diff options
author | Michal Schmidt <mschmidt@redhat.com> | 2014-10-15 01:27:16 +0200 |
---|---|---|
committer | Michal Schmidt <mschmidt@redhat.com> | 2014-10-30 19:50:51 +0100 |
commit | 89439d4fc0d29f04ac68432fd06ab84bc4e36e20 (patch) | |
tree | 45cfb719ba5dd1e1d54678ca1f4159e0dd3b0c1f /src/shared/set.h | |
parent | ce79279bff6e7a1a17070509a039ab635796f129 (diff) |
hashmap: rewrite the implementation
This is a rewrite of the hashmap implementation. Its advantage is lower
memory usage.
It uses open addressing (entries are stored in an array, as opposed to
linked lists). Hash collisions are resolved with linear probing and
Robin Hood displacement policy. See the references in hashmap.c.
Some fun empirical findings about hashmap usage in systemd on my laptop:
- 98 % of allocated hashmaps are Sets.
- Sets contain 78 % of all entries, plain Hashmaps 17 %, and
OrderedHashmaps 5 %.
- 60 % of allocated hashmaps contain only 1 entry.
- 90 % of allocated hashmaps contain 5 or fewer entries.
- 75 % of all entries are in hashmaps that use trivial_hash_ops.
Clearly it makes sense to:
- store entries in distinct entry types. Especially for Sets - their
entries are the most numerous and they require the least information
to store an entry.
- have a way to store small numbers of entries directly in the hashmap
structs, and only allocate the usual entry arrays when the direct
storage is full.
The implementation has an optional debugging feature (enabled by
defining the ENABLE_HASHMAP_DEBUG macro), where it:
- tracks all allocated hashmaps in a linked list so that one can
easily find them in gdb,
- tracks which function/line allocated a given hashmap, and
- checks for invalid mixing of hashmap iteration and modification.
Since entries are not allocated one-by-one anymore, mempools are not
used for entries. Originally I meant to drop mempools entirely, but it's
still worth it to use them for the hashmap structs. My testing indicates
that it makes loading of units about 5 % faster (a test with 10000 units
where more than 200000 hashmaps are allocated - pure malloc: 449±4 ms,
mempools: 427±7 ms).
Here are some memory usage numbers, taken on my laptop with a more or
less normal Fedora setup after booting with SELinux disabled (SELinux
increases systemd's memory usage significantly):
systemd (PID 1) Original New Change
dirty memory (from pmap -x 1) [KiB] 2152 1264 -41 %
total heap allocations (from gdb-heap) [KiB] 1623 756 -53 %
Diffstat (limited to 'src/shared/set.h')
-rw-r--r-- | src/shared/set.h | 120 |
1 files changed, 89 insertions, 31 deletions
diff --git a/src/shared/set.h b/src/shared/set.h index d2622d17ea..4605ecd2c1 100644 --- a/src/shared/set.h +++ b/src/shared/set.h @@ -21,56 +21,114 @@ along with systemd; If not, see <http://www.gnu.org/licenses/>. ***/ -/* Pretty straightforward set implementation. Internally based on the - * hashmap. That means that as a minor optimization a NULL set - * object will be treated as empty set for all read - * operations. That way it is not necessary to instantiate an object - * for each set use. */ - #include "hashmap.h" #include "util.h" -typedef struct Set Set; +Set *internal_set_new(const struct hash_ops *hash_ops HASHMAP_DEBUG_PARAMS); +#define set_new(ops) internal_set_new(ops HASHMAP_DEBUG_SRC_ARGS) -Set *set_new(const struct hash_ops *hash_ops); -void set_free(Set* s); -void set_free_free(Set *s); -Set* set_copy(Set *s); -int set_ensure_allocated(Set **s, const struct hash_ops *hash_ops); +static inline void set_free(Set *s) { + internal_hashmap_free(HASHMAP_BASE(s)); +} -int