Changeset: 7a4e53e8a05d for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=7a4e53e8a05d
Modified Files:
gdk/gdk_hash.c
monetdb5/modules/atoms/uuid.c
Branch: Jun2020
Log Message:
Don't make UUIDs go through a 32 bit bottleneck when calculating a hash.
diffs (73 lines):
diff --git a/gdk/gdk_hash.c b/gdk/gdk_hash.c
--- a/gdk/gdk_hash.c
+++ b/gdk/gdk_hash.c
@@ -97,7 +97,11 @@ HASHclear(Hash *h)
memset(h->Bckt, 0xFF, h->nbucket * h->width);
}
-#define HASH_VERSION 3
+#define HASH_VERSION 4
+/* this is only for the change of hash function of the UUID type; if
+ * HASH_VERSION is increased again from 4, the code associated with
+ * HASH_VERSION_NOUUID must be deleted */
+#define HASH_VERSION_NOUUID 3
#define HASH_HEADER_SIZE 7 /* nr of size_t fields in header */
static void
@@ -442,11 +446,21 @@ BATcheckhash(BAT *b)
struct stat st;
if (read(fd, hdata, sizeof(hdata)) ==
sizeof(hdata) &&
- hdata[0] == (
+ (hdata[0] == (
#ifdef PERSISTENTHASH
((size_t) 1 << 24) |
#endif
- HASH_VERSION) &&
+ HASH_VERSION)
+#ifdef HASH_VERSION_NOUUID
+ /* if not uuid, also allow
previous version */
+ || (hdata[0] == (
+#ifdef PERSISTENTHASH
+ ((size_t) 1 << 24) |
+#endif
+ HASH_VERSION_NOUUID) &&
+ strcmp(ATOMname(b->ttype),
"uuid") != 0)
+#endif
+ ) &&
hdata[1] > 0 &&
hdata[4] == (size_t) BATcount(b) &&
fstat(fd, &st) == 0 &&
diff --git a/monetdb5/modules/atoms/uuid.c b/monetdb5/modules/atoms/uuid.c
--- a/monetdb5/modules/atoms/uuid.c
+++ b/monetdb5/modules/atoms/uuid.c
@@ -319,17 +319,19 @@ BUN
UUIDhash(const void *v)
{
const uuid *u = (const uuid *) v;
- unsigned int u1, u2, u3, u4;
+ ulng u1, u2;
- u1 = (unsigned int) u->u[0] << 24 | (unsigned int) u->u[1] << 16 |
- (unsigned int) u->u[2] << 8 | (unsigned int) u->u[3];
- u2 = (unsigned int) u->u[4] << 24 | (unsigned int) u->u[5] << 16 |
- (unsigned int) u->u[6] << 8 | (unsigned int) u->u[7];
- u3 = (unsigned int) u->u[8] << 24 | (unsigned int) u->u[9] << 16 |
- (unsigned int) u->u[10] << 8 | (unsigned int) u->u[11];
- u4 = (unsigned int) u->u[12] << 24 | (unsigned int) u->u[13] << 16 |
- (unsigned int) u->u[14] << 8 | (unsigned int) u->u[15];
- return (BUN) mix_int(u1 ^ u2 ^ u3 ^ u4);
+ u1 = (ulng) u->u[0] << 56 | (ulng) u->u[1] << 48 |
+ (ulng) u->u[2] << 40 | (ulng) u->u[3] << 32 |
+ (ulng) u->u[4] << 24 | (ulng) u->u[5] << 16 |
+ (ulng) u->u[6] << 8 | (ulng) u->u[7];
+ u2 = (ulng) u->u[8] << 56 | (ulng) u->u[9] << 48 |
+ (ulng) u->u[10] << 40 | (ulng) u->u[11] << 32 |
+ (ulng) u->u[12] << 24 | (ulng) u->u[13] << 16 |
+ (ulng) u->u[14] << 8 | (ulng) u->u[15];
+ /* we're not using mix_hge since this we way get the same result
+ * on systems with and without 128 bit integer support */
+ return (BUN) (mix_lng(u1) ^ mix_lng(u2));
}
const uuid *
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
