Merge pull request #1465 from teg/siphash24

hashmap/siphash24: refactor hash functions

4 files changed, 145 insertions, 81 deletions

diff --git a/src/basic/hashmap.c b/src/basic/hashmap.c
index 7d2a4160c6..3e17ed30df 100644
--- a/src/basic/hashmap.c
+++ b/src/basic/hashmap.c
@@ -276,10 +276,8 @@ static const struct hashmap_type_info hashmap_type_info[_HASHMAP_TYPE_MAX] = {
         },
 };
 
-unsigned long string_hash_func(const void *p, const uint8_t hash_key[HASH_KEY_SIZE]) {
-        uint64_t u;
-        siphash24((uint8_t*) &u, p, strlen(p), hash_key);
-        return (unsigned long) u;
+void string_hash_func(const void *p, struct siphash *state) {
+        siphash24_compress(p, strlen(p) + 1, state);
 }
 
 int string_compare_func(const void *a, const void *b) {
@@ -291,10 +289,8 @@ const struct hash_ops string_hash_ops = {
         .compare = string_compare_func
 };
 
-unsigned long trivial_hash_func(const void *p, const uint8_t hash_key[HASH_KEY_SIZE]) {
-        uint64_t u;
-        siphash24((uint8_t*) &u, &p, sizeof(p), hash_key);
-        return (unsigned long) u;
+void trivial_hash_func(const void *p, struct siphash *state) {
+        siphash24_compress(&p, sizeof(p), state);
 }
 
 int trivial_compare_func(const void *a, const void *b) {
@@ -306,10 +302,8 @@ const struct hash_ops trivial_hash_ops = {
         .compare = trivial_compare_func
 };
 
-unsigned long uint64_hash_func(const void *p, const uint8_t hash_key[HASH_KEY_SIZE]) {
-        uint64_t u;
-        siphash24((uint8_t*) &u, p, sizeof(uint64_t), hash_key);
-        return (unsigned long) u;
+void uint64_hash_func(const void *p, struct siphash *state) {
+        siphash24_compress(p, sizeof(uint64_t), state);
 }
 
 int uint64_compare_func(const void *_a, const void *_b) {
@@ -325,10 +319,8 @@ const struct hash_ops uint64_hash_ops = {
 };
 
 #if SIZEOF_DEV_T != 8
-unsigned long devt_hash_func(const void *p, const uint8_t hash_key[HASH_KEY_SIZE]) {
-        uint64_t u;
-        siphash24((uint8_t*) &u, p, sizeof(dev_t), hash_key);
-        return (unsigned long) u;
+void devt_hash_func(const void *p, struct siphash *state) {
+        siphash24_compress(p, sizeof(dev_t), state);
 }
 
 int devt_compare_func(const void *_a, const void *_b) {
@@ -379,7 +371,13 @@ static uint8_t *hash_key(HashmapBase *h) {
 }
 
 static unsigned base_bucket_hash(HashmapBase *h, const void *p) {
-        return (unsigned) (h->hash_ops->hash(p, hash_key(h)) % n_buckets(h));
+        struct siphash state;
+
+        siphash_init(&state, hash_key(h));
+
+        h->hash_ops->hash(p, &state);
+
+        return (unsigned) (siphash24_finalize(&state) % n_buckets(h));
 }
 #define bucket_hash(h, p) base_bucket_hash(HASHMAP_BASE(h), p)
 
diff --git a/src/basic/hashmap.h b/src/basic/hashmap.h
index 2af23024de..ed6a092d82 100644
--- a/src/basic/hashmap.h
+++ b/src/basic/hashmap.h
@@ -25,6 +25,7 @@
 #include <stdbool.h>
 
 #include "macro.h"
+#include "siphash24.h"
 #include "util.h"
 
 /*
@@ -67,7 +68,7 @@ typedef struct {
 #define _IDX_ITERATOR_FIRST (UINT_MAX - 1)
 #define ITERATOR_FIRST ((Iterator) { .idx = _IDX_ITERATOR_FIRST, .next_key = NULL })
 
-typedef unsigned long (*hash_func_t)(const void *p, const uint8_t hash_key[HASH_KEY_SIZE]);
+typedef void (*hash_func_t)(const void *p, struct siphash *state);
 typedef int (*compare_func_t)(const void *a, const void *b);
 
 struct hash_ops {
@@ -75,28 +76,28 @@ struct hash_ops {
         compare_func_t compare;
 };
 
