Implemended basic scrubbing / detection of disk corruption. Still needs more code to propagate errors and make adjustments to account for them in higher level data structures

1 files changed, 78 insertions, 17 deletions

diff --git a/zenstore/filecas.cpp b/zenstore/filecas.cpp
index 5fdf505d4..3314beb7e 100644
--- a/zenstore/filecas.cpp
+++ b/zenstore/filecas.cpp
@@ -10,6 +10,7 @@
 #include <zencore/string.h>
 #include <zencore/thread.h>
 #include <zencore/uid.h>
+#include <zenstore/basicfile.h>
 
 #include <gsl/gsl-lite.hpp>
 
@@ -17,6 +18,7 @@
 #include <functional>
 #include <unordered_map>
 
+// clang-format off
 #include <zencore/prewindows.h>
 
 struct IUnknown;  // Workaround for "combaseapi.h(229): error C2187: syntax error: 'identifier' was unexpected here" when using /permissive-
@@ -24,13 +26,19 @@ struct IUnknown;  // Workaround for "combaseapi.h(229): error C2187: syntax erro
 
 #include <zencore/postwindows.h>
 // clang-format on
-//
-//////////////////////////////////////////////////////////////////////////
 
 namespace zen {
 
 using namespace fmt::literals;
 
+FileCasStrategy::FileCasStrategy(const CasStoreConfiguration& Config) : m_Config(Config)
+{
+}
+
+FileCasStrategy::~FileCasStrategy()
+{
+}
+
 WideStringBuilderBase&
 FileCasStrategy::MakeShardedPath(WideStringBuilderBase& ShardedPath, const IoHash& ChunkHash, size_t& OutShard2len)
 {
@@ -56,7 +64,7 @@ FileCasStrategy::MakeShardedPath(WideStringBuilderBase& ShardedPath, const IoHas
 	OutShard2len = ShardedPath.Size();
 
 	ShardedPath.Append('\\');
-	ShardedPath.AppendAsciiRange(str + 6, str + 64);
+	ShardedPath.AppendAsciiRange(str + 5, str + 64);
 
 	return ShardedPath;
 }
@@ -259,12 +267,9 @@ FileCasStrategy::InsertChunk(const void* const ChunkData, const size_t ChunkSize
 	}
 
 	// We cannot rely on RAII to close the file handle since it would be closed
-	// *after* the lock is released due to the initialization order.
+	// *after* the lock is released due to the initialization order
 	PayloadFile.Close();
 
-	AtomicIncrement(m_Stats.PutCount);
-	AtomicAdd(m_Stats.PutBytes, ChunkSize);
-
 	return {.New = true};
 }
 
@@ -279,15 +284,7 @@ FileCasStrategy::FindChunk(const IoHash& ChunkHash)
 
 	RwLock::SharedLockScope _(LockForHash(ChunkHash));
 
-	auto Chunk = IoBufferBuilder::MakeFromFile(ShardedPath.c_str());
-
-	if (Chunk)
-	{
-		AtomicIncrement(m_Stats.GetCount);
-		AtomicAdd(m_Stats.GetBytes, Chunk.Size());
-	}
-
-	return Chunk;
+	return IoBufferBuilder::MakeFromFile(ShardedPath.c_str());
 }
 
 bool
@@ -338,6 +335,56 @@ FileCasStrategy::FilterChunks(CasChunkSet& InOutChunks)
 }
 
 void
+FileCasStrategy::IterateChunks(std::function<void(const IoHash& Hash, BasicFile& PayloadFile)>&& Callback)
+{
+	struct Visitor : public FileSystemTraversal::TreeVisitor
+	{
+		Visitor(const std::filesystem::path& RootDir) : RootDirectory(RootDir) {}
+		virtual void VisitFile(const std::filesystem::path& Parent, const std::wstring_view& File, uint64_t FileSize) override
+		{
+			std::filesystem::path RelPath = std::filesystem::relative(Parent, RootDirectory);
+
+			std::wstring PathString = RelPath.native();
+
+			if ((PathString.size() == (3 + 2 + 1)) && (File.size() == (40 - 3 - 2)))
+			{
+				if (PathString.at(3) == std::filesystem::path::preferred_separator)
+				{
+					PathString.erase(3, 1);
+				}
+				PathString.append(File);
+
+				StringBuilder<64> Utf8;
+				WideToUtf8(PathString, Utf8);
+
+				// TODO: should validate that we're actually dealing with a valid hex string here
+
+				IoHash NameHash = IoHash::FromHexString({Utf8.Data(), Utf8.Size()});
+
+				BasicFile		PayloadFile;
+				std::error_code Ec;
+				PayloadFile.Open(Parent / File, false, Ec);
+
+				if (!Ec)
+				{
+					Callback(NameHash, PayloadFile);
+				}
+			}
+		}
+
+		virtual bool VisitDirectory(const std::filesystem::path& Parent, const std::wstring_view& DirectoryName) { return true; }
+
+		const std::filesystem::path&									RootDirectory;
+		std::function<void(const IoHash& Hash, BasicFile& PayloadFile)> Callback;
+	} CasVisitor{m_Config.RootDirectory};
+
+	CasVisitor.Callback = std::move(Callback);
+
+	FileSystemTraversal Traversal;
+	Traversal.TraverseFileSystem(m_Config.RootDirectory, CasVisitor);
+}
+
+void
 FileCasStrategy::Flush()
 {
 	// Since we don't keep files open after writing there's nothing specific
@@ -353,8 +400,22 @@ FileCasStrategy::Flush()
 }
 
 void
-FileCasStrategy::Scrub()
+FileCasStrategy::Scrub(ScrubContext& Ctx)
 {
+	std::vector<IoHash> BadHashes;
+
+	IterateChunks([&](const IoHash& Hash, BasicFile& Payload) {
+		IoHashStream Hasher;
+		Payload.StreamFile([&](const void* Data, size_t Size) { Hasher.Append(Data, Size); });
+		IoHash ComputedHash = Hasher.GetHash();
+
+		if (ComputedHash != Hash)
+		{
+			BadHashes.push_back(Hash);
+		}
+	});
+
+	Ctx.ReportBadChunks(BadHashes);
 }
 
 void