[PDB] Move PDB functions to a separate file.

[ruiu]

Summary

We are going to use the hash functions from TPI streams.

[zturner]

There is also SigForPbCb in Microsoft headers. I haven't ported this one over yet, and I don't know the details about when each particular function is used.

If possible, could you add a comment over each function in Hash.h that briefly documents the situations in which that particular is used? i.e. what types of records, Tpi Stream version, etc. We can improve the comments as we learn more, but it would be nice for someone to look at the header file and have a general idea of why there are so many different hash functions.

[ruiu]

Added brief comments about where these hash functions are used.
But I think it is better to describe how they are used where they
are used (instead of where the hash functions are defined,) so
it's not that long comments.

[majnemer]

lib/DebugInfo/PDB/Raw/Hash.cpp

21 ↗

(On Diff #60057)

size_t ?

30 ↗

(On Diff #60057)

Isn't this just Size % 4?

[ruiu]

lib/DebugInfo/PDB/Raw/Hash.cpp

22 ↗

(On Diff #60065)

It shouldn't be that large as Str always points to a string in PDB (in which all streams are at most 2^32 bytes.)

31 ↗

(On Diff #60065)

Done.

[ruiu]

• address review comments.

[christof]

Hi Rui.

This patch breaks the LLVM build on Windows. For example:
http://lab.llvm.org:8011/builders/clang-x64-ninja-win7/builds/12912

Seems the new Hash.cpp and Hash.h need <cstdint>.

Regards,
Christof

[aaboud]

llvm/trunk/include/llvm/DebugInfo/PDB/Raw/Hash.h

14

This patch broke the build, please add this line to fix it:
#include <stdint.h>

[ruiu]

I'm sorry about the breakage. Hans fixed it for me in r272269.

llvm/trunk/include/llvm/DebugInfo/PDB/Raw/Hash.h

+//===- Hash.h - PDB hash functions ------------------------------*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_DEBUGINFO_PDB_RAW_HASH_H
+#define LLVM_DEBUGINFO_PDB_RAW_HASH_H
+
+#include "llvm/ADT/StringRef.h"
+
+namespace llvm {
+namespace pdb {
+uint32_t HashStringV1(StringRef Str);
+uint32_t HashStringV2(StringRef Str);
+}
+}
+
+#endif

llvm/trunk/lib/DebugInfo/PDB/CMakeLists.txt

add_pdb_impl_folder(DIA
     )
 
     set(LIBPDB_ADDITIONAL_HEADER_DIRS "${LLVM_MAIN_INCLUDE_DIR}/llvm/DebugInfo/PDB/DIA")
 endif()
 
 add_pdb_impl_folder(Raw
   Raw/DbiStream.cpp
   Raw/EnumTables.cpp
+  Raw/Hash.cpp
   Raw/IndexedStreamData.cpp
   Raw/InfoStream.cpp
   Raw/MappedBlockStream.cpp
   Raw/ModInfo.cpp
   Raw/ModStream.cpp
   Raw/NameHashTable.cpp
   Raw/NameMap.cpp
   Raw/PDBFile.cpp

llvm/trunk/lib/DebugInfo/PDB/Raw/Hash.cpp

+//===- Hash.cpp - PDB Hash Functions --------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/DebugInfo/PDB/Raw/Hash.h"
+
+#include "llvm/ADT/ArrayRef.h"
+#include "llvm/Support/Endian.h"
+
+using namespace llvm;
+using namespace llvm::support;
+
+// Corresponds to `Hasher::lhashPbCb` in PDB/include/misc.h.
+// Used for name hash table and TPI/IPI hashes.
+uint32_t pdb::HashStringV1(StringRef Str) {
+  uint32_t Result = 0;
+  uint32_t Size = Str.size();
+
+  ArrayRef<ulittle32_t> Longs(reinterpret_cast<const ulittle32_t *>(Str.data()),
+                              Size / 4);
+
+  for (auto Value : Longs)
+    Result ^= Value;
+
+  const uint8_t *Remainder = reinterpret_cast<const uint8_t *>(Longs.end());
+  uint32_t RemainderSize = Size % 4;
+
+  // Maximum of 3 bytes left.  Hash a 2 byte word if possible, then hash the
+  // possibly remaining 1 byte.
+  if (RemainderSize >= 2) {
+    uint16_t Value = *reinterpret_cast<const ulittle16_t *>(Remainder);
+    Result ^= static_cast<uint32_t>(Value);
+    Remainder += 2;
+    RemainderSize -= 2;
+  }
+
+  // hash possible odd byte
+  if (RemainderSize == 1) {
+    Result ^= *(Remainder++);
+  }
+
+  const uint32_t toLowerMask = 0x20202020;
+  Result |= toLowerMask;
+  Result ^= (Result >> 11);
+
+  return Result ^ (Result >> 16);
+}
+
+// Corresponds to `HasherV2::HashULONG` in PDB/include/misc.h.
+// Used for name hash table.
+uint32_t pdb::HashStringV2(StringRef Str) {
+  uint32_t Hash = 0xb170a1bf;
+
+  ArrayRef<char> Buffer(Str.begin(), Str.end());
+
+  ArrayRef<ulittle32_t> Items(
+      reinterpret_cast<const ulittle32_t *>(Buffer.data()),
+      Buffer.size() / sizeof(ulittle32_t));
+  for (ulittle32_t Item : Items) {
+    Hash += Item;
+    Hash += (Hash << 10);
+    Hash ^= (Hash >> 6);
+  }
+  Buffer = Buffer.slice(Items.size() * sizeof(ulittle32_t));
+  for (uint8_t Item : Buffer) {
+    Hash += Item;
+    Hash += (Hash << 10);
+    Hash ^= (Hash >> 6);
+  }
+
+  return Hash * 1664525L + 1013904223L;
+}