-unsigned long string_hash_func(const void *p, const uint8_t hash_key[HASH_KEY_SIZE]) _pure_;
+void string_hash_func(const void *p, struct siphash *state);
 int string_compare_func(const void *a, const void *b) _pure_;
 extern const struct hash_ops string_hash_ops;
 
 /* This will compare the passed pointers directly, and will not
  * dereference them. This is hence not useful for strings or
  * suchlike. */
-unsigned long trivial_hash_func(const void *p, const uint8_t hash_key[HASH_KEY_SIZE]) _pure_;
+void trivial_hash_func(const void *p, struct siphash *state);
 int trivial_compare_func(const void *a, const void *b) _const_;
 extern const struct hash_ops trivial_hash_ops;
 
 /* 32bit values we can always just embedd in the pointer itself, but
  * in order to support 32bit archs we need store 64bit values
  * indirectly, since they don't fit in a pointer. */
-unsigned long uint64_hash_func(const void *p, const uint8_t hash_key[HASH_KEY_SIZE]) _pure_;
+void uint64_hash_func(const void *p, struct siphash *state);
 int uint64_compare_func(const void *a, const void *b) _pure_;
 extern const struct hash_ops uint64_hash_ops;
 
 /* On some archs dev_t is 32bit, and on others 64bit. And sometimes
  * it's 64bit on 32bit archs, and sometimes 32bit on 64bit archs. Yuck! */
 #if SIZEOF_DEV_T != 8
-unsigned long devt_hash_func(const void *p, const uint8_t hash_key[HASH_KEY_SIZE]) _pure_;
+void devt_hash_func(const void *p, struct siphash *state) _pure_;
 int devt_compare_func(const void *a, const void *b) _pure_;
 extern const struct hash_ops devt_hash_ops = {
         .hash = devt_hash_func,
diff --git a/src/basic/siphash24.c b/src/basic/siphash24.c
index f68bd283a1..308e4230c5 100644
--- a/src/basic/siphash24.c
+++ b/src/basic/siphash24.c
@@ -44,92 +44,144 @@ typedef uint8_t u8;
    ((u64)((p)[6]) << 48) | \
    ((u64)((p)[7]) << 56))
 
-#define SIPROUND            \
+#define SIPROUND(state)         \
   do {              \
-    v0 += v1; v1=ROTL(v1,13); v1 ^= v0; v0=ROTL(v0,32); \
-    v2 += v3; v3=ROTL(v3,16); v3 ^= v2;     \
-    v0 += v3; v3=ROTL(v3,21); v3 ^= v0;     \
-    v2 += v1; v1=ROTL(v1,17); v1 ^= v2; v2=ROTL(v2,32); \
+    (state)->v0 += (state)->v1; (state)->v1=ROTL((state)->v1,13); (state)->v1 ^= (state)->v0; (state)->v0=ROTL((state)->v0,32); \
+    (state)->v2 += (state)->v3; (state)->v3=ROTL((state)->v3,16); (state)->v3 ^= (state)->v2;     \
+    (state)->v0 += (state)->v3; (state)->v3=ROTL((state)->v3,21); (state)->v3 ^= (state)->v0;     \
+    (state)->v2 += (state)->v1; (state)->v1=ROTL((state)->v1,17); (state)->v1 ^= (state)->v2; (state)->v2=ROTL((state)->v2,32); \
   } while(0)
 
