ceph: make CRUSH hash function a bucket property
Make the integer hash function a property of the bucket it is used on. This
allows us to gracefully add support for new hash functions without starting
from scatch.
Signed-off-by: Sage Weil <sage@newdream.net>
diff --git a/fs/ceph/crush/crush.h b/fs/ceph/crush/crush.h
index 92c6b3c..dcd7e75 100644
--- a/fs/ceph/crush/crush.h
+++ b/fs/ceph/crush/crush.h
@@ -102,7 +102,8 @@
struct crush_bucket {
__s32 id; /* this'll be negative */
__u16 type; /* non-zero; type=0 is reserved for devices */
- __u16 alg; /* one of CRUSH_BUCKET_* */
+ __u8 alg; /* one of CRUSH_BUCKET_* */
+ __u8 hash; /* which hash function to use, CRUSH_HASH_* */
__u32 weight; /* 16-bit fixed point */
__u32 size; /* num items */
__s32 *items;
diff --git a/fs/ceph/crush/hash.c b/fs/ceph/crush/hash.c
index b438c5d..5873aed 100644
--- a/fs/ceph/crush/hash.c
+++ b/fs/ceph/crush/hash.c
@@ -1,5 +1,6 @@
#include <linux/types.h>
+#include "hash.h"
/*
* Robert Jenkins' function for mixing 32-bit values
@@ -20,7 +21,7 @@
#define crush_hash_seed 1315423911
-__u32 crush_hash32(__u32 a)
+static __u32 crush_hash32_rjenkins1(__u32 a)
{
__u32 hash = crush_hash_seed ^ a;
__u32 b = a;
@@ -31,7 +32,7 @@
return hash;
}
-__u32 crush_hash32_2(__u32 a, __u32 b)
+static __u32 crush_hash32_rjenkins1_2(__u32 a, __u32 b)
{
__u32 hash = crush_hash_seed ^ a ^ b;
__u32 x = 231232;
@@ -42,7 +43,7 @@
return hash;
}
-__u32 crush_hash32_3(__u32 a, __u32 b, __u32 c)
+static __u32 crush_hash32_rjenkins1_3(__u32 a, __u32 b, __u32 c)
{
__u32 hash = crush_hash_seed ^ a ^ b ^ c;
__u32 x = 231232;
@@ -55,7 +56,7 @@
return hash;
}
-__u32 crush_hash32_4(__u32 a, __u32 b, __u32 c, __u32 d)
+static __u32 crush_hash32_rjenkins1_4(__u32 a, __u32 b, __u32 c, __u32 d)
{
__u32 hash = crush_hash_seed ^ a ^ b ^ c ^ d;
__u32 x = 231232;
@@ -69,7 +70,8 @@
return hash;
}
-__u32 crush_hash32_5(__u32 a, __u32 b, __u32 c, __u32 d, __u32 e)
+static __u32 crush_hash32_rjenkins1_5(__u32 a, __u32 b, __u32 c, __u32 d,
+ __u32 e)
{
__u32 hash = crush_hash_seed ^ a ^ b ^ c ^ d ^ e;
__u32 x = 231232;
@@ -84,3 +86,64 @@
crush_hashmix(y, e, hash);
return hash;
}
+
+
+__u32 crush_hash32(int type, __u32 a)
+{
+ switch (type) {
+ case CRUSH_HASH_RJENKINS1:
+ return crush_hash32_rjenkins1(a);
+ default:
+ return 0;
+ }
+}
+
+__u32 crush_hash32_2(int type, __u32 a, __u32 b)
+{
+ switch (type) {
+ case CRUSH_HASH_RJENKINS1:
+ return crush_hash32_rjenkins1_2(a, b);
+ default:
+ return 0;
+ }
+}
+
+__u32 crush_hash32_3(int type, __u32 a, __u32 b, __u32 c)
+{
+ switch (type) {
+ case CRUSH_HASH_RJENKINS1:
+ return crush_hash32_rjenkins1_3(a, b, c);
+ default:
+ return 0;
+ }
+}
+
+__u32 crush_hash32_4(int type, __u32 a, __u32 b, __u32 c, __u32 d)
+{
+ switch (type) {
+ case CRUSH_HASH_RJENKINS1:
+ return crush_hash32_rjenkins1_4(a, b, c, d);
+ default:
+ return 0;
+ }
+}
+
+__u32 crush_hash32_5(int type, __u32 a, __u32 b, __u32 c, __u32 d, __u32 e)
+{
+ switch (type) {
+ case CRUSH_HASH_RJENKINS1:
+ return crush_hash32_rjenkins1_5(a, b, c, d, e);
+ default:
+ return 0;
+ }
+}
+
+const char *crush_hash_name(int type)
+{
+ switch (type) {
+ case CRUSH_HASH_RJENKINS1:
+ return "rjenkins1";
