//
// The LLVM Compiler Infrastructure
//
-// This file was developed by Chris Lattner and is distributed under
-// the University of Illinois Open Source License. See LICENSE.TXT for details.
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
//===----------------------------------------------------------------------===//
#include "llvm/ADT/StringMap.h"
+#include "llvm/ADT/StringExtras.h"
#include <cassert>
using namespace llvm;
StringMapImpl::StringMapImpl(unsigned InitSize, unsigned itemSize) {
+ ItemSize = itemSize;
+
+ // If a size is specified, initialize the table with that many buckets.
+ if (InitSize) {
+ init(InitSize);
+ return;
+ }
+
+ // Otherwise, initialize it with zero buckets to avoid the allocation.
+ TheTable = 0;
+ NumBuckets = 0;
+ NumItems = 0;
+ NumTombstones = 0;
+}
+
+void StringMapImpl::init(unsigned InitSize) {
assert((InitSize & (InitSize-1)) == 0 &&
"Init Size must be a power of 2 or zero!");
- NumBuckets = InitSize ? InitSize : 512;
- ItemSize = itemSize;
+ NumBuckets = InitSize ? InitSize : 16;
NumItems = 0;
+ NumTombstones = 0;
- TheTable = new ItemBucket[NumBuckets+1]();
- memset(TheTable, 0, NumBuckets*sizeof(ItemBucket));
+ TheTable = (ItemBucket*)calloc(NumBuckets+1, sizeof(ItemBucket));
// Allocate one extra bucket, set it to look filled so the iterators stop at
// end.
}
-/// HashString - Compute a hash code for the specified string.
-///
-static unsigned HashString(const char *Start, const char *End) {
- // Bernstein hash function.
- unsigned int Result = 0;
- // TODO: investigate whether a modified bernstein hash function performs
- // better: http://eternallyconfuzzled.com/tuts/algorithms/jsw_tut_hashing.aspx
- // X*33+c -> X*33^c
- while (Start != End)
- Result = Result * 33 + *Start++;
- Result = Result + (Result >> 5);
- return Result;
-}
-
/// LookupBucketFor - Look up the bucket that the specified string should end
/// up in. If it already exists as a key in the map, the Item pointer for the
/// specified bucket will be non-null. Otherwise, it will be null. In either
/// case, the FullHashValue field of the bucket will be set to the hash value
/// of the string.
-unsigned StringMapImpl::LookupBucketFor(const char *NameStart,
- const char *NameEnd) {
+unsigned StringMapImpl::LookupBucketFor(StringRef Name) {
unsigned HTSize = NumBuckets;
- unsigned FullHashValue = HashString(NameStart, NameEnd);
+ if (HTSize == 0) { // Hash table unallocated so far?
+ init(16);
+ HTSize = NumBuckets;
+ }
+ unsigned FullHashValue = HashString(Name);
unsigned BucketNo = FullHashValue & (HTSize-1);
unsigned ProbeAmt = 1;
+ int FirstTombstone = -1;
while (1) {
ItemBucket &Bucket = TheTable[BucketNo];
StringMapEntryBase *BucketItem = Bucket.Item;
// If we found an empty bucket, this key isn't in the table yet, return it.
if (BucketItem == 0) {
+ // If we found a tombstone, we want to reuse the tombstone instead of an
+ // empty bucket. This reduces probing.
+ if (FirstTombstone != -1) {
+ TheTable[FirstTombstone].FullHashValue = FullHashValue;
+ return FirstTombstone;
+ }
+
Bucket.FullHashValue = FullHashValue;
return BucketNo;
}
- // If the full hash value matches, check deeply for a match. The common
- // case here is that we are only looking at the buckets (for item info
- // being non-null and for the full hash value) not at the items. This
- // is important for cache locality.
- if (Bucket.FullHashValue == FullHashValue) {
- // Do the comparison like this because NameStart isn't necessarily
+ if (BucketItem == getTombstoneVal()) {
+ // Skip over tombstones. However, remember the first one we see.
+ if (FirstTombstone == -1) FirstTombstone = BucketNo;
+ } else if (Bucket.FullHashValue == FullHashValue) {
+ // If the full hash value matches, check deeply for a match. The common
+ // case here is that we are only looking at the buckets (for item info
+ // being non-null and for the full hash value) not at the items. This
+ // is important for cache locality.
