Rewrite the logic for generating the path ffor cas storage

Author: ojhunt

llvm/lib/CAS/OnDiskGraphDB.cpp

@@ -63,7 +63,6 @@
 #include "llvm/Support/MemoryBuffer.h"
 #include "llvm/Support/Path.h"
 #include "llvm/Support/Process.h"
-#include "llvm/Support/xxhash.h"
 #include <optional>
 
 #if __has_include(<sys/mount.h>)
@@ -618,7 +617,6 @@ Error OnDiskGraphDB::TempFile::keep(const Twine &Name) {
   sys::fs::file_t File = sys::fs::convertFDToNativeFile(FD);
   if (std::error_code EC = sys::fs::closeFile(File))
     return errorCodeToError(EC);
-
   FD = -1;
 
   return errorCodeToError(RenameEC);
@@ -1270,24 +1268,49 @@ void OnDiskGraphDB::getStandalonePath(StringRef Suffix, const IndexProxy &I,
   PersistentPath.assign(RootPath.begin(), RootPath.end());
   if (TempPath)
     TempPath->assign(RootPath.begin(), RootPath.end());
-  SmallVector<char, 256> FileNameBuffer;
-  StringRef FileName =
-    (FilePrefix + Twine(I.Offset.get()) + Suffix).toStringRef(FileNameBuffer);
-
-  uint64_t FileNameHash = xxHash64(FileName);
-
-  // This is around 10 bits of entropy given we're using radix-36
-  static const unsigned IntermediateDirNameLength = 2;
-  static const char Radix36Chars[] = "0123456789abcdefghijklmnopqrstuvwxyz";
-  static_assert(sizeof(Radix36Chars) == 36 + /* null */ 1);
-  SmallString<IntermediateDirNameLength> IntermediateDirName;
-  for (unsigned Ch = 0; Ch < IntermediateDirNameLength; ++Ch) {
-    IntermediateDirName.push_back(Radix36Chars[FileNameHash % 36]);
-    FileNameHash /= 36;
+
+  static const unsigned IntermediateDirCount = 2;
+  static const unsigned EntropyPerIntermediateDir = 8;
+  static const unsigned CharactersPerDirName =
+    ((EntropyPerIntermediateDir + 15) & ~7) / 8;
+
+  static const unsigned TotalEntropyBits =
+    IntermediateDirCount * EntropyPerIntermediateDir;
+  static const unsigned HashByteCount = ((TotalEntropyBits + 7) & ~7) / 8;
+  static_assert(HashByteCount <= sizeof(uint32_t));
+
+  // We will never need 32 bits of entropy, so we just assert that now as it
+  // means we don't need to worry about overflows in the shifts below.
+  static_assert(TotalEntropyBits < 32);
+
+  // It would be nice if there was a way to statically assert the size of an
+  // ArrayRef.
+  assert(I.Hash.size() >= sizeof(uint32_t));
+
+  uint32_t Hash;
+  memcpy(&Hash, I.Hash.data(), sizeof(Hash));
+
+  for (unsigned IntermediateDir = 0; IntermediateDir < IntermediateDirCount; ++IntermediateDir) {
+    unsigned DirValue = Hash & ((1u << EntropyPerIntermediateDir) - 1);
+    Hash >>= EntropyPerIntermediateDir;
+    char DirNameCharacters[CharactersPerDirName];
+    for (unsigned CharIdx = 0; CharIdx < CharactersPerDirName; ++CharIdx) {
+      // Mod and divide for clarity, this gets lowered to bit operations as
+      // expected.
+      unsigned CharValue = DirValue % 16;
+      DirValue /= 16;
+      DirNameCharacters[CharIdx] =
+        hexdigit(CharValue, /*LowerCase=*/false);
+    }
+    sys::path::append(PersistentPath, StringRef(DirNameCharacters, CharactersPerDirName));
   }
-  sys::path::append(PersistentPath, IntermediateDirName, FileName);
-  if (TempPath)
-    sys::path::append(*TempPath, FileName);
+  auto AppendFileName = [&](Twine FileName) {
+    sys::path::append(PersistentPath, FileName);
+
+    if (TempPath)
+      sys::path::append(*TempPath, FileName);
+  };
+  AppendFileName(FilePrefix + Twine(I.Offset.get()) + Suffix);
 }
 
 OnDiskContent OnDiskGraphDB::getContentFromHandle(ObjectHandle OH) const {