+ default:
+ return "unknown";
+ }
+}
diff --git a/fs/ceph/crush/hash.h b/fs/ceph/crush/hash.h
index 9ce89f8..ff48e110 100644
--- a/fs/ceph/crush/hash.h
+++ b/fs/ceph/crush/hash.h
@@ -1,12 +1,17 @@
#ifndef _CRUSH_HASH_H
#define _CRUSH_HASH_H
-extern __u32 crush_hash32(__u32 a);
-extern __u32 crush_hash32_2(__u32 a, __u32 b);
-extern __u32 crush_hash32_3(__u32 a, __u32 b, __u32 c);
-extern __u32 crush_hash32_4(__u32 a, __u32 b, __u32 c,
- __u32 d);
-extern __u32 crush_hash32_5(__u32 a, __u32 b, __u32 c,
- __u32 d, __u32 e);
+#define CRUSH_HASH_RJENKINS1 0
+
+#define CRUSH_HASH_DEFAULT CRUSH_HASH_RJENKINS1
+
+extern const char *crush_hash_name(int type);
+
+extern __u32 crush_hash32(int type, __u32 a);
+extern __u32 crush_hash32_2(int type, __u32 a, __u32 b);
+extern __u32 crush_hash32_3(int type, __u32 a, __u32 b, __u32 c);
+extern __u32 crush_hash32_4(int type, __u32 a, __u32 b, __u32 c, __u32 d);
+extern __u32 crush_hash32_5(int type, __u32 a, __u32 b, __u32 c, __u32 d,
+ __u32 e);
#endif
diff --git a/fs/ceph/crush/mapper.c b/fs/ceph/crush/mapper.c
index 54f3f40..2523d44 100644
--- a/fs/ceph/crush/mapper.c
+++ b/fs/ceph/crush/mapper.c
@@ -78,7 +78,7 @@
/* optimize common r=0 case */
if (pr == 0) {
- s = crush_hash32_3(x, bucket->id, 0) %
+ s = crush_hash32_3(bucket->hash, x, bucket->id, 0) %
bucket->size;
bucket->perm[0] = s;
bucket->perm_n = 0xffff; /* magic value, see below */
@@ -103,7 +103,7 @@
unsigned p = bucket->perm_n;
/* no point in swapping the final entry */
if (p < bucket->size - 1) {
- i = crush_hash32_3(x, bucket->id, p) %
+ i = crush_hash32_3(bucket->hash, x, bucket->id, p) %
(bucket->size - p);
if (i) {
unsigned t = bucket->perm[p + i];
@@ -138,8 +138,8 @@
int i;
for (i = bucket->h.size-1; i >= 0; i--) {
- __u64 w = crush_hash32_4(x, bucket->h.items[i], r,
- bucket->h.id);
+ __u64 w = crush_hash32_4(bucket->h.hash,x, bucket->h.items[i],
+ r, bucket->h.id);
w &= 0xffff;
dprintk("list_choose i=%d x=%d r=%d item %d weight %x "
"sw %x rand %llx",
@@ -198,7 +198,8 @@
while (!terminal(n)) {
/* pick point in [0, w) */
w = bucket->node_weights[n];
- t = (__u64)crush_hash32_4(x, n, r, bucket->h.id) * (__u64)w;
+ t = (__u64)crush_hash32_4(bucket->h.hash, x, n, r,
+ bucket->h.id) * (__u64)w;
t = t >> 32;
/* descend to the left or right? */
@@ -224,7 +225,7 @@
__u64 draw;
for (i = 0; i < bucket->h.size; i++) {
- draw = crush_hash32_3(x, bucket->h.items[i], r);
+ draw = crush_hash32_3(bucket->h.hash, x, bucket->h.items[i], r);
draw &= 0xffff;
draw *= bucket->straws[i];
if (i == 0 || draw > high_draw) {
@@ -267,7 +268,8 @@
return 0;
if (weight[item] == 0)
return 1;
- if ((crush_hash32_2(x, item) & 0xffff) < weight[item])
+ if ((crush_hash32_2(CRUSH_HASH_RJENKINS1, x, item) & 0xffff)
+ < weight[item])
return 0;
return 1;
}
diff --git a/fs/ceph/osdmap.c b/fs/ceph/osdmap.c
index 6847827..8c994c7 100644
--- a/fs/ceph/osdmap.c
+++ b/fs/ceph/osdmap.c
@@ -210,7 +210,8 @@
ceph_decode_need(p, end, 4*sizeof(u32), bad);
b->id = ceph_decode_32(p);
b->type = ceph_decode_16(p);
- b->alg = ceph_decode_16(p);
+ b->alg = ceph_decode_8(p);
+ b->hash = ceph_decode_8(p);
b->weight = ceph_decode_32(p);
b->size = ceph_decode_32(p);