Commit 4feb7c7a authored by NeilBrown's avatar NeilBrown Committed by David S. Miller

rhashtable: don't hold lock on first table throughout insertion.

rhashtable_try_insert() currently holds a lock on the bucket in
the first table, while also locking buckets in subsequent tables.
This is unnecessary and looks like a hold-over from some earlier
version of the implementation.

As insert and remove always lock a bucket in each table in turn, and
as insert only inserts in the final table, there cannot be any races
that are not covered by simply locking a bucket in each table in turn.

When an insert call reaches that last table it can be sure that there
is no matchinf entry in any other table as it has searched them all, and
insertion never happens anywhere but in the last table.  The fact that
code tests for the existence of future_tbl while holding a lock on
the relevant bucket ensures that two threads inserting the same key
will make compatible decisions about which is the "last" table.

This simplifies the code and allows the ->rehash field to be
discarