diff options
Diffstat (limited to 'include')
| -rw-r--r-- | include/linux/dcache.h | 27 | ||||
| -rw-r--r-- | include/linux/hash.h | 108 | ||||
| -rw-r--r-- | include/linux/stringhash.h | 76 | ||||
| -rw-r--r-- | include/linux/sunrpc/svcauth.h | 40 |
4 files changed, 141 insertions, 110 deletions
diff --git a/include/linux/dcache.h b/include/linux/dcache.h index f8506e8dd4d4..484c8792da82 100644 --- a/include/linux/dcache.h +++ b/include/linux/dcache.h @@ -10,6 +10,7 @@ #include <linux/cache.h> #include <linux/rcupdate.h> #include <linux/lockref.h> +#include <linux/stringhash.h> struct path; struct vfsmount; @@ -52,9 +53,6 @@ struct qstr { }; #define QSTR_INIT(n,l) { { { .len = l } }, .name = n } -#define hashlen_hash(hashlen) ((u32) (hashlen)) -#define hashlen_len(hashlen) ((u32)((hashlen) >> 32)) -#define hashlen_create(hash,len) (((u64)(len)<<32)|(u32)(hash)) struct dentry_stat_t { long nr_dentry; @@ -65,29 +63,6 @@ struct dentry_stat_t { }; extern struct dentry_stat_t dentry_stat; -/* Name hashing routines. Initial hash value */ -/* Hash courtesy of the R5 hash in reiserfs modulo sign bits */ -#define init_name_hash() 0 - -/* partial hash update function. Assume roughly 4 bits per character */ -static inline unsigned long -partial_name_hash(unsigned long c, unsigned long prevhash) -{ - return (prevhash + (c << 4) + (c >> 4)) * 11; -} - -/* - * Finally: cut down the number of bits to a int value (and try to avoid - * losing bits) - */ -static inline unsigned long end_name_hash(unsigned long hash) -{ - return (unsigned int) hash; -} - -/* Compute the hash for a name string. */ -extern unsigned int full_name_hash(const unsigned char *, unsigned int); - /* * Try to keep struct dentry aligned on 64 byte cachelines (this will * give reasonable cacheline footprint with larger lines without the diff --git a/include/linux/hash.h b/include/linux/hash.h index 79c52fa81cac..ad6fa21d977b 100644 --- a/include/linux/hash.h +++ b/include/linux/hash.h @@ -3,92 +3,94 @@ /* Fast hashing routine for ints, longs and pointers. (C) 2002 Nadia Yvette Chambers, IBM */ -/* - * Knuth recommends primes in approximately golden ratio to the maximum - * integer representable by a machine word for multiplicative hashing. - * Chuck Lever verified the effectiveness of this technique: - * http://www.citi.umich.edu/techreports/reports/citi-tr-00-1.pdf - * - * These primes are chosen to be bit-sparse, that is operations on - * them can use shifts and additions instead of multiplications for - * machines where multiplications are slow. - */ - #include <asm/types.h> #include <linux/compiler.h> -/* 2^31 + 2^29 - 2^25 + 2^22 - 2^19 - 2^16 + 1 */ -#define GOLDEN_RATIO_PRIME_32 0x9e370001UL -/* 2^63 + 2^61 - 2^57 + 2^54 - 2^51 - 2^18 + 1 */ -#define GOLDEN_RATIO_PRIME_64 0x9e37fffffffc0001UL - +/* + * The "GOLDEN_RATIO_PRIME" is used in ifs/btrfs/brtfs_inode.h and + * fs/inode.c. It's not actually prime any more (the previous primes + * were actively bad for hashing), but the name remains. + */ #if BITS_PER_LONG == 32 -#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_PRIME_32 +#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_32 #define hash_long(val, bits) hash_32(val, bits) #elif BITS_PER_LONG == 64 #define hash_long(val, bits) hash_64(val, bits) -#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_PRIME_64 +#define GOLDEN_RATIO_PRIME GOLDEN_RATIO_64 #else #error Wordsize not 32 or 64 #endif /* - * The above primes are actively bad for hashing, since they are - * too sparse. The 32-bit one is mostly ok, the 64-bit one causes - * real problems. Besides, the "prime" part is pointless for the - * multiplicative hash. + * This hash multiplies the input by a large odd number and takes the + * high bits. Since multiplication propagates changes to the