s3: Use jenkins hash for str_checksum, fix bug 8010

From the bugreport:

I have a folder with ~90 photos: IMG_XXXX.JPG where XXXX is a four digit
number, almost consecutive (photos from camera for one day).
Current implementation gives about 30 different checksums for this set of
files.

Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Wed Mar 16 01:15:41 CET 2011 on sn-devel-104

diff --git a/source3/lib/util.c b/source3/lib/util.c
index 81d2a78d3ce9620d63cf79bbc1e4a5f4d1f4b392..b6128feaf6591f3fe6a434ab6484d144cd27bf5f 100644 (file)
--- a/source3/lib/util.c
+++ b/source3/lib/util.c
@@ -1899,17 +1899,8 @@ const char *tab_depth(int level, int depth)
 
 int str_checksum(const char *s)
 {
-       int res = 0;
-       int c;
-       int i=0;
-
-       while(*s) {
-               c = *s;
-               res ^= (c << (i % 15)) ^ (c >> (15-(i%15)));
-               s++;
-               i++;
-       }
-       return(res);
+       TDB_DATA key = string_tdb_data(s);
+       return tdb_jenkins_hash(&key);
 }
 
 /*****************************************************************