1 files changed, 80 insertions, 2 deletions

diff --git a/zenstore/cidstore.cpp b/zenstore/cidstore.cpp
index 100054a0e..5e266f9d3 100644
--- a/zenstore/cidstore.cpp
+++ b/zenstore/cidstore.cpp
@@ -73,24 +73,96 @@ struct CidStore::CidState
 	bool ContainsChunk(const IoHash& DecompressedId)
 	{
 		RwLock::SharedLockScope _(m_Lock);
+		// Note that we do not check CAS here. This is optimistic but usually
+		// what we want.
 		return m_CidMap.find(DecompressedId) != m_CidMap.end();
 	}
 
 	void InitializeIndex(const std::filesystem::path& RootDir)
 	{
-		zen::CreateDirectories(RootDir);
+		CreateDirectories(RootDir);
 		std::filesystem::path SlogPath{RootDir / "cid.slog"};
 
 		bool IsNew = !std::filesystem::exists(SlogPath);
 
 		m_LogFile.Open(SlogPath, IsNew);
 
-		m_LogFile.Replay([&](const IndexEntry& Ie) { m_CidMap.insert_or_assign(Ie.Uncompressed, Ie.Compressed); });
+		m_LogFile.Replay([&](const IndexEntry& Ie) {
+			if (Ie.Compressed != IoHash::Zero)
+			{
+				// Update
+				m_CidMap.insert_or_assign(Ie.Uncompressed, Ie.Compressed);
+			}
+			else
+			{
+				// Tombstone
+				m_CidMap.erase(Ie.Uncompressed);
+			}
+		});
 
 		ZEN_DEBUG("CID index initialized: {} entries found", m_CidMap.size());
 	}
 
 	void Flush() { m_LogFile.Flush(); }
+
+	void Scrub(ScrubContext& Ctx)
+	{
+		CasChunkSet ChunkSet;
+
+		{
+			RwLock::SharedLockScope _(m_Lock);
+
+			for (auto& Kv : m_CidMap)
+			{
+				ChunkSet.AddChunk(Kv.second);
+			}
+		}
+
+		m_CasStore.FilterChunks(ChunkSet);
+
+		if (ChunkSet.IsEmpty())
+		{
+			// All good - we have all the chunks
+			return;
+		}
+
+		ZEN_ERROR("Scrubbing found that {} cid mappings mapped to non-existent CAS chunks", ChunkSet.GetChunkSet().size());
+
+		// Erase all mappings to chunks which are not present in the underlying CAS store
+		// we do this by removing mappings from the in-memory lookup structure and also
+		// by emitting tombstone records to the commit log
+
+		const auto&			MissingChunks = ChunkSet.GetChunkSet();
+		std::vector<IoHash> BadChunks;
+		{
+			RwLock::SharedLockScope _(m_Lock);
+
+			for (auto It = begin(m_CidMap), ItEnd = end(m_CidMap); It != ItEnd;)
+			{
+				if (auto MissingIt = MissingChunks.find(It->second); MissingIt != MissingChunks.end())
+				{
+					const IoHash& BadHash = It->first;
+
+					// Log a tombstone record
+					m_LogFile.Append({.Uncompressed = BadHash, .Compressed = IoHash::Zero});
+
+					BadChunks.push_back(BadHash);
+
+					It = m_CidMap.erase(It);
+				}
+				else
+				{
+					++It;
+				}
+			}
+		}
+
+		m_LogFile.Flush();
+
+		// TODO: Should compute a snapshot index here
+
+		Ctx.ReportBadChunks(BadChunks);
+	}
 };
 
 //////////////////////////////////////////////////////////////////////////
@@ -134,4 +206,10 @@ CidStore::Flush()
 	m_Impl->Flush();
 }
 
+void
+CidStore::Scrub(ScrubContext& Ctx)
+{
+	m_Impl->Scrub(Ctx);
+}
+
 }  // namespace zen