Merged main

1 files changed, 195 insertions, 80 deletions

diff --git a/zenserver/cache/structuredcachestore.cpp b/zenserver/cache/structuredcachestore.cpp
index 5e93ebaa9..580446473 100644
--- a/zenserver/cache/structuredcachestore.cpp
+++ b/zenserver/cache/structuredcachestore.cpp
@@ -32,6 +32,8 @@ ZenCacheStore::ZenCacheStore(CasStore& Cas, const std::filesystem::path& RootDir
 {
 	ZEN_INFO("initializing structured cache at '{}'", RootDir);
 	CreateDirectories(RootDir);
+
+	m_DiskLayer.DiscoverBuckets();
 }
 
 ZenCacheStore::~ZenCacheStore()
@@ -116,6 +118,13 @@ ZenCacheStore::Scrub(ScrubContext& Ctx)
 	m_DiskLayer.Scrub(Ctx);
 	m_MemLayer.Scrub(Ctx);
 }
+
+void
+ZenCacheStore::GarbageCollect(GcContext& GcCtx)
+{
+	ZEN_UNUSED(GcCtx);
+}
+
 //////////////////////////////////////////////////////////////////////////
 
 ZenCacheMemoryLayer::ZenCacheMemoryLayer()
@@ -142,6 +151,10 @@ ZenCacheMemoryLayer::Get(std::string_view InBucket, const IoHash& HashKey, ZenCa
 
 	_.ReleaseNow();
 
+	// There's a race here. Since the lock is released early to allow
+	// inserts, the bucket delete path could end up deleting the
+	// underlying data structure
+
 	return Bucket->Get(HashKey, OutValue);
 }
 
@@ -195,13 +208,21 @@ ZenCacheMemoryLayer::Scrub(ScrubContext& Ctx)
 }
 
 void
+ZenCacheMemoryLayer::GarbageCollect(GcContext& GcCtx)
+{
+	ZEN_UNUSED(GcCtx);
+}
+
+void
 ZenCacheMemoryLayer::CacheBucket::Scrub(ScrubContext& Ctx)
 {
+	RwLock::SharedLockScope _(m_bucketLock);
+
 	std::vector<IoHash> BadHashes;
 
 	for (auto& Kv : m_cacheMap)
 	{
-		if (Kv.first != IoHash::HashBuffer(Kv.second))
+		if (Kv.first != IoHash::HashBuffer(Kv.second.Payload))
 		{
 			BadHashes.push_back(Kv.first);
 		}
@@ -209,10 +230,16 @@ ZenCacheMemoryLayer::CacheBucket::Scrub(ScrubContext& Ctx)
 
 	if (!BadHashes.empty())
 	{
-		Ctx.ReportBadChunks(BadHashes);
+		Ctx.ReportBadCasChunks(BadHashes);
 	}
 }
 
+void
+ZenCacheMemoryLayer::CacheBucket::GarbageCollect(GcContext& GcCtx)
+{
+	ZEN_UNUSED(GcCtx);
+}
+
 bool
 ZenCacheMemoryLayer::CacheBucket::Get(const IoHash& HashKey, ZenCacheValue& OutValue)
 {
@@ -224,18 +251,26 @@ ZenCacheMemoryLayer::CacheBucket::Get(const IoHash& HashKey, ZenCacheValue& OutV
 	}
 	else
 	{
-		OutValue.Value = bucketIt->second;
+		BucketValue& Value = bucketIt.value();
+		OutValue.Value	   = Value.Payload;
+		Value.LastAccess   = GetCurrentTimeStamp();
 
 		return true;
 	}
 }
 
+uint64_t
+ZenCacheMemoryLayer::CacheBucket::GetCurrentTimeStamp()
+{
+	return GetLofreqTimerValue();
+}
+
 void
 ZenCacheMemoryLayer::CacheBucket::Put(const IoHash& HashKey, const ZenCacheValue& Value)
 {
 	RwLock::ExclusiveLockScope _(m_bucketLock);
 