-/* SipHash-2-4 */
-void siphash24(uint8_t out[8], const void *_in, size_t inlen, const uint8_t k[16])
-{
+void siphash_init(struct siphash *state, const uint8_t k[16]) {
+  u64 k0, k1;
+
+  k0 = U8TO64_LE( k );
+  k1 = U8TO64_LE( k + 8 );
+
   /* "somepseudorandomlygeneratedbytes" */
-  u64 v0 = 0x736f6d6570736575ULL;
-  u64 v1 = 0x646f72616e646f6dULL;
-  u64 v2 = 0x6c7967656e657261ULL;
-  u64 v3 = 0x7465646279746573ULL;
-  u64 b;
-  u64 k0 = U8TO64_LE( k );
-  u64 k1 = U8TO64_LE( k + 8 );
+  state->v0 = 0x736f6d6570736575ULL ^ k0;
+  state->v1 = 0x646f72616e646f6dULL ^ k1;
+  state->v2 = 0x6c7967656e657261ULL ^ k0;
+  state->v3 = 0x7465646279746573ULL ^ k1;
+  state->padding = 0;
+  state->inlen = 0;
+}
+
+void siphash24_compress(const void *_in, size_t inlen, struct siphash *state) {
   u64 m;
   const u8 *in = _in;
-  const u8 *end = in + inlen - ( inlen % sizeof( u64 ) );
-  const int left = inlen & 7;
-  b = ( ( u64 )inlen ) << 56;
-  v3 ^= k1;
-  v2 ^= k0;
-  v1 ^= k1;
-  v0 ^= k0;
-
-  for ( ; in != end; in += 8 )
+  const u8 *end = in + inlen;
+  int left = state->inlen & 7;
+
+  /* update total length */
+  state->inlen += inlen;
+
+  /* if padding exists, fill it out */
+  if (left > 0) {
+    for ( ; in < end && left < 8; in ++, left ++ )
+      state->padding |= ( ( u64 )*in ) << (left * 8);
+
+    if (in == end && left < 8)
+      /* we did not have enough input to fill out the padding completely */
+      return;
+
+#ifdef DEBUG
+    printf( "(%3d) v0 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v0 >> 32 ), ( u32 )state->v0 );
+    printf( "(%3d) v1 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v1 >> 32 ), ( u32 )state->v1 );
+    printf( "(%3d) v2 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v2 >> 32 ), ( u32 )state->v2 );
+    printf( "(%3d) v3 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v3 >> 32 ), ( u32 )state->v3 );
+    printf( "(%3d) compress padding %08x %08x\n", ( int )state->inlen, ( u32 )( state->padding >> 32 ), ( u32 )state->padding );
+#endif
+    state->v3 ^= state->padding;
+    SIPROUND(state);
+    SIPROUND(state);
+    state->v0 ^= state->padding;
+
+    state->padding = 0;
+  }
+
+  end -= ( state->inlen % sizeof (u64) );
+
+  for ( ; in < end; in += 8 )
   {
     m = U8TO64_LE( in );
 #ifdef DEBUG
-    printf( "(%3d) v0 %08x %08x\n", ( int )inlen, ( u32 )( v0 >> 32 ), ( u32 )v0 );
-    printf( "(%3d) v1 %08x %08x\n", ( int )inlen, ( u32 )( v1 >> 32 ), ( u32 )v1 );
-    printf( "(%3d) v2 %08x %08x\n", ( int )inlen, ( u32 )( v2 >> 32 ), ( u32 )v2 );
-    printf( "(%3d) v3 %08x %08x\n", ( int )inlen, ( u32 )( v3 >> 32 ), ( u32 )v3 );
-    printf( "(%3d) compress %08x %08x\n", ( int )inlen, ( u32 )( m >> 32 ), ( u32 )m );
+    printf( "(%3d) v0 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v0 >> 32 ), ( u32 )state->v0 );
+    printf( "(%3d) v1 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v1 >> 32 ), ( u32 )state->v1 );
+    printf( "(%3d) v2 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v2 >> 32 ), ( u32 )state->v2 );
+    printf( "(%3d) v3 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v3 >> 32 ), ( u32 )state->v3 );
+    printf( "(%3d) compress %08x %08x\n", ( int )state->inlen, ( u32 )( m >> 32 ), ( u32 )m );
 #endif
-    v3 ^= m;
-    SIPROUND;
-    SIPROUND;
-    v0 ^= m;
+    state->v3 ^= m;
+    SIPROUND(state);
+    SIPROUND(state);
+    state->v0 ^= m;
   }
 
+  left = state->inlen & 7;
+
   switch( left )
   {
-  case 7: b |= ( ( u64 )in[ 6] )  << 48;
+  case 7: state->padding |= ( ( u64 )in[ 6] )  << 48;
 
-  case 6: b |= ( ( u64 )in[ 5] )  << 40;
+  case 6: state->padding |= ( ( u64 )in[ 5] )  << 40;
 
-  case 5: b |= ( ( u64 )in[ 4] )  << 32;
+  case 5: state->padding |= ( ( u64 )in[ 4] )  << 32;
 
-  case 4: b |= ( ( u64 )in[ 3] )  << 24;
+  case 4: state->padding |= ( ( u64 )in[ 3] )  << 24;
 
-  case 3: b |= ( ( u64 )in[ 2] )  << 16;
+  case 3: state->padding |= ( ( u64 )in[ 2] )  << 16;
 
-  case 2: b |= ( ( u64 )in[ 1] )  <<  8;
+  case 2: state->padding |= ( ( u64 )in[ 1] )  <<  8;
 
-  case 1: b |= ( ( u64 )in[ 0] ); break;
+  case 1: state->padding |= ( ( u64 )in[ 0] ); break;
 
   case 0: break;
   }
+}
+
+uint64_t siphash24_finalize(struct siphash *state) {
+  u64 b;
 
