Remove SSZ Union usage in BlockHeaderWithProof type (#3019)

* Remove SSZ Union usage in BlockHeaderWithProof type

Remove SSZ Union in BlockHeaderWithProof type by making the proof
an SSZ encoded ByteList. The right type for the proof can be
selected at the decoding step by first looking into the header
for the timestamp and selecting the right type based on the
hardfork the block is in.

* Add content db migration function to fcli_db tool

* Update test vector repo

fluffy/database/content_db_migrate_deprecated.nim

@@ -0,0 +1,269 @@
+# Fluffy
+# Copyright (c) 2025 Status Research & Development GmbH
+# Licensed and distributed under either of
+#   * MIT license (license terms in the root directory or at https://opensource.org/licenses/MIT).
+#   * Apache v2 license (license terms in the root directory or at https://www.apache.org/licenses/LICENSE-2.0).
+# at your option. This file may not be copied, modified, or distributed except according to those terms.
+
+{.push raises: [].}
+
+import
+  chronicles,
+  metrics,
+  stint,
+  results,
+  stew/ptrops,
+  sqlite3_abi,
+  eth/db/kvstore,
+  eth/db/kvstore_sqlite3,
+  eth/common/[headers_rlp, blocks_rlp, receipts_rlp, transactions_rlp],
+  ../network/history/history_content,
+  ../network/history/history_type_conversions,
+  ../network/state/state_utils,
+  ../network/state/state_content,
+  ../network/history/content/content_values_deprecated
+
+export kvstore_sqlite3
+
+type
+  ContentPair = tuple[contentKey: array[32, byte], contentItem: seq[byte]]
+
+  ContentDBDeprecated* = ref object
+    backend: SqStoreRef
+    kv: KvStoreRef
+    selectAllStmt: SqliteStmt[NoParams, ContentPair]
+    deleteBatchStmt: SqliteStmt[NoParams, void]
+    updateBatchStmt: SqliteStmt[NoParams, void]
+
+func isWithoutProofImpl(content: openArray[byte]): bool =
+  let headerWithProof = decodeSsz(content, BlockHeaderWithProofDeprecated).valueOr:
+    # Leave all other content as it is
+    return false
+
+  if headerWithProof.proof.proofType ==
+      BlockHeaderProofType.historicalHashesAccumulatorProof:
+    false
+  elif headerWithProof.proof.proofType == BlockHeaderProofType.none:
+    true
+  else:
+    false
+
+func isWithoutProof*(
+    ctx: SqliteContext, n: cint, v: SqliteValue
+) {.cdecl, gcsafe, raises: [].} =
+  doAssert(n == 1)
+
+  let
+    ptrs = makeUncheckedArray(v)
+    blob1Len = sqlite3_value_bytes(ptrs[][0])
+
+  if isWithoutProofImpl(makeOpenArray(sqlite3_value_blob(ptrs[][0]), byte, blob1Len)):
+    ctx.sqlite3_result_int(cint 1)
+  else:
+    ctx.sqlite3_result_int(cint 0)
+
+func isWithInvalidEncodingImpl(content: openArray[byte]): bool =
+  let headerWithProof = decodeSsz(content, BlockHeaderWithProofDeprecated).valueOr:
+    # Leave all other content as it is
+    return false
+
+  if headerWithProof.proof.proofType ==
+      BlockHeaderProofType.historicalHashesAccumulatorProof: true else: false
+
+func isWithInvalidEncoding*(
+    ctx: SqliteContext, n: cint, v: SqliteValue
+) {.cdecl, gcsafe, raises: [].} =
+  doAssert(n == 1)
+
+  let
+    ptrs = makeUncheckedArray(v)
+    blobLen = sqlite3_value_bytes(ptrs[][0])
+
+  if isWithInvalidEncodingImpl(
+    makeOpenArray(sqlite3_value_blob(ptrs[][0]), byte, blobLen)
+  ):
+    ctx.sqlite3_result_int(cint 1)
+  else:
+    ctx.sqlite3_result_int(cint 0)
+
+func adjustContentImpl(a: openArray[byte]): seq[byte] =
+  let headerWithProof = decodeSsz(a, BlockHeaderWithProofDeprecated).valueOr:
+    raiseAssert("Should not occur as decoding check is already done")
+
+  let accumulatorProof = headerWithProof.proof.historicalHashesAccumulatorProof
+  let adjustedContent = BlockHeaderWithProof(
+    header: headerWithProof.header,
+    proof: ByteList[MAX_HEADER_PROOF_LENGTH].init(SSZ.encode(accumulatorProof)),
+  )
+
+  SSZ.encode(adjustedContent)
+
+func adjustContent*(
+    ctx: SqliteContext, n: cint, v: SqliteValue
+) {.cdecl, gcsafe, raises: [].} =
+  doAssert(n == 1)
+
+  let
+    ptrs = makeUncheckedArray(v)
+    blobLen = sqlite3_value_bytes(ptrs[][0])
+
+    bytes =
+      adjustContentImpl(makeOpenArray(sqlite3_value_blob(ptrs[][0]), byte, blobLen))
+
+  sqlite3_result_blob(ctx, baseAddr bytes, cint bytes.len, SQLITE_TRANSIENT)
+
+template expectDb(x: auto): untyped =
+  # There's no meaningful error handling implemented for a corrupt database or
+  # full disk - this requires manual intervention, so we'll panic for now
+  x.expect("working database (disk broken/full?)")
+
+## Public calls to get database size, content size and similar.
+
+proc new*(
+    T: type ContentDBDeprecated,
+    path: string,
+    inMemory = false,
+    manualCheckpoint = false,