-	m_cacheMap[HashKey] = Value.Value;
+	m_cacheMap.insert_or_assign(HashKey, BucketValue{.LastAccess = GetCurrentTimeStamp(), .Payload = Value.Value});
 }
 
 //////////////////////////////////////////////////////////////////////////
@@ -245,11 +280,17 @@ ZenCacheMemoryLayer::CacheBucket::Put(const IoHash& HashKey, const ZenCacheValue
 
 struct DiskLocation
 {
-	uint64_t OffsetAndFlags;
-	uint32_t Size;
-	uint32_t IndexDataSize;
+	inline DiskLocation() = default;
+
+	inline DiskLocation(uint64_t Offset, uint64_t ValueSize, uint32_t IndexSize, uint64_t Flags)
+	: OffsetAndFlags(CombineOffsetAndFlags(Offset, Flags))
+	, LowerSize(ValueSize & 0xFFFFffff)
+	, IndexDataSize(IndexSize)
+	{
+	}
 
-	static const uint64_t kOffsetMask	  = 0x00FF'ffFF'ffFF'ffFFull;
+	static const uint64_t kOffsetMask	  = 0x0000'ffFF'ffFF'ffFFull;
+	static const uint64_t kSizeMask		  = 0x00FF'0000'0000'0000ull;
 	static const uint64_t kFlagsMask	  = 0xff00'0000'0000'0000ull;
 	static const uint64_t kStandaloneFile = 0x8000'0000'0000'0000ull;
 	static const uint64_t kStructured	  = 0x4000'0000'0000'0000ull;
@@ -257,6 +298,7 @@ struct DiskLocation
 	static uint64_t CombineOffsetAndFlags(uint64_t Offset, uint64_t Flags) { return Offset | Flags; }
 
 	inline uint64_t		  Offset() const { return OffsetAndFlags & kOffsetMask; }
+	inline uint64_t		  Size() const { return LowerSize; }
 	inline uint64_t		  IsFlagSet(uint64_t Flag) const { return OffsetAndFlags & Flag; }
 	inline ZenContentType GetContentType() const
 	{
@@ -269,6 +311,11 @@ struct DiskLocation
 
 		return ContentType;
 	}
+
+private:
+	uint64_t OffsetAndFlags = 0;
+	uint32_t LowerSize		= 0;
+	uint32_t IndexDataSize	= 0;
 };
 
 struct DiskIndexEntry
@@ -286,7 +333,7 @@ struct ZenCacheDiskLayer::CacheBucket
 	CacheBucket(CasStore& Cas);
 	~CacheBucket();
 
-	void		OpenOrCreate(std::filesystem::path BucketDir);
+	void		OpenOrCreate(std::filesystem::path BucketDir, bool AllowCreate = true);
 	static bool Delete(std::filesystem::path BucketDir);
 
 	bool Get(const IoHash& HashKey, ZenCacheValue& OutValue);
@@ -294,14 +341,15 @@ struct ZenCacheDiskLayer::CacheBucket
 	void Drop();
 	void Flush();
 	void Scrub(ScrubContext& Ctx);
+	void GarbageCollect(GcContext& GcCtx);
 
-	inline bool IsOk() const { return m_Ok; }
+	inline bool IsOk() const { return m_IsOk; }
 
 private:
 	CasStore&			  m_CasStore;
 	std::filesystem::path m_BucketDir;
 	Oid					  m_BucketId;
-	bool				  m_Ok					 = false;
+	bool				  m_IsOk				 = false;
 	uint64_t			  m_LargeObjectThreshold = 64 * 1024;
 
