contrib/xxhash: Add xxhash library

xxhash is a faster non-cryptographic hash.
https://github.com/Cyan4973/xxHash

Release Taken: "xxHash v0.6.2"
--------------

Files added:
  contrib/xxhash/xxhash.c
  contrib/xxhash/xxhash.h
  contrib/xxhash/xxhsum.c

Modifications to source:
------------------------
Following functions and data types got 'GF_' prefix
as below to avoid any form of name collisions in future.

    ---- Functions ----
    GF_XXH_versionNumber
    GF_XXH32
    GF_XXH32_createState
    GF_XXH32_freeState
    GF_XXH32_copyState
    GF_XXH32_reset
    GF_XXH32_update
    GF_XXH32_digest
    GF_XXH32_canonicalFromHash
    GF_XXH32_hashFromCanonical
    GF_XXH64
    GF_XXH64_createState
    GF_XXH64_freeState
    GF_XXH64_copyState
    GF_XXH64_reset
    GF_XXH64_update
    GF_XXH64_digest
    GF_XXH64_canonicalFromHash
    GF_XXH64_hashFromCanonical

    ---- Data Types ----
    GF_XXH_errorcode
    GF_XXH32_state_t*
    GF_XXH32_canonical_t*
    GF_XXH32_hash_t
    GF_XXH64_state_t*
    GF_XXH64_canonical_t*
    GF_XXH64_hash_t

It is linked with libglusterfs.so. A wrapper
funtion is also added for the easy usage in
common-utils.c.

xxhash can be used for the all the usecases where
a faster non-cryptographic hash is required.
gfid to path infra would be using this for now.

NOTE:
----
The gluster coding guidelines check is ignored
as maintaining it further would be difficult.

Updates: #253
Change-Id: Ib143f90d91d4ee99864a10246d5983e92900173b
Signed-off-by: Kotresh HR <khiremat@redhat.com>
Reviewed-on: https://review.gluster.org/17641
Smoke: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Amar Tumballi <amarts@redhat.com>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
Reviewed-by: Niels de Vos <ndevos@redhat.com>

3 files changed, 27 insertions, 1 deletions

diff --git a/libglusterfs/src/Makefile.am b/libglusterfs/src/Makefile.am
index 52b73166ddb..768fe042a10 100644
--- a/libglusterfs/src/Makefile.am
+++ b/libglusterfs/src/Makefile.am
@@ -9,7 +9,8 @@ libglusterfs_la_CPPFLAGS = $(GF_CPPFLAGS) -D__USE_FILE_OFFSET64 \
 	-I$(top_srcdir)/rpc/xdr/src/ -I$(top_builddir)/rpc/xdr/src/ \
 	-I$(top_srcdir)/rpc/rpc-lib/src/ -I$(CONTRIBDIR)/rbtree \
 	-I$(CONTRIBDIR)/libexecinfo ${ARGP_STANDALONE_CPPFLAGS} \
-	-DSBIN_DIR=\"$(sbindir)\" -I$(CONTRIBDIR)/timer-wheel
+	-DSBIN_DIR=\"$(sbindir)\" -I$(CONTRIBDIR)/timer-wheel \
+	-I$(CONTRIBDIR)/xxhash
 
 libglusterfs_la_LIBADD = @LEXLIB@ $(ZLIB_LIBS) $(MATH_LIB) $(UUID_LIBS)
 libglusterfs_la_LDFLAGS = -version-info $(LIBGLUSTERFS_LT_VERSION)
@@ -33,6 +34,7 @@ libglusterfs_la_SOURCES = dict.c xlator.c logging.c \
 	$(CONTRIBDIR)/libexecinfo/execinfo.c quota-common-utils.c rot-buffs.c \
 	$(CONTRIBDIR)/timer-wheel/timer-wheel.c \
 	$(CONTRIBDIR)/timer-wheel/find_last_bit.c default-args.c locking.c \
+	$(CONTRIBDIR)/xxhash/xxhash.c \
 	compound-fop-utils.c throttle-tbf.c
 
 nodist_libglusterfs_la_SOURCES = y.tab.c graph.lex.c defaults.c
@@ -62,6 +64,7 @@ noinst_HEADERS = unittest/unittest.h \
 	$(CONTRIBDIR)/mount/mntent_compat.h \
 	$(CONTRIBDIR)/libexecinfo/execinfo_compat.h \
 	$(CONTRIBDIR)/timer-wheel/timer-wheel.h \
+	$(CONTRIBDIR)/xxhash/xxhash.h \
 	tier-ctr-interface.h
 
 if !HAVE_LIBUUID
diff --git a/libglusterfs/src/common-utils.c b/libglusterfs/src/common-utils.c
index 65557fefba2..5015d9666b6 100644
--- a/libglusterfs/src/common-utils.c
+++ b/libglusterfs/src/common-utils.c
@@ -49,6 +49,7 @@
 #include "lkowner.h"
 #include "syscall.h"
 #include "cli1-xdr.h"
+#include "xxhash.h"
 #include <ifaddrs.h>
 #include "libglusterfs-messages.h"
 
@@ -82,6 +83,23 @@ md5_wrapper(const unsigned char *data, size_t len, char *md5)
                 snprintf(md5 + i * 2, lim-i*2, "%02x", scratch[i]);
 }
 
+void
+gf_xxh64_wrapper(const unsigned char *data, size_t len, unsigned long long seed,
+                 char *xxh64)
+{
+        unsigned short         i      = 0;
+        unsigned short         lim    = GF_XXH64_DIGEST_LENGTH*2+1;
+        GF_XXH64_hash_t        hash   = 0;
+        GF_XXH64_canonical_t   c_hash = {{0,},};
+        const uint8_t         *p      = (const uint8_t *) &c_hash;
+
+        hash = GF_XXH64(data, len, seed);
+        GF_XXH64_canonicalFromHash(&c_hash, hash);
+
+        for (i = 0; i < GF_XXH64_DIGEST_LENGTH; i++)
+                snprintf(xxh64 + i * 2, lim-i*2, "%02x", p[i]);
+}
+
 /* works similar to mkdir(1) -p.
  */
 int
diff --git a/libglusterfs/src/common-utils.h b/libglusterfs/src/common-utils.h
index 6bd24179592..c3c5ec77350 100644
--- a/libglusterfs/src/common-utils.h
+++ b/libglusterfs/src/common-utils.h
@@ -113,6 +113,9 @@ void trap (void);
 /* Default value of signing waiting time to sign a file for bitrot */
 #define SIGNING_TIMEOUT "120"
 
+/* xxhash */
+#define GF_XXH64_DIGEST_LENGTH 8
+
 /* Shard */
 #define GF_XATTR_SHARD_FILE_SIZE  "trusted.glusterfs.shard.file-size"
 #define SHARD_ROOT_GFID "be318638-e8a0-4c6d-977d-7a937aa84806"
@@ -823,6 +826,8 @@ int gf_get_hostname_from_ip (char *client_ip, char **hostname);
 gf_boolean_t gf_is_local_addr (char *hostname);
 gf_boolean_t gf_is_same_address (char *host1, char *host2);
 void md5_wrapper(const unsigned char *data, size_t len, char *md5);
+void gf_xxh64_wrapper(const unsigned char *data, size_t len,
+                      unsigned long long seed, char *xxh64);
 int gf_set_timestamp  (const char *src, const char* dest);
 
 int gf_thread_create (pthread_t *thread, const pthread_attr_t *attr,