[cfe-commits] r39056 - in /cfe/cfe/trunk: Lex/IdentifierTable.cpp include/clang/Lex/IdentifierTable.h

[sabre at cs.uiuc.edu]

Author: sabre
Date: Wed Jul 11 11:27:11 2007
New Revision: 39056

URL: http://llvm.org/viewvc/llvm-project?rev=39056&view=rev
Log:
Track the full (not mod the hash table size) hash value for each token.
This lets us find interesting properties of the hash distribution.

Modified:
    cfe/cfe/trunk/Lex/IdentifierTable.cpp
    cfe/cfe/trunk/include/clang/Lex/IdentifierTable.h

Modified: cfe/cfe/trunk/Lex/IdentifierTable.cpp
URL: http://llvm.org/viewvc/llvm-project/cfe/cfe/trunk/Lex/IdentifierTable.cpp?rev=39056&r1=39055&r2=39056&view=diff

==============================================================================
--- cfe/cfe/trunk/Lex/IdentifierTable.cpp (original)
+++ cfe/cfe/trunk/Lex/IdentifierTable.cpp Wed Jul 11 11:27:11 2007
@@ -179,7 +179,8 @@
                                      const char *NameEnd) {
   IdentifierBucket **TableArray = (IdentifierBucket**)TheTable;
 
-  unsigned Hash = HashString(NameStart, NameEnd) & (HASH_TABLE_SIZE-1);
+  unsigned FullHash = HashString(NameStart, NameEnd);
+  unsigned Hash = FullHash & (HASH_TABLE_SIZE-1);
   unsigned Length = NameEnd-NameStart;
   
   IdentifierBucket *IdentHead = TableArray[Hash];
@@ -218,6 +219,7 @@
   Identifier->TokInfo.IsPoisoned = false;
   Identifier->TokInfo.IsOtherTargetMacro = false;
   Identifier->TokInfo.FETokenInfo = 0;
+  Identifier->TokInfo.HashValue = FullHash;
 
   // Copy the string information.
   char *StrBuffer = (char*)(Identifier+1);
@@ -350,8 +352,19 @@
         MaxIdentifierLength = Id->TokInfo.getNameLength();
       ++NumIdentifiersInBucket;
     }
-    if (NumIdentifiersInBucket > MaxBucketLength) 
+    if (NumIdentifiersInBucket > MaxBucketLength) {
       MaxBucketLength = NumIdentifiersInBucket;
+      
+#if 0 // This code can be enabled to see (with -stats) a sample of some of the
+      // longest buckets in the hash table.  Useful for inspecting density of
+      // buckets etc.
+      std::cerr << "Bucket length " << MaxBucketLength << ":\n";
+      for (IdentifierBucket *Id = TableArray[i]; Id; Id = Id->Next) {
+        std::cerr << "  " << Id->TokInfo.getName() << " hash = "
+                  << Id->TokInfo.HashValue << "\n";
+      }
+#endif
+    }
     if (NumIdentifiersInBucket == 0)
       ++NumEmptyBuckets;
 

Modified: cfe/cfe/trunk/include/clang/Lex/IdentifierTable.h
URL: http://llvm.org/viewvc/llvm-project/cfe/cfe/trunk/include/clang/Lex/IdentifierTable.h?rev=39056&r1=39055&r2=39056&view=diff

==============================================================================
--- cfe/cfe/trunk/include/clang/Lex/IdentifierTable.h (original)
+++ cfe/cfe/trunk/include/clang/Lex/IdentifierTable.h Wed Jul 11 11:27:11 2007
@@ -30,7 +30,7 @@
 /// variable or function name).  The preprocessor keeps this information in a
 /// set, and all tok::identifier tokens have a pointer to one of these.  
 class IdentifierInfo {
-  unsigned NameLen;                // String that is the identifier.
+  unsigned NameLen;                // Length of the identifier string.
   MacroInfo *Macro;                // Set if this identifier is #define'd.
   tok::TokenKind TokenID      : 8; // Front-end token ID or tok::identifier.
   tok::PPKeywordKind PPID     : 5; // ID for preprocessor command like 'ifdef'.
@@ -39,6 +39,7 @@
   bool IsPoisoned             : 1; // True if identifier is poisoned.
   bool IsOtherTargetMacro     : 1; // True if ident is macro on another target.
   void *FETokenInfo;               // Managed by the language front-end.
+  unsigned HashValue;              // The full (non-truncated) hash value.
   friend class IdentifierTable;
 public:
   /// getName - Return the actual string for this identifier.  The length of