 	// These files are used to manage storage of small objects for this bucket
@@ -314,9 +362,19 @@ private:
 	uint64_t											 m_WriteCursor = 0;
 
 	void BuildPath(WideStringBuilderBase& Path, const IoHash& HashKey);
-	void PutLargeObject(const IoHash& HashKey, const ZenCacheValue& Value);
-	bool GetStandaloneCacheValue(const IoHash& HashKey, ZenCacheValue& OutValue, const DiskLocation& Loc);
+	void PutStandaloneCacheValue(const IoHash& HashKey, const ZenCacheValue& Value);
+	bool GetStandaloneCacheValue(const DiskLocation& Loc, const IoHash& HashKey, ZenCacheValue& OutValue);
 	bool GetInlineCacheValue(const DiskLocation& Loc, ZenCacheValue& OutValue);
+
+	// These locks are here to avoid contention on file creation, therefore it's sufficient
+	// that we take the same lock for the same hash
+	//
+	// These locks are small and should really be spaced out so they don't share cache lines,
+	// but we don't currently access them at particularly high frequency so it should not be
+	// an issue in practice
+
+	RwLock		   m_ShardedLocks[256];
+	inline RwLock& LockForHash(const IoHash& Hash) { return m_ShardedLocks[Hash.Hash[19]]; }
 };
 
 ZenCacheDiskLayer::CacheBucket::CacheBucket(CasStore& Cas) : m_CasStore(Cas)
@@ -341,7 +399,7 @@ ZenCacheDiskLayer::CacheBucket::Delete(std::filesystem::path BucketDir)
 }
 
 void
-ZenCacheDiskLayer::CacheBucket::OpenOrCreate(std::filesystem::path BucketDir)
+ZenCacheDiskLayer::CacheBucket::OpenOrCreate(std::filesystem::path BucketDir, bool AllowCreate)
 {
 	CreateDirectories(BucketDir);
 
@@ -368,17 +426,23 @@ ZenCacheDiskLayer::CacheBucket::OpenOrCreate(std::filesystem::path BucketDir)
 		{
 			ManifestFile.Read(&m_BucketId, sizeof(Oid), 0);
 
-			m_Ok = true;
+			m_IsOk = true;
 		}
 
-		if (!m_Ok)
+		if (!m_IsOk)
 		{
 			ManifestFile.Close();
 		}
 	}
 
-	if (!m_Ok)
+	if (!m_IsOk)
 	{
+		if (AllowCreate == false)
+		{
+			// Invalid bucket
+			return;
+		}
+
 		// No manifest file found, this is a new bucket
 
 		ManifestFile.Open(ManifestPath, /* IsCreate */ true, Ec);
@@ -410,13 +474,13 @@ ZenCacheDiskLayer::CacheBucket::OpenOrCreate(std::filesystem::path BucketDir)
 		m_SlogFile.Replay([&](const DiskIndexEntry& Record) {
 			m_Index[Record.Key] = Record.Location;
 
-			MaxFileOffset = std::max<uint64_t>(MaxFileOffset, Record.Location.Offset() + Record.Location.Size);
+			MaxFileOffset = std::max<uint64_t>(MaxFileOffset, Record.Location.Offset() + Record.Location.Size());
 		});
 
 		m_WriteCursor = (MaxFileOffset + 15) & ~15;
 	}
 
-	m_Ok = true;
+	m_IsOk = true;
 }
 
 void
@@ -437,23 +501,25 @@ ZenCacheDiskLayer::CacheBucket::BuildPath(WideStringBuilderBase& Path, const IoH
 bool
 ZenCacheDiskLayer::CacheBucket::GetInlineCacheValue(const DiskLocation& Loc, ZenCacheValue& OutValue)
 {
-	if (!Loc.IsFlagSet(DiskLocation::kStandaloneFile))
+	if (Loc.IsFlagSet(DiskLocation::kStandaloneFile))
 	{
-		OutValue.Value = IoBufferBuilder::MakeFromFileHandle(m_SobsFile.Handle(), Loc.Offset(), Loc.Size);
-		OutValue.Value.SetContentType(Loc.GetContentType());
-
-		return true;
+		return false;
 	}
 