+  b = state->padding | (( ( u64 )state->inlen ) << 56);
 #ifdef DEBUG
-  printf( "(%3d) v0 %08x %08x\n", ( int )inlen, ( u32 )( v0 >> 32 ), ( u32 )v0 );
-  printf( "(%3d) v1 %08x %08x\n", ( int )inlen, ( u32 )( v1 >> 32 ), ( u32 )v1 );
-  printf( "(%3d) v2 %08x %08x\n", ( int )inlen, ( u32 )( v2 >> 32 ), ( u32 )v2 );
-  printf( "(%3d) v3 %08x %08x\n", ( int )inlen, ( u32 )( v3 >> 32 ), ( u32 )v3 );
-  printf( "(%3d) padding   %08x %08x\n", ( int )inlen, ( u32 )( b >> 32 ), ( u32 )b );
+  printf( "(%3d) v0 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v0 >> 32 ), ( u32 )state->v0 );
+  printf( "(%3d) v1 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v1 >> 32 ), ( u32 )state->v1 );
+  printf( "(%3d) v2 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v2 >> 32 ), ( u32 )state->v2 );
+  printf( "(%3d) v3 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v3 >> 32 ), ( u32 )state->v3 );
+  printf( "(%3d) padding   %08x %08x\n", ( int )state->inlen, ( u32 )( state->padding >> 32 ), ( u32 )state->padding );
 #endif
-  v3 ^= b;
-  SIPROUND;
-  SIPROUND;
-  v0 ^= b;
+  state->v3 ^= b;
+  SIPROUND(state);
+  SIPROUND(state);
+  state->v0 ^= b;
+
 #ifdef DEBUG
-  printf( "(%3d) v0 %08x %08x\n", ( int )inlen, ( u32 )( v0 >> 32 ), ( u32 )v0 );
-  printf( "(%3d) v1 %08x %08x\n", ( int )inlen, ( u32 )( v1 >> 32 ), ( u32 )v1 );
-  printf( "(%3d) v2 %08x %08x\n", ( int )inlen, ( u32 )( v2 >> 32 ), ( u32 )v2 );
-  printf( "(%3d) v3 %08x %08x\n", ( int )inlen, ( u32 )( v3 >> 32 ), ( u32 )v3 );
+  printf( "(%3d) v0 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v0 >> 32 ), ( u32 )state->v0 );
+  printf( "(%3d) v1 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v1 >> 32 ), ( u32 )state->v1 );
+  printf( "(%3d) v2 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v2 >> 32 ), ( u32 )state->v2 );
+  printf( "(%3d) v3 %08x %08x\n", ( int )state->inlen, ( u32 )( state->v3 >> 32 ), ( u32 )state->v3 );
 #endif
-  v2 ^= 0xff;
-  SIPROUND;
-  SIPROUND;
-  SIPROUND;
-  SIPROUND;
-  b = v0 ^ v1 ^ v2  ^ v3;
+  state->v2 ^= 0xff;
+  SIPROUND(state);
+  SIPROUND(state);
+  SIPROUND(state);
+  SIPROUND(state);
+
+  return state->v0 ^ state->v1 ^ state->v2  ^ state->v3;
+}
+
+/* SipHash-2-4 */
+void siphash24(uint8_t out[8], const void *_in, size_t inlen, const uint8_t k[16])
+{
+  struct siphash state;
+  u64 b;
+
+  siphash_init(&state, k);
+
+  siphash24_compress(_in, inlen, &state);
+
+  b = siphash24_finalize(&state);
+
   U64TO8_LE( out, b );
 }
diff --git a/src/basic/siphash24.h b/src/basic/siphash24.h
index 62e1168a79..c107bdd213 100644
--- a/src/basic/siphash24.h
+++ b/src/basic/siphash24.h
@@ -3,4 +3,17 @@
 #include <inttypes.h>
 #include <sys/types.h>
 
+struct siphash {
+  uint64_t v0;
+  uint64_t v1;
+  uint64_t v2;
+  uint64_t v3;
+  uint64_t padding;
+  size_t inlen;
+};
+
+void siphash_init(struct siphash *state, const uint8_t k[16]);
+void siphash24_compress(const void *in, size_t inlen, struct siphash *state);
+uint64_t siphash24_finalize(struct siphash *state);
+
 void siphash24(uint8_t out[8], const void *in, size_t inlen, const uint8_t k[16]);