte{4, 5, 6},\n+\t\t\t})}}},\n+\t\t},\n+\t\t{\n+\t\t\tname:  \"good block\",\n+\t\t\tblock: originalBlock,\n+\t\t},\n+\t\t{\n+\t\t\tname:          \"forged block\",\n+\t\t\tblock:         forgedBlock,\n+\t\t\texpectedError: \"transaction 0 has 10 trailing bytes\",\n+\t\t},\n+\t\t{\n+\t\t\tname:          \"no signature\",\n+\t\t\texpectedError: \"transaction 0 has no signature\",\n+\t\t\tblock: &cb.Block{\n+\t\t\t\tData: &cb.BlockData{\n+\t\t\t\t\tData: [][]byte{\n+\t\t\t\t\t\tmarshalOrPanic(&cb.Envelope{\n+\t\t\t\t\t\t\tPayload: []byte{1, 2, 3},\n+\t\t\t\t\t\t}),\n+\t\t\t\t\t},\n+\t\t\t\t},\n+\t\t\t},\n+\t\t},\n+\t\t{\n+\t\t\tname:          \"no payload\",\n+\t\t\texpectedError: \"transaction 0 has no payload\",\n+\t\t\tblock: &cb.Block{\n+\t\t\t\tData: &cb.BlockData{\n+\t\t\t\t\tData: [][]byte{\n+\t\t\t\t\t\tmarshalOrPanic(&cb.Envelope{\n+\t\t\t\t\t\t\tSignature: []byte{4, 5, 6},\n+\t\t\t\t\t\t}),\n+\t\t\t\t\t},\n+\t\t\t\t},\n+\t\t\t},\n+\t\t},\n+\t\t{\n+\t\t\tname:          \"transaction invalid\",\n+\t\t\texpectedError: \"transaction 0 is invalid: proto: cannot parse invalid wire-format data\",\n+\t\t\tblock: &cb.Block{\n+\t\t\t\tData: &cb.BlockData{\n+\t\t\t\t\tData: [][]byte{\n+\t\t\t\t\t\tmarshalOrPanic(&cb.Envelope{\n+\t\t\t\t\t\t\tPayload:   []byte{1, 2, 3},\n+\t\t\t\t\t\t\tSignature: []byte{4, 5, 6},\n+\t\t\t\t\t\t})[9:],\n+\t\t\t\t\t},\n+\t\t\t\t},\n+\t\t\t},\n+\t\t},\n+\t} {\n+\t\tt.Run(tst.name, func(t *testing.T) {\n+\t\t\terr := protoutil.VerifyTransactionsAreWellFormed(tst.block)\n+\t\t\tif tst.expectedError == \"\" {\n+\t\t\t\trequire.NoError(t, err)\n+\t\t\t} else {\n+\t\t\t\trequire.EqualError(t, err, tst.expectedError)\n+\t\t\t}\n+\t\t})\n+\t}\n+}\n\n```\n\nThe idea is as follows:\n\nWhen we validate that the block's transactions match the hash in the header, we not only hash the transactions are earlier, \n\nbut also ensure that if the transactions in the block are encoded into bytes, they re-create the exact split in the original block: `|$$$$$$$$$$$$|***********|@@@@@@@@@|%%%%%%%%%%%|`\n\nMore specifically, each transaction in the block is parsed and then re-encoded to bytes, and we check that the original encoding of a transaction is as the second encoding after parsing the original bytes of the transaction.\n\nThis fix keeps the legacy way of hashing transactions to create the block data hash, but also aims to check if some manipulation was done.\n\n \nTo understand why the fix works, we need to understand how protobuf, the wire protocol that Fabric uses to encode transactions (and almost anything it sends over the wire or writes to disk) encodes a transaction.\n\nA transaction is a protobuf message with two fields of bytes: (1) Payload and (2) Signature.\n\nWhen encoding a field of bytes, protobuf first writes a tag for the field (a byte) and then writes the length of the field in variable-length encoding, and then the bytes themselves.\n\nFor example, to encode a transaction, protobuf writes 10 (the tag for payload), then two bytes for the length of the payload, then the payload, and then 18, the tag for the signature, and then a single byte for the length of the signature, and finally the signature.\n\nNow, we can understand a proof sketch of why my solution works:\n\nAssume in contradiction that an adversary takes a block of transactions and changes the split of the concatenation in a way that changes the transactions for a peer:\n\nFrom `|$$$$$$$$$$$$|************|@@@@@@@@@@@|...|%%%%%%%|` to (for example): From `|$$$$$$$$$$$$************|@@@@@@@@@@@|...|%%%%%%%|` \n\nSince this split is not identical to the original split, there exists at least one transaction index of different size between the two splits. Let's look at the first transaction that is of different size.\n\nFor example, for the split:\n\n`|$$$$$$$$$$$$|************|@@@@@@@@@@@|...|%%%%%%%|`  we have two options:\n\n1.  The first transaction of different size is smaller in the new split:  `|$$$$$$$$$$$$|*****|*******|@@@@@@@@@@@|...|%%%%%%%|`  In such a case, it must contain both a payload and a signature, so it needs two fields (we can say we will return an error if one of the two is missing). If the protobuf parser detects it lacks bytes to parse a payload, it will fail with an error. Else, it has enough bytes to parse the payload, and then the signature is parsed. If the signature field is too short then we also error similarly.\n\n2. The first transaction of different size is bigger in the new split: `|$$$$$$$$$$$$|************@@@@|@@@@@@@|...|%%%%%%%|` \nIn that case, once this transaction is parsed, the extra bytes are skipped, so encoding the transaction to bytes yields a shorter byte array, and we detect a tempering.\n\n",
  "severity" : [ {
    "type" : "CVSS_V3",
    "score" : "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:H/A:L"
  } ],
  "affected" : [ {
    "package" : {
      "ecosystem" : "Go",
      "name" : "github.com/hyperledger/fabric"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "1.0.0-alpha"
      }, {
        "fixed" : "2.2.14"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "Go",
      "name" : "github.com/hyperledger/fabric"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.3.0"
      }, {
        "fixed" : "2.5.5"
      } ]
    } ]
  } ],
  "references" : [ {
    "type" : "WEB",
    "url" : "https://github.com/hyperledger/fabric/security/advisories/GHSA-v9w2-543f-h69m"
  }, {
    "type" : "ADVISORY",
    "url" : "https://nvd.nist.gov/vuln/detail/CVE-2023-46132"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/hyperledger/fabric/pull/4503"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/hyperledger/fabric/pull/4504"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/hyperledger/fabric/commit/389b2e66de9a6fbc6043216d554c97bbbdf0e008"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/hyperledger/fabric/commit/93bef10bd3ce3c54d7f3b064f765dbde61da7def"
  }, {
    "type" : "PACKAGE",
    "url" : "https://github.com/hyperledger/fabric"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/hyperledger/fabric/releases/tag/v2.2.14"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/hyperledger/fabric/releases/tag/v2.5.5"
  } ],
  "database_specific" : {
    "cwe_ids" : [ "CWE-362" ],
    "severity" : "HIGH",
    "github_reviewed" : true,
    "github_reviewed_at" : "2023-11-14T20:28:34Z",
    "nvd_published_at" : "2023-11-14T21:15:11Z"
  }
}