-	return false;
+	OutValue.Value = IoBufferBuilder::MakeFromFileHandle(m_SobsFile.Handle(), Loc.Offset(), Loc.Size());
+	OutValue.Value.SetContentType(Loc.GetContentType());
+
+	return true;
 }
 
 bool
-ZenCacheDiskLayer::CacheBucket::GetStandaloneCacheValue(const IoHash& HashKey, ZenCacheValue& OutValue, const DiskLocation& Loc)
+ZenCacheDiskLayer::CacheBucket::GetStandaloneCacheValue(const DiskLocation& Loc, const IoHash& HashKey, ZenCacheValue& OutValue)
 {
 	WideStringBuilder<128> DataFilePath;
 	BuildPath(DataFilePath, HashKey);
 
+	RwLock::SharedLockScope ValueLock(LockForHash(HashKey));
+
 	if (IoBuffer Data = IoBufferBuilder::MakeFromFile(DataFilePath.c_str()))
 	{
 		OutValue.Value = Data;
@@ -468,7 +534,7 @@ ZenCacheDiskLayer::CacheBucket::GetStandaloneCacheValue(const IoHash& HashKey, Z
 bool
 ZenCacheDiskLayer::CacheBucket::Get(const IoHash& HashKey, ZenCacheValue& OutValue)
 {
-	if (!m_Ok)
+	if (!m_IsOk)
 	{
 		return false;
 	}
@@ -486,7 +552,7 @@ ZenCacheDiskLayer::CacheBucket::Get(const IoHash& HashKey, ZenCacheValue& OutVal
 
 		_.ReleaseNow();
 
-		return GetStandaloneCacheValue(HashKey, OutValue, Loc);
+		return GetStandaloneCacheValue(Loc, HashKey, OutValue);
 	}
 
 	return false;
@@ -495,14 +561,14 @@ ZenCacheDiskLayer::CacheBucket::Get(const IoHash& HashKey, ZenCacheValue& OutVal
 void
 ZenCacheDiskLayer::CacheBucket::Put(const IoHash& HashKey, const ZenCacheValue& Value)
 {
-	if (!m_Ok)
+	if (!m_IsOk)
 	{
 		return;
 	}
 
 	if (Value.Value.Size() >= m_LargeObjectThreshold)
 	{
-		return PutLargeObject(HashKey, Value);
+		return PutStandaloneCacheValue(HashKey, Value);
 	}
 	else
 	{
@@ -517,10 +583,9 @@ ZenCacheDiskLayer::CacheBucket::Put(const IoHash& HashKey, const ZenCacheValue&
 
 		RwLock::ExclusiveLockScope _(m_IndexLock);
 
-		DiskLocation Loc{.OffsetAndFlags = DiskLocation::CombineOffsetAndFlags(m_WriteCursor, EntryFlags),
-						 .Size			 = gsl::narrow<uint32_t>(Value.Value.Size())};
+		DiskLocation Loc(m_WriteCursor, Value.Value.Size(), 0, EntryFlags);
 
-		m_WriteCursor = RoundUp(m_WriteCursor + Loc.Size, 16);
+		m_WriteCursor = RoundUp(m_WriteCursor + Loc.Size(), 16);
 
 		if (auto it = m_Index.find(HashKey); it == m_Index.end())
 		{
@@ -530,11 +595,13 @@ ZenCacheDiskLayer::CacheBucket::Put(const IoHash& HashKey, const ZenCacheValue&
 		else
 		{
 			// TODO: should check if write is idempotent and bail out if it is?
+			// this would requiring comparing contents on disk unless we add a
+			// content hash to the index entry
 			it.value() = Loc;
 		}
 
 		m_SlogFile.Append({.Key = HashKey, .Location = Loc});
-		m_SobsFile.Write(Value.Value.Data(), Loc.Size, Loc.Offset());
+		m_SobsFile.Write(Value.Value.Data(), Loc.Size(), Loc.Offset());
 	}
 }
 