most + * significant end only, it is essential that the high bits of the + * product be used for the hash value. + * + * Chuck Lever verified the effectiveness of this technique: + * http://www.citi.umich.edu/techreports/reports/citi-tr-00-1.pdf * * Although a random odd number will do, it turns out that the golden * ratio phi = (sqrt(5)-1)/2, or its negative, has particularly nice - * properties. + * properties. (See Knuth vol 3, section 6.4, exercise 9.) * - * These are the negative, (1 - phi) = (phi^2) = (3 - sqrt(5))/2. - * (See Knuth vol 3, section 6.4, exercise 9.) + * These are the negative, (1 - phi) = phi**2 = (3 - sqrt(5))/2, + * which is very slightly easier to multiply by and makes no + * difference to the hash distribution. */ #define GOLDEN_RATIO_32 0x61C88647 #define GOLDEN_RATIO_64 0x61C8864680B583EBull -static __always_inline u64 hash_64(u64 val, unsigned int bits) -{ - u64 hash = val; +#ifdef CONFIG_HAVE_ARCH_HASH +/* This header may use the GOLDEN_RATIO_xx constants */ +#include <asm/hash.h> +#endif -#if BITS_PER_LONG == 64 - hash = hash * GOLDEN_RATIO_64; -#else - /* Sigh, gcc can't optimise this alone like it does for 32 bits. */ - u64 n = hash; - n <<= 18; - hash -= n; - n <<= 33; - hash -= n; - n <<= 3; - hash += n; - n <<= 3; - hash -= n; - n <<= 4; - hash += n; - n <<= 2; - hash += n; +/* + * The _generic versions exist only so lib/test_hash.c can compare + * the arch-optimized versions with the generic. + * + * Note that if you change these, any <asm/hash.h> that aren't updated + * to match need to have their HAVE_ARCH_* define values updated so the + * self-test will not false-positive. + */ +#ifndef HAVE_ARCH__HASH_32 +#define __hash_32 __hash_32_generic #endif +static inline u32 __hash_32_generic(u32 val) +{ + return val * GOLDEN_RATIO_32; +} +#ifndef HAVE_ARCH_HASH_32 +#define hash_32 hash_32_generic +#endif +static inline u32 hash_32_generic(u32 val, unsigned int bits) +{ /* High bits are more random, so use them. */ - return hash >> (64 - bits); + return __hash_32(val) >> (32 - bits); } -static inline u32 hash_32(u32 val, unsigned int bits) +#ifndef HAVE_ARCH_HASH_64 +#define hash_64 hash_64_generic +#endif +static __always_inline u32 hash_64_generic(u64 val, unsigned int bits) { - /* On some cpus multiply is faster, on others gcc will do shifts */ - u32 hash = val * GOLDEN_RATIO_PRIME_32; - - /* High bits are more random, so use them. */ - return hash >> (32 - bits); +#if BITS_PER_LONG == 64 + /* 64x64-bit multiply is efficient on all 64-bit processors */ + return val * GOLDEN_RATIO_64 >> (64 - bits); +#else + /* Hash 64 bits using only 32x32-bit multiply. */ + return hash_32((u32)val ^ __hash_32(val >> 32), bits); +#endif } -static inline unsigned long hash_ptr(const void *ptr, unsigned int bits) +static inline u32 hash_ptr(const void *ptr, unsigned int bits) { return hash_long((unsigned long)ptr, bits); } +/* This really should be called fold32_ptr; it does no hashing to speak of. */ static inline u32 hash32_ptr(const void *ptr) { unsigned long val = (unsigned long)ptr; diff --git a/include/linux/stringhash.h b/include/linux/stringhash.h new file mode 100644 index 000000000000..451771d9b9c0 --- /dev/null +++ b/include/linux/stringhash.h @@ -0,0 +1,76 @@ +#ifndef __LINUX_STRINGHASH_H +#define __LINUX_STRINGHASH_H + +#include <linux/compiler.h> /* For __pure */ +#include <linux/types.h> /* For u32, u64 */ + +/* + * Routines for hashing strings of bytes to a 32-bit hash value. + * + * These hash functions are NOT GUARANTEED STABLE between kernel + * versions, architectures, or even repeated boots of the same kernel. + * (E.g. they may depend on boot-time hardware detection or be + * deliberately randomized.) + * + * They are also not intended to be secure against collisions caused by + * malicious