From 35b72573e977ed6b18b094136a4fa3e0ffb13603 Mon Sep 17 00:00:00 2001 From: David Howells Date: Thu, 17 Jun 2021 14:21:00 +0100 Subject: fscache: Fix cookie key hashing The current hash algorithm used for hashing cookie keys is really bad, producing almost no dispersion (after a test kernel build, ~30000 files were split over just 18 out of the 32768 hash buckets). Borrow the full_name_hash() hash function into fscache to do the hashing for cookie keys and, in the future, volume keys. I don't want to use full_name_hash() as-is because I want the hash value to be consistent across arches and over time as the hash value produced may get used on disk. I can also optimise parts of it away as the key will always be a padded array of aligned 32-bit words. Fixes: ec0328e46d6e ("fscache: Maintain a catalogue of allocated cookies") Signed-off-by: David Howells Reviewed-by: Jeff Layton cc: linux-cachefs@redhat.com Link: https://lore.kernel.org/r/162431201844.2908479.8293647220901514696.stgit@warthog.procyon.org.uk/ --- fs/fscache/main.c | 39 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 39 insertions(+) (limited to 'fs/fscache/main.c') diff --git a/fs/fscache/main.c b/fs/fscache/main.c index c1e6cc9091aa..4207f98e405f 100644 --- a/fs/fscache/main.c +++ b/fs/fscache/main.c @@ -93,6 +93,45 @@ static struct ctl_table fscache_sysctls_root[] = { }; #endif +/* + * Mixing scores (in bits) for (7,20): + * Input delta: 1-bit 2-bit + * 1 round: 330.3 9201.6 + * 2 rounds: 1246.4 25475.4 + * 3 rounds: 1907.1 31295.1 + * 4 rounds: 2042.3 31718.6 + * Perfect: 2048 31744 + * (32*64) (32*31/2 * 64) + */ +#define HASH_MIX(x, y, a) \ + ( x ^= (a), \ + y ^= x, x = rol32(x, 7),\ + x += y, y = rol32(y,20),\ + y *= 9 ) + +static inline unsigned int fold_hash(unsigned long x, unsigned long y) +{ + /* Use arch-optimized multiply if one exists */ + return __hash_32(y ^ __hash_32(x)); +} + +/* + * Generate a hash. This is derived from full_name_hash(), but we want to be + * sure it is arch independent and that it doesn't change as bits of the + * computed hash value might appear on disk. The caller also guarantees that + * the hashed data will be a series of aligned 32-bit words. + */ +unsigned int fscache_hash(unsigned int salt, unsigned int *data, unsigned int n) +{ + unsigned int a, x = 0, y = salt; + + for (; n; n--) { + a = *data++; + HASH_MIX(x, y, a); + } + return fold_hash(x, y); +} + /* * initialise the fs caching module */ -- cgit