Add BlobHeader integrity validation

conanfile.py

@@ -9,7 +9,7 @@
 
 class HomeObjectConan(ConanFile):
     name = "homeobject"
-    version = "2.2.11"
+    version = "2.2.12"
 
     homepage = "https://github.com/eBay/HomeObject"
     description = "Blob Store built on HomeReplication"

src/lib/homestore_backend/hs_blob_manager.cpp

@@ -156,6 +156,7 @@ BlobManager::AsyncResult< blob_id_t > HSHomeObject::_put_blob(ShardInfo const& s
     compute_blob_payload_hash(req->blob_header()->hash_algorithm, blob.body.cbytes(), blob_size,
                               (uint8_t*)blob.user_key.data(), blob.user_key.size(), req->blob_header()->hash,
                               BlobHeader::blob_max_hash_len);
+    req->blob_header()->seal();
 
     // Add blob body to the request
     req->add_data_sg(std::move(blob.body));
@@ -360,8 +361,7 @@ HSHomeObject::blob_put_get_blk_alloc_hints(sisl::blob const& header, cintrusive<
 
     auto hs_pg = get_hs_pg(msg_header->pg_id);
     if (hs_pg == nullptr) {
-        LOGW("Received a blob_put on an unknown pg:{}, underlying engine will retry this later",
-             msg_header->pg_id);
+        LOGW("Received a blob_put on an unknown pg:{}, underlying engine will retry this later", msg_header->pg_id);
         return folly::makeUnexpected(homestore::ReplServiceError::RESULT_NOT_EXIST_YET);
     }
 

src/lib/homestore_backend/hs_homeobject.hpp

@@ -157,7 +157,7 @@ class HSHomeObject : public HomeObjectImpl {
         data_type_t type{data_type_t::BLOB_INFO};
     };
 
-    struct shard_info_superblk : public DataHeader {
+    struct shard_info_superblk : DataHeader {
         ShardInfo info;
         homestore::chunk_num_t p_chunk_id;
         homestore::chunk_num_t v_chunk_id;
@@ -301,7 +301,7 @@ class HSHomeObject : public HomeObjectImpl {
 #pragma pack(1)
     // Every blob payload stored in disk as blob header | blob data | blob metadata(optional) | padding.
     // Padding of zeroes is added to make sure the whole payload be aligned to device block size.
-    struct BlobHeader : public DataHeader {
+    struct BlobHeader : DataHeader {
         static constexpr uint64_t blob_max_hash_len = 32;
 
         enum class HashAlgorithm : uint8_t {
@@ -312,6 +312,7 @@ class HSHomeObject : public HomeObjectImpl {
         };
 
         HashAlgorithm hash_algorithm;
+        mutable uint8_t header_hash[blob_max_hash_len]{};
         uint8_t hash[blob_max_hash_len]{};
         shard_id_t shard_id;
         blob_id_t blob_id;
@@ -325,6 +326,46 @@ class HSHomeObject : public HomeObjectImpl {
                                magic, version, shard_id, blob_size, user_key_size, (uint8_t)hash_algorithm,
                                spdlog::to_hex(hash, hash + blob_max_hash_len));
         }
+
+        bool valid() const {
+            if (!DataHeader::valid()) { return false; }
+
+            uint8_t hash_arr[blob_max_hash_len];
+            std::memcpy(hash_arr, header_hash, blob_max_hash_len);
+            if (!_do_seal()) {
+                std::memcpy(header_hash, hash_arr, blob_max_hash_len);
+                return false;
+            }
+
+            bool hash_valid = std::memcmp(header_hash, hash_arr, blob_max_hash_len) == 0;
+            std::memcpy(header_hash, hash_arr, blob_max_hash_len);
+            return hash_valid;
+        }
+
+        void seal() const {
+            if (!_do_seal()) {
+                DEBUG_ASSERT(false, "Invalid hash algorithm"); // Not implemented
+            }
+        }
+
+    private:
+        bool _do_seal() const {
+            switch (hash_algorithm) {
+            case HashAlgorithm::NONE:
+                return true;
+            case HashAlgorithm::CRC32: {
+                std::memset(header_hash, 0, blob_max_hash_len);
+                uint32_t computed_hash = crc32_ieee(0, (uint8_t*)this, sizeof(BlobHeader));
+                std::memcpy(header_hash, &computed_hash, sizeof(uint32_t));
+                return true;
+            }
+            case HashAlgorithm::MD5:
+            case HashAlgorithm::SHA1:
+            default:
+                break; // Not implemented
+            }
+            return false;
+        }
     };
 #pragma pack()
 

src/lib/homestore_backend/tests/homeobj_misc_tests.cpp

@@ -303,6 +303,7 @@ TEST_F(HomeObjectFixture, SnapshotReceiveHandler) {
                                                      reinterpret_cast< uint8_t* >(blob.user_key.data()),
                                                      blob.user_key.size(), hdr.hash,
                                                      HSHomeObject::BlobHeader::blob_max_hash_len);
+                hdr.seal();
 
                 std::memcpy(blob_raw.bytes(), &hdr, sizeof(HSHomeObject::BlobHeader));
                 if (!blob.user_key.empty()) {