inputs; much slower hash functions are required for that. + * + * They are optimized for pathname components, meaning short strings. + * Even if a majority of files have longer names, the dynamic profile of + * pathname components skews short due to short directory names. + * (E.g. /usr/lib/libsesquipedalianism.so.3.141.) + */ + +/* + * Version 1: one byte at a time. Example of use: + * + * unsigned long hash = init_name_hash; + * while (*p) + * hash = partial_name_hash(tolower(*p++), hash); + * hash = end_name_hash(hash); + * + * Although this is designed for bytes, fs/hfsplus/unicode.c + * abuses it to hash 16-bit values. + */ + +/* Hash courtesy of the R5 hash in reiserfs modulo sign bits */ +#define init_name_hash() 0 + +/* partial hash update function. Assume roughly 4 bits per character */ +static inline unsigned long +partial_name_hash(unsigned long c, unsigned long prevhash) +{ + return (prevhash + (c << 4) + (c >> 4)) * 11; +} + +/* + * Finally: cut down the number of bits to a int value (and try to avoid + * losing bits) + */ +static inline unsigned long end_name_hash(unsigned long hash) +{ + return (unsigned int)hash; +} + +/* + * Version 2: One word (32 or 64 bits) at a time. + * If CONFIG_DCACHE_WORD_ACCESS is defined (meaning <asm/word-at-a-time.h> + * exists, which describes major Linux platforms like x86 and ARM), then + * this computes a different hash function much faster. + * + * If not set, this falls back to a wrapper around the preceding. + */ +extern unsigned int __pure full_name_hash(const char *, unsigned int); + +/* + * A hash_len is a u64 with the hash of a string in the low + * half and the length in the high half. + */ +#define hashlen_hash(hashlen) ((u32)(hashlen)) +#define hashlen_len(hashlen) ((u32)((hashlen) >> 32)) +#define hashlen_create(hash, len) ((u64)(len)<<32 | (u32)(hash)) + +/* Return the "hash_len" (hash and length) of a null-terminated string */ +extern u64 __pure hashlen_string(const char *name); + +#endif /* __LINUX_STRINGHASH_H */ diff --git a/include/linux/sunrpc/svcauth.h b/include/linux/sunrpc/svcauth.h index c00f53a4ccdd..91d5a5d6f52b 100644 --- a/include/linux/sunrpc/svcauth.h +++ b/include/linux/sunrpc/svcauth.h @@ -16,6 +16,7 @@ #include <linux/sunrpc/cache.h> #include <linux/sunrpc/gss_api.h> #include <linux/hash.h> +#include <linux/stringhash.h> #include <linux/cred.h> struct svc_cred { @@ -165,41 +166,18 @@ extern int svcauth_unix_set_client(struct svc_rqst *rqstp); extern int unix_gid_cache_create(struct net *net); extern void unix_gid_cache_destroy(struct net *net); -static inline unsigned long hash_str(char *name, int bits) +/* + * The <stringhash.h> functions are good enough that we don't need to + * use hash_32() on them; just extracting the high bits is enough. + */ +static inline unsigned long hash_str(char const *name, int bits) { - unsigned long hash = 0; - unsigned long l = 0; - int len = 0; - unsigned char c; - do { - if (unlikely(!(c = *name++))) { - c = (char)len; len = -1; - } - l = (l << 8) | c; - len++; - if ((len & (BITS_PER_LONG/8-1))==0) - hash = hash_long(hash^l, BITS_PER_LONG); - } while (len); - return hash >> (BITS_PER_LONG - bits); + return hashlen_hash(hashlen_string(name)) >> (32 - bits); } -static inline unsigned long hash_mem(char *buf, int length, int bits) +static inline unsigned long hash_mem(char const *buf, int length, int bits) { - unsigned long hash = 0; - unsigned long l = 0; - int len = 0; - unsigned char c; - do { - if (len == length) { - c = (char)len; len = -1; - } else - c = *buf++; - l = (l << 8) | c; - len++; - if ((len & (BITS_PER_LONG/8-1))==0) - hash = hash_long(hash^l, BITS_PER_LONG); - } while (len); - return hash >> (BITS_PER_LONG - bits); + return full_name_hash(buf, length) >> (32 - bits); } #endif /* __KERNEL__ */ |