set_put(Set *s, void *value); -int set_consume(Set *s, void *value); -int set_put_strdup(Set *s, const char *p); -int set_put_strdupv(Set *s, char **l); -int set_replace(Set *s, void *value); -void *set_get(Set *s, void *value); -bool set_contains(Set *s, void *value); -void *set_remove(Set *s, void *value); -int set_remove_and_put(Set *s, void *old_value, void *new_value); +static inline void set_free_free(Set *s) { + internal_hashmap_free_free(HASHMAP_BASE(s)); +} + +/* no set_free_free_free */ + +static inline Set *set_copy(Set *s) { + return (Set*) internal_hashmap_copy(HASHMAP_BASE(s)); +} + +int internal_set_ensure_allocated(Set **s, const struct hash_ops *hash_ops HASHMAP_DEBUG_PARAMS); +#define set_ensure_allocated(h, ops) internal_set_ensure_allocated(h, ops HASHMAP_DEBUG_SRC_ARGS) +int set_put(Set *s, const void *key); +/* no set_update */ +/* no set_replace */ +static inline void *set_get(Set *s, void *key) { + return internal_hashmap_get(HASHMAP_BASE(s), key); +} +/* no set_get2 */ + +static inline bool set_contains(Set *s, const void *key) { + return internal_hashmap_contains(HASHMAP_BASE(s), key); +} + +static inline void *set_remove(Set *s, void *key) { + return internal_hashmap_remove(HASHMAP_BASE(s), key); +} + +/* no set_remove2 */ +/* no set_remove_value */ +int set_remove_and_put(Set *s, const void *old_key, const void *new_key); +/* no set_remove_and_replace */ int set_merge(Set *s, Set *other); -int set_reserve(Set *s, unsigned entries_add); -int set_move(Set *s, Set *other); -int set_move_one(Set *s, Set *other, void *value); -unsigned set_size(Set *s); -bool set_isempty(Set *s); +static inline int set_reserve(Set *h, unsigned entries_add) { + return internal_hashmap_reserve(HASHMAP_BASE(h), entries_add); +} + +static inline int set_move(Set *s, Set *other) { + return internal_hashmap_move(HASHMAP_BASE(s), HASHMAP_BASE(other)); +} + +static inline int set_move_one(Set *s, Set *other, const void *key) { + return internal_hashmap_move_one(HASHMAP_BASE(s), HASHMAP_BASE(other), key); +} + +static inline unsigned set_size(Set *s) { + return internal_hashmap_size(HASHMAP_BASE(s)); +} + +static inline bool set_isempty(Set *s) { + return set_size(s) == 0; +} + +static inline unsigned set_buckets(Set *s) { + return internal_hashmap_buckets(HASHMAP_BASE(s)); +} void *set_iterate(Set *s, Iterator *i); -void set_clear(Set *s); -void set_clear_free(Set *s); +static inline void set_clear(Set *s) { + internal_hashmap_clear(HASHMAP_BASE(s)); +} + +static inline void set_clear_free(Set *s) { + internal_hashmap_clear_free(HASHMAP_BASE(s)); +} + +/* no set_clear_free_free */ + +static inline void *set_steal_first(Set *s) { + return internal_hashmap_steal_first(HASHMAP_BASE(s)); +} + +/* no set_steal_first_key */ +/* no set_first_key */ -void *set_steal_first(Set *s); -void* set_first(Set *s); +static inline void *set_first(Set *s) { + return internal_hashmap_first(HASHMAP_BASE(s)); +} -char **set_get_strv(Set *s); +/* no set_next */ + +static inline char **set_get_strv(Set *s) { + return internal_hashmap_get_strv(HASHMAP_BASE(s)); +} + +int set_consume(Set *s, void *value); +int set_put_strdup(Set *s, const char *p); +int set_put_strdupv(Set *s, char **l); #define SET_FOREACH(e, s, i) \ for ((i) = ITERATOR_FIRST, (e) = set_iterate((s), &(i)); (e); (e) = set_iterate((s), &(i))) DEFINE_TRIVIAL_CLEANUP_FUNC(Set*, set_free); DEFINE_TRIVIAL_CLEANUP_FUNC(Set*, set_free_free); + #define _cleanup_set_free_ _cleanup_(set_freep) #define _cleanup_set_free_free_ _cleanup_(set_free_freep) |