+): ContentDBDeprecated =
+  let db =
+    if inMemory:
+      SqStoreRef.init("", "fluffy-test", inMemory = true).expect(
+        "working database (out of memory?)"
+      )
+    else:
+      SqStoreRef.init(path, "fluffy", manualCheckpoint = false).expectDb()
+
+  db.createCustomFunction("isWithoutProof", 1, isWithoutProof).expect(
+    "Custom function isWithoutProof creation OK"
+  )
+
+  db.createCustomFunction("isWithInvalidEncoding", 1, isWithInvalidEncoding).expect(
+    "Custom function isWithInvalidEncoding creation OK"
+  )
+
+  db.createCustomFunction("adjustContent", 1, adjustContent).expect(
+    "Custom function adjustContent creation OK"
+  )
+
+  let selectAllStmt =
+    db.prepareStmt("SELECT key, value FROM kvstore", NoParams, ContentPair)[]
+
+  let deleteBatchStmt = db.prepareStmt(
+    "DELETE FROM kvstore WHERE key IN (SELECT key FROM kvstore WHERE isWithoutProof(value) == 1)",
+    NoParams, void,
+  )[]
+
+  let updateBatchStmt = db.prepareStmt(
+    "UPDATE kvstore SET value = adjustContent(value) WHERE key IN (SELECT key FROM kvstore WHERE isWithInvalidEncoding(value) == 1)",
+    NoParams, void,
+  )[]
+
+  let kvStore = kvStore db.openKvStore().expectDb()
+
+  let contentDb = ContentDBDeprecated(
+    kv: kvStore,
+    backend: db,
+    selectAllStmt: selectAllStmt,
+    deleteBatchStmt: deleteBatchStmt,
+    updateBatchStmt: updateBatchStmt,
+  )
+
+  contentDb
+
+template disposeSafe(s: untyped): untyped =
+  if distinctBase(s) != nil:
+    s.dispose()
+    s = typeof(s)(nil)
+
+proc close*(db: ContentDBDeprecated) =
+  db.selectAllStmt.disposeSafe()
+  db.deleteBatchStmt.disposeSafe()
+  db.updateBatchStmt.disposeSafe()
+  discard db.kv.close()
+
+proc deleteAllHeadersWithoutProof*(db: ContentDBDeprecated) =
+  notice "ContentDB migration: deleting all headers without proof"
+  db.deleteBatchStmt.exec().expectDb()
+  notice "ContentDB migration done"
+
+proc updateAllHeadersWithInvalidEncoding*(db: ContentDBDeprecated) =
+  notice "ContentDB migration: updating all headers with invalid encoding"
+  db.updateBatchStmt.exec().expectDb()
+  notice "ContentDB migration done"
+
+proc iterateAllAndCountTypes*(db: ContentDBDeprecated) =
+  ## Ugly debugging call to print out count of content types in case of issues.
+  var
+    contentPair: ContentPair
+    contentTotal = 0
+    contentOldHeaders = 0
+    contentNewHeaders = 0
+    contentBodies = 0
+    contentReceipts = 0
+    contentAccount = 0
+    contentContract = 0
+    contentCode = 0
+    contentOther = 0
+
+  notice "ContentDB type count: iterating over all content"
+  for e in db.selectAllStmt.exec(contentPair):
+    contentTotal.inc()
+    block:
+      let res = decodeSsz(contentPair.contentItem, BlockHeaderWithProofDeprecated)
+      if res.isOk():
+        if decodeRlp(res.value().header.asSeq(), Header).isOk():
+          contentOldHeaders.inc()
+          continue
+    block:
+      let res = decodeSsz(contentPair.contentItem, BlockHeaderWithProof)
+      if res.isOk():
+        if decodeRlp(res.value().header.asSeq(), Header).isOk():
+          contentNewHeaders.inc()
+          continue
+
+    block:
+      let res = decodeSsz(contentPair.contentItem, PortalReceipts)
+      if res.isOk():
+        if fromPortalReceipts(seq[Receipt], res.value()).isOk():
+          contentReceipts.inc()
+          continue
+
+    block:
+      let res = decodeSsz(contentPair.contentItem, PortalBlockBodyShanghai)
+      if res.isOk():
+        if fromPortalBlockBody(BlockBody, res.value()).isOk():
+          contentBodies.inc()
+          continue
+
+    block:
+      let res = decodeSsz(contentPair.contentItem, PortalBlockBodyLegacy)
+      if res.isOk():
+        if fromPortalBlockBody(BlockBody, res.value()).isOk():
+          contentBodies.inc()
+          continue
+
+    block:
+      let res = decodeSsz(contentPair.contentItem, AccountTrieNodeRetrieval)
+      if res.isOk():
+        if rlpDecodeAccountTrieNode(res.value().node).isOk():
+          contentAccount.inc()
+          continue
+
+    block:
+      let res = decodeSsz(contentPair.contentItem, ContractTrieNodeRetrieval)
+      if res.isOk():
+        if rlpDecodeContractTrieNode(res.value().node).isOk():
+          contentContract.inc()
+          continue
+
+    block:
+      let res = decodeSsz(contentPair.contentItem, ContractCodeRetrieval)
+      if res.isOk():
+        contentCode.inc()
+        continue
+
+    contentOther.inc()
+
+  notice "ContentDB type count done: ",
+    contentTotal, contentOldHeaders, contentNewHeaders, contentReceipts, contentBodies,
+    contentAccount, contentContract, contentCode, contentOther