@@ -558,61 +625,69 @@ ZenCacheDiskLayer::CacheBucket::Flush()
 void
 ZenCacheDiskLayer::CacheBucket::Scrub(ScrubContext& Ctx)
 {
-	std::vector<DiskIndexEntry> StandaloneFiles;
+	std::atomic<uint64_t> ScrubbedChunks{0}, ScrubbedBytes{0};
 
-	std::vector<IoHash>	  BadChunks;
-	std::vector<IoBuffer> BadStandaloneChunks;
+	std::vector<IoHash> BadChunks;
 
 	{
 		RwLock::SharedLockScope _(m_IndexLock);
 
 		for (auto& Kv : m_Index)
 		{
-			const IoHash&		Hash = Kv.first;
-			const DiskLocation& Loc	 = Kv.second;
+			const IoHash&		HashKey = Kv.first;
+			const DiskLocation& Loc		= Kv.second;
 
 			ZenCacheValue Value;
 
-			if (!GetInlineCacheValue(Loc, Value))
+			if (GetInlineCacheValue(Loc, Value))
 			{
-				ZEN_ASSERT(Loc.IsFlagSet(DiskLocation::kStandaloneFile));
-				StandaloneFiles.push_back({.Key = Hash, .Location = Loc});
+				// Validate contents
 			}
 			else
 			{
-				if (GetStandaloneCacheValue(Hash, Value, Loc))
+				if (Loc.IsFlagSet(DiskLocation::kStandaloneFile))
 				{
-					// Hash contents
-
-					const IoHash ComputedHash = HashBuffer(Value.Value);
-
-					if (ComputedHash != Hash)
+					if (GetStandaloneCacheValue(Loc, HashKey, Value))
 					{
-						BadChunks.push_back(Hash);
+						// Note: we cannot currently validate contents since we don't
+						// have a content hash!
+					}
+					else
+					{
+						// Value not found
+						BadChunks.push_back(HashKey);
 					}
-				}
-				else
-				{
-					// Non-existent
 				}
 			}
 		}
 	}
 
-	if (Ctx.RunRecovery())
+	Ctx.ReportScrubbed(ScrubbedChunks, ScrubbedBytes);
+
+	if (BadChunks.empty())
 	{
-		// Clean out bad chunks
+		return;
 	}
 
-	if (!BadChunks.empty())
+	Ctx.ReportBadCasChunks(BadChunks);
+
+	if (Ctx.RunRecovery())
 	{
-		Ctx.ReportBadChunks(BadChunks);
+		// Clean out bad data
 	}
 }
 
 void
-ZenCacheDiskLayer::CacheBucket::PutLargeObject(const IoHash& HashKey, const ZenCacheValue& Value)
+ZenCacheDiskLayer::CacheBucket::GarbageCollect(GcContext& GcCtx)
 {
+	ZEN_UNUSED(GcCtx);
+}
+
+void
+ZenCacheDiskLayer::CacheBucket::PutStandaloneCacheValue(const IoHash& HashKey, const ZenCacheValue& Value)
+{
+	RwLock::ExclusiveLockScope ValueLock(LockForHash(HashKey));
+
 	WideStringBuilder<128> DataFilePath;
 	BuildPath(DataFilePath, HashKey);
 
@@ -661,7 +736,7 @@ ZenCacheDiskLayer::CacheBucket::PutLargeObject(const IoHash& HashKey, const ZenC
 
 	RwLock::ExclusiveLockScope _(m_IndexLock);
 
-	DiskLocation Loc{.OffsetAndFlags = DiskLocation::CombineOffsetAndFlags(0, EntryFlags), .Size = 0};
+	DiskLocation Loc(/* Offset */ 0, Value.Value.Size(), 0, EntryFlags);
 
 	if (auto it = m_Index.find(HashKey); it == m_Index.end())
 	{
@@ -719,7 +794,7 @@ ZenCacheDiskLayer::Get(std::string_view InBucket, const IoHash& HashKey, ZenCach
 			std::filesystem::path BucketPath = m_RootDir;
 			BucketPath /= std::string(InBucket);
 