llvm/trunk/lib/DebugInfo/PDB/Raw/NameHashTable.cpp

 //===- NameHashTable.cpp - PDB Name Hash Table ------------------*- C++ -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 
 #include "llvm/DebugInfo/PDB/Raw/NameHashTable.h"
 
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/DebugInfo/CodeView/StreamReader.h"
+#include "llvm/DebugInfo/PDB/Raw/Hash.h"
 #include "llvm/DebugInfo/PDB/Raw/RawError.h"
 #include "llvm/Support/Endian.h"
 
 using namespace llvm;
 using namespace llvm::support;
 using namespace llvm::pdb;
 
-// Corresponds to `Hasher::lhashPbCb` in PDB/include/misc.h.
-static inline uint32_t HashStringV1(StringRef Str) {
-  uint32_t Result = 0;
-  uint32_t Size = Str.size();
-
-  ArrayRef<ulittle32_t> Longs(reinterpret_cast<const ulittle32_t *>(Str.data()),
-                              Size / 4);
-
-  for (auto Value : Longs)
-    Result ^= Value;
-
-  const uint8_t *Remainder = reinterpret_cast<const uint8_t *>(Longs.end());
-  uint32_t RemainderSize = Size - Longs.size() * 4;
-
-  // Maximum of 3 bytes left.  Hash a 2 byte word if possible, then hash the
-  // possibly remaining 1 byte.
-  if (RemainderSize >= 2) {
-    uint16_t Value = *reinterpret_cast<const ulittle16_t *>(Remainder);
-    Result ^= static_cast<uint32_t>(Value);
-    Remainder += 2;
-    RemainderSize -= 2;
-  }
-
-  // hash possible odd byte
-  if (RemainderSize == 1) {
-    Result ^= *(Remainder++);
-  }
-
-  const uint32_t toLowerMask = 0x20202020;
-  Result |= toLowerMask;
-  Result ^= (Result >> 11);
-
-  return Result ^ (Result >> 16);
-}
-
-// Corresponds to `HasherV2::HashULONG` in PDB/include/misc.h.
-static inline uint32_t HashStringV2(StringRef Str) {
-  uint32_t Hash = 0xb170a1bf;
-
-  ArrayRef<char> Buffer(Str.begin(), Str.end());
-
-  ArrayRef<ulittle32_t> Items(
-      reinterpret_cast<const ulittle32_t *>(Buffer.data()),
-      Buffer.size() / sizeof(ulittle32_t));
-  for (ulittle32_t Item : Items) {
-    Hash += Item;
-    Hash += (Hash << 10);
-    Hash ^= (Hash >> 6);
-  }
-  Buffer = Buffer.slice(Items.size() * sizeof(ulittle32_t));
-  for (uint8_t Item : Buffer) {
-    Hash += Item;
-    Hash += (Hash << 10);
-    Hash ^= (Hash >> 6);
-  }
-
-  return Hash * 1664525L + 1013904223L;
-}
-
 NameHashTable::NameHashTable() : Signature(0), HashVersion(0), NameCount(0) {}
 
 Error NameHashTable::load(codeview::StreamReader &Stream) {
   struct Header {
     support::ulittle32_t Signature;
     support::ulittle32_t HashVersion;
     support::ulittle32_t ByteSize;
   };