fluffy/network/history/content/content_values.nim

@@ -26,25 +26,15 @@ const
   MAX_WITHDRAWALS_COUNT = MAX_WITHDRAWALS_PER_PAYLOAD
 
   MAX_EPHEMERAL_HEADER_PAYLOAD = 256
+  MAX_HEADER_PROOF_LENGTH* = 1024
 
 type
   ## BlockHeader types
   HistoricalHashesAccumulatorProof* = array[15, Digest]
 
-  BlockHeaderProofType* = enum
-    none = 0x00 # An SSZ Union None
-    historicalHashesAccumulatorProof = 0x01
-
-  BlockHeaderProof* = object
-    case proofType*: BlockHeaderProofType
-    of none:
-      discard
-    of historicalHashesAccumulatorProof:
-      historicalHashesAccumulatorProof*: HistoricalHashesAccumulatorProof
-
   BlockHeaderWithProof* = object
     header*: ByteList[MAX_HEADER_LENGTH] # RLP data
-    proof*: BlockHeaderProof
+    proof*: ByteList[MAX_HEADER_PROOF_LENGTH]
 
   ## Ephemeral BlockHeader list
   EphemeralBlockHeaderList* =
@@ -73,11 +63,3 @@ type
   ## Receipts types
   ReceiptByteList* = ByteList[MAX_RECEIPT_LENGTH] # RLP data
   PortalReceipts* = List[ReceiptByteList, MAX_TRANSACTION_COUNT]
-
-func init*(T: type BlockHeaderProof, proof: HistoricalHashesAccumulatorProof): T =
-  BlockHeaderProof(
-    proofType: historicalHashesAccumulatorProof, historicalHashesAccumulatorProof: proof
-  )
-
-func init*(T: type BlockHeaderProof): T =
-  BlockHeaderProof(proofType: none)

fluffy/network/history/content/content_values_deprecated.nim

@@ -0,0 +1,33 @@
+# fluffy
+# Copyright (c) 2025 Status Research & Development GmbH
+# Licensed and distributed under either of
+#   * MIT license (license terms in the root directory or at https://opensource.org/licenses/MIT).
+#   * Apache v2 license (license terms in the root directory or at https://www.apache.org/licenses/LICENSE-2.0).
+# at your option. This file may not be copied, modified, or distributed except according to those terms.
+
+{.push raises: [].}
+
+import std/math, nimcrypto/hash, ssz_serialization
+
+export ssz_serialization, hash
+
+const MAX_HEADER_LENGTH = 2 ^ 11 # = 2048
+
+type
+  ## BlockHeader types
+  HistoricalHashesAccumulatorProof* = array[15, Digest]
+
+  BlockHeaderProofType* = enum
+    none = 0x00 # An SSZ Union None
+    historicalHashesAccumulatorProof = 0x01
+
+  BlockHeaderProof* = object
+    case proofType*: BlockHeaderProofType
+    of none:
+      discard
+    of historicalHashesAccumulatorProof:
+      historicalHashesAccumulatorProof*: HistoricalHashesAccumulatorProof
+
+  BlockHeaderWithProofDeprecated* = object
+    header*: ByteList[MAX_HEADER_LENGTH] # RLP data
+    proof*: BlockHeaderProof

fluffy/network/history/history_validation.nim

@@ -50,20 +50,23 @@ func validateHeaderBytes*(
   ok(header)
 
 func verifyBlockHeaderProof*(
-    a: FinishedHistoricalHashesAccumulator, header: Header, proof: BlockHeaderProof
+    a: FinishedHistoricalHashesAccumulator,
+    header: Header,
+    proof: ByteList[MAX_HEADER_PROOF_LENGTH],
 ): Result[void, string] =
-  case proof.proofType
+  let timestamp = Moment.init(header.timestamp.int64, Second)
-  of BlockHeaderProofType.historicalHashesAccumulatorProof:
+
-    a.verifyAccumulatorProof(header, proof.historicalHashesAccumulatorProof)
+  if isShanghai(chainConfig, timestamp):
-  of BlockHeaderProofType.none:
+    # TODO: Add verification post merge based on historical_summaries
-    if header.isPreMerge():
+    err("Shanghai block verification not implemented")
-      err("Pre merge header requires HistoricalHashesAccumulatorProof")
+  elif isPoSBlock(chainConfig, header.number):
-    else:
+    # TODO: Add verification post merge based on historical_roots
-      # TODO:
+    err("PoS block verification not implemented")
-      # Add verification post merge based on historical_roots & historical_summaries
+  else:
-      # Lets for now no longer accept other headers without a proof and the most
+    let accumulatorProof = decodeSsz(proof.asSeq(), HistoricalHashesAccumulatorProof).valueOr:
-      # recent ones are now a different type.
+      return err("Failed decoding accumulator proof: " & error)
-      err("Post merge header proofs not yet activated")
+
+    a.verifyAccumulatorProof(header, accumulatorProof)
 
 func validateCanonicalHeaderBytes*(
     bytes: openArray[byte], id: uint64 | Hash32, a: FinishedHistoricalHashesAccumulator

fluffy/network/history/validation/historical_hashes_accumulator.nim

@@ -1,5 +1,5 @@
 # Nimbus
-# Copyright (c) 2022-2024 Status Research & Development GmbH
+# Copyright (c) 2022-2025 Status Research & Development GmbH
 # Licensed and distributed under either of
 #   * MIT license (license terms in the root directory or at https://opensource.org/licenses/MIT).
 #   * Apache v2 license (license terms in the root directory or at https://www.apache.org/licenses/LICENSE-2.0).
@@ -198,7 +198,7 @@ func buildHeaderWithProof*(
 
   ok(
     BlockHeaderWithProof(
-      header: ByteList[2048].init(rlp.encode(header)),
+      header: ByteList[MAX_HEADER_LENGTH].init(rlp.encode(header)),
-      proof: BlockHeaderProof.init(proof),
+      proof: ByteList[MAX_HEADER_PROOF_LENGTH].init(SSZ.encode(proof)),
     )
   )

fluffy/tests/rpc_tests/test_portal_rpc_client.nim

@@ -1,5 +1,5 @@
 # Nimbus - Portal Network
-# Copyright (c) 2021-2024 Status Research & Development GmbH
+# Copyright (c) 2021-2025 Status Research & Development GmbH
 # Licensed and distributed under either of
 #   * MIT license (license terms in the root directory or at https://opensource.org/licenses/MIT).
 #   * Apache v2 license (license terms in the root directory or at https://www.apache.org/licenses/LICENSE-2.0).
@@ -63,7 +63,8 @@ proc store*(hn: HistoryNode, blockHash: Hash32, blockHeader: Header) =
   let
     headerRlp = rlp.encode(blockHeader)
     blockHeaderWithProof = BlockHeaderWithProof(
-      header: ByteList[2048].init(headerRlp), proof: BlockHeaderProof.init()
+      header: ByteList[MAX_HEADER_LENGTH].init(headerRlp),
+      proof: ByteList[MAX_HEADER_PROOF_LENGTH].init(@[]),
     )
     contentKeyBytes = blockHeaderContentKey(blockHash).encode()
     contentId = history_content.toContentId(contentKeyBytes)

fluffy/tests/state_network_tests/state_test_helpers.nim

@@ -154,7 +154,8 @@ proc mockStateRootLookup*(
     blockHeader = Header(stateRoot: stateRoot)
     headerRlp = rlp.encode(blockHeader)
     blockHeaderWithProof = BlockHeaderWithProof(
-      header: ByteList[2048].init(headerRlp), proof: BlockHeaderProof.init()
+      header: ByteList[MAX_HEADER_LENGTH].init(headerRlp),
+      proof: ByteList[MAX_HEADER_PROOF_LENGTH].init(@[]),
     )
     contentKeyBytes = blockHeaderContentKey(blockNumOrHash).encode()
     contentId = history_content.toContentId(contentKeyBytes)

fluffy/tools/fcli_db.nim

@@ -1,12 +1,18 @@
 # Fluffy
-# Copyright (c) 2023-2024 Status Research & Development GmbH
+# Copyright (c) 2023-2025 Status Research & Development GmbH
 # Licensed and distributed under either of
 #   * MIT license (license terms in the root directory or at https://opensource.org/licenses/MIT).
 #   * Apache v2 license (license terms in the root directory or at https://www.apache.org/licenses/LICENSE-2.0).
 # at your option. This file may not be copied, modified, or distributed except according to those terms.
 
 import
-  chronicles, confutils, stint, eth/common/keys, ../database/content_db, ./benchmark
+  chronicles,
+  confutils,
+  stint,
+  eth/common/keys,
+  ../database/content_db,
+  ../database/content_db_migrate_deprecated,
+  ./benchmark
 
 when defined(posix):
   import system/ansi_c
@@ -29,6 +35,7 @@ type
     generate = "Generate random content into the database, for testing purposes."
     prune = "Prune the ContentDb in case of resizing or selecting a different local id"
     validate = "Validate all the content in the ContentDb"
+    migrate = "Migrate the ContentDb for new HeaderWithProof format"
 
   DbConf = object
     databaseDir* {.
@@ -60,6 +67,22 @@ type
       .}: bool
     of DbCmd.validate:
       discard
+    of DbCmd.migrate:
+      deleteNoProof* {.
+        desc: "Delete old HeaderWithProof content without proof",
+        defaultValue: true,
+        name: "delete-no-proof"
+      .}: bool
+      updateWithProof* {.
+        desc: "Update old HeaderWithProof per-merge content with proof",
+        defaultValue: true,
+        name: "update-with-proof"
+      .}: bool
+      checkTypes* {.
+        desc: "Iterate and count all content types",
+        defaultValue: false,
+        name: "debug-check-types"
+      .}: bool
 
 const maxDbSize = 4_000_000_000'u64
 
@@ -153,6 +176,21 @@ proc cmdPrune(conf: DbConf) =
     notice "Functionality not yet implemented"
     quit QuitSuccess
 
+proc cmdMigrate(conf: DbConf) =
+  let db = ContentDBDeprecated.new(conf.databaseDir.string)
+
+  if conf.checkTypes:
+    db.iterateAllAndCountTypes()
+
+  if conf.deleteNoProof:
+    db.deleteAllHeadersWithoutProof()
+
+  if conf.updateWithProof:
+    db.updateAllHeadersWithInvalidEncoding()
+
+  if conf.checkTypes and (conf.deleteNoProof or conf.updateWithProof):
+    db.iterateAllAndCountTypes()
+
 proc controlCHook() {.noconv.} =
   notice "Shutting down after having received SIGINT."
   quit QuitSuccess
@@ -177,3 +215,5 @@ when isMainModule:
     cmdPrune(conf)
   of DbCmd.validate:
     notice "Functionality not yet implemented"
+  of DbCmd.migrate:
+    cmdMigrate(conf)

vendor/portal-spec-tests

@@ -1 +1 @@
-Subproject commit 9d48845c2f01ccd95e9d4d6ffe4eb2b231b7cc40
+Subproject commit 6f2a2a8a14225c2acabfd2d2707c5c75edbb6b46