+
+ // Do the comparison like this because Name isn't necessarily
// null-terminated!
char *ItemStr = (char*)BucketItem+ItemSize;
- unsigned ItemStrLen = BucketItem->getKeyLength();
- if (unsigned(NameEnd-NameStart) == ItemStrLen &&
- memcmp(ItemStr, NameStart, ItemStrLen) == 0) {
+ if (Name == StringRef(ItemStr, BucketItem->getKeyLength())) {
// We found a match!
return BucketNo;
}
/// FindKey - Look up the bucket that contains the specified key. If it exists
/// in the map, return the bucket number of the key. Otherwise return -1.
/// This does not modify the map.
-int StringMapImpl::FindKey(const char *KeyStart, const char *KeyEnd) const {
+int StringMapImpl::FindKey(StringRef Key) const {
unsigned HTSize = NumBuckets;
- unsigned FullHashValue = HashString(KeyStart, KeyEnd);
+ if (HTSize == 0) return -1; // Really empty table?
+ unsigned FullHashValue = HashString(Key);
unsigned BucketNo = FullHashValue & (HTSize-1);
unsigned ProbeAmt = 1;
if (BucketItem == 0)
return -1;
- // If the full hash value matches, check deeply for a match. The common
- // case here is that we are only looking at the buckets (for item info
- // being non-null and for the full hash value) not at the items. This
- // is important for cache locality.
- if (Bucket.FullHashValue == FullHashValue) {
+ if (BucketItem == getTombstoneVal()) {
+ // Ignore tombstones.
+ } else if (Bucket.FullHashValue == FullHashValue) {
+ // If the full hash value matches, check deeply for a match. The common
+ // case here is that we are only looking at the buckets (for item info
+ // being non-null and for the full hash value) not at the items. This
+ // is important for cache locality.
+
// Do the comparison like this because NameStart isn't necessarily
// null-terminated!
char *ItemStr = (char*)BucketItem+ItemSize;
- unsigned ItemStrLen = BucketItem->getKeyLength();
- if (unsigned(KeyEnd-KeyStart) == ItemStrLen &&
- memcmp(ItemStr, KeyStart, ItemStrLen) == 0) {
+ if (Key == StringRef(ItemStr, BucketItem->getKeyLength())) {
// We found a match!
return BucketNo;
}
}
}
+/// RemoveKey - Remove the specified StringMapEntry from the table, but do not
+/// delete it. This aborts if the value isn't in the table.
+void StringMapImpl::RemoveKey(StringMapEntryBase *V) {
+ const char *VStr = (char*)V + ItemSize;
+ StringMapEntryBase *V2 = RemoveKey(StringRef(VStr, V->getKeyLength()));
+ V2 = V2;
+ assert(V == V2 && "Didn't find key?");
+}
+
+/// RemoveKey - Remove the StringMapEntry for the specified key from the
+/// table, returning it. If the key is not in the table, this returns null.
+StringMapEntryBase *StringMapImpl::RemoveKey(StringRef Key) {
+ int Bucket = FindKey(Key);
+ if (Bucket == -1) return 0;
+
+ StringMapEntryBase *Result = TheTable[Bucket].Item;
+ TheTable[Bucket].Item = getTombstoneVal();
+ --NumItems;
+ ++NumTombstones;
+ return Result;
+}
+
+
/// RehashTable - Grow the table, redistributing values into the buckets with
/// the appropriate mod-of-hashtable-size.
unsigned NewSize = NumBuckets*2;
// Allocate one extra bucket which will always be non-empty. This allows the
// iterators to stop at end.
- ItemBucket *NewTableArray = new ItemBucket[NewSize+1]();
- memset(NewTableArray, 0, NewSize*sizeof(ItemBucket));
+ ItemBucket *NewTableArray =(ItemBucket*)calloc(NewSize+1, sizeof(ItemBucket));
NewTableArray[NewSize].Item = (StringMapEntryBase*)2;
// Rehash all the items into their new buckets. Luckily :) we already have
// the hash values available, so we don't have to rehash any strings.
for (ItemBucket *IB = TheTable, *E = TheTable+NumBuckets; IB != E; ++IB) {
- if (IB->Item) {
+ if (IB->Item && IB->Item != getTombstoneVal()) {
// Fast case, bucket available.
unsigned FullHash = IB->FullHashValue;
unsigned NewBucket = FullHash & (NewSize-1);
continue;
}
+ // Otherwise probe for a spot.
unsigned ProbeSize = 1;
do {
NewBucket = (NewBucket + ProbeSize++) & (NewSize-1);
}
}
- delete[] TheTable;
+ free(TheTable);
TheTable = NewTableArray;
NumBuckets = NewSize;