X-Git-Url: https://mattmccutchen.net/rsync/rsync.git/blobdiff_plain/936fa865e1fd2ea0c97ef1a361f3f35a4153be75..df476bfcff992fb66765d338128dffcf58369722:/match.c

diff --git a/match.c b/match.c
index 8367d158..280e09fc 100644
--- a/match.c
+++ b/match.c
@@ -3,7 +3,7 @@
  *
  * Copyright (C) 1996 Andrew Tridgell
  * Copyright (C) 1996 Paul Mackerras
- * Copyright (C) 2003-2007 Wayne Davison
+ * Copyright (C) 2003-2008 Wayne Davison
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -39,29 +39,50 @@ static int total_matches;
 
 extern struct stats stats;
 
-#define TABLESIZE (1<<16)
+#define TRADITIONAL_TABLESIZE (1<<16)
 
+static uint32 tablesize;
 static int32 *hash_table;
 
 #define SUM2HASH2(s1,s2) (((s1) + (s2)) & 0xFFFF)
 #define SUM2HASH(sum) SUM2HASH2((sum)&0xFFFF,(sum)>>16)
 
+#define BIG_SUM2HASH(sum) ((sum)%tablesize)
+
 static void build_hash_table(struct sum_struct *s)
 {
+	static uint32 alloc_size;
 	int32 i;
 
-	if (!hash_table) {
-		hash_table = new_array(int32, TABLESIZE);
+	/* Dynamically calculate the hash table size so that the hash load
+	 * for big files is about 80%.  A number greater than the traditional
+	 * size must be odd or s2 will not be able to span the entire set. */
+	tablesize = (uint32)(s->count/8) * 10 + 11;
+	if (tablesize < TRADITIONAL_TABLESIZE)
+		tablesize = TRADITIONAL_TABLESIZE;
+	if (tablesize > alloc_size || tablesize < alloc_size - 16*1024) {
+		if (hash_table)
+			free(hash_table);
+		hash_table = new_array(int32, tablesize);
 		if (!hash_table)
 			out_of_memory("build_hash_table");
+		alloc_size = tablesize;
 	}
 
-	memset(hash_table, 0xFF, TABLESIZE * sizeof hash_table[0]);
+	memset(hash_table, 0xFF, tablesize * sizeof hash_table[0]);
 
-	for (i = 0; i < s->count; i++) {
-		uint32 t = SUM2HASH(s->sums[i].sum1);
-		s->sums[i].chain = hash_table[t];
-		hash_table[t] = i;
+	if (tablesize == TRADITIONAL_TABLESIZE) {
+		for (i = 0; i < s->count; i++) {
+			uint32 t = SUM2HASH(s->sums[i].sum1);
+			s->sums[i].chain = hash_table[t];
+			hash_table[t] = i;
+		}
+	} else {
+		for (i = 0; i < s->count; i++) {
+			uint32 t = BIG_SUM2HASH(s->sums[i].sum1);
+			s->sums[i].chain = hash_table[t];
+			hash_table[t] = i;
+		}
 	}
 }
 
@@ -164,11 +185,16 @@ static void hash_search(int f,struct sum_struct *s,
 				(double)offset, s2 & 0xFFFF, s1 & 0xFFFF);
 		}
 
-		i = hash_table[SUM2HASH2(s1,s2)];
-		if (i < 0)
-			goto null_hash;
+		if (tablesize == TRADITIONAL_TABLESIZE) {
+			if ((i = hash_table[SUM2HASH2(s1,s2)]) < 0)
+				goto null_hash;
+			sum = (s1 & 0xffff) | (s2 << 16);
+		} else {
+			sum = (s1 & 0xffff) | (s2 << 16);
+			if ((i = hash_table[BIG_SUM2HASH(sum)]) < 0)
+				goto null_hash;
+		}
 
-		sum = (s1 & 0xffff) | (s2 << 16);
 		hash_hits++;
 		do {
 			int32 l;