-			Bucket->OpenOrCreate(BucketPath.c_str());
+			Bucket->OpenOrCreate(BucketPath);
 		}
 	}
 
@@ -762,7 +837,7 @@ ZenCacheDiskLayer::Put(std::string_view InBucket, const IoHash& HashKey, const Z
 			std::filesystem::path bucketPath = m_RootDir;
 			bucketPath /= std::string(InBucket);
 
-			Bucket->OpenOrCreate(bucketPath.c_str());
+			Bucket->OpenOrCreate(bucketPath);
 		}
 	}
 
@@ -774,6 +849,63 @@ ZenCacheDiskLayer::Put(std::string_view InBucket, const IoHash& HashKey, const Z
 	}
 }
 
+void
+ZenCacheDiskLayer::DiscoverBuckets()
+{
+	FileSystemTraversal Traversal;
+	struct Visitor : public FileSystemTraversal::TreeVisitor
+	{
+		virtual void VisitFile([[maybe_unused]] const std::filesystem::path& Parent,
+							   [[maybe_unused]] const path_view&			 File,
+							   [[maybe_unused]] uint64_t					 FileSize) override
+		{
+		}
+
+		virtual bool VisitDirectory([[maybe_unused]] const std::filesystem::path& Parent, const path_view& DirectoryName) override
+		{
+			Dirs.push_back(std::wstring(DirectoryName));
+			return false;
+		}
+
+		std::vector<std::wstring> Dirs;
+	} Visit;
+
+	Traversal.TraverseFileSystem(m_RootDir, Visit);
+
+	// Initialize buckets
+
+	RwLock::ExclusiveLockScope _(m_Lock);
+
+	for (const std::wstring& BucketName : Visit.Dirs)
+	{
+		// New bucket needs to be created
+
+		std::string BucketName8 = WideToUtf8(BucketName);
+
+		if (auto It = m_Buckets.find(BucketName8); It != m_Buckets.end())
+		{
+		}
+		else
+		{
+			auto InsertResult = m_Buckets.try_emplace(BucketName8, m_CasStore);
+
+			std::filesystem::path BucketPath = m_RootDir;
+			BucketPath /= BucketName8;
+
+			CacheBucket& Bucket = InsertResult.first->second;
+
+			Bucket.OpenOrCreate(BucketPath, /* AllowCreate */ false);
+
+			if (!Bucket.IsOk())
+			{
+				ZEN_WARN("Found directory '{}' in our base directory '{}' but it is not a valid bucket", BucketName8, m_RootDir);
+
+				m_Buckets.erase(InsertResult.first);
+			}
+		}
+	}
+}
+
 bool
 ZenCacheDiskLayer::DropBucket(std::string_view InBucket)
 {
@@ -830,27 +962,10 @@ ZenCacheDiskLayer::Scrub(ScrubContext& Ctx)
 	}
 }
 
-//////////////////////////////////////////////////////////////////////////
-
-ZenCacheTracker::ZenCacheTracker(ZenCacheStore& CacheStore)
-{
-	ZEN_UNUSED(CacheStore);
-}
-
-ZenCacheTracker::~ZenCacheTracker()
-{
-}
-
-void
-ZenCacheTracker::TrackAccess(std::string_view Bucket, const IoHash& HashKey)
-{
-	ZEN_UNUSED(Bucket);
-	ZEN_UNUSED(HashKey);
-}
-
 void
-ZenCacheTracker::Flush()
+ZenCacheDiskLayer::GarbageCollect(GcContext& GcCtx)
 {
+	ZEN_UNUSED(GcCtx);
 }
 
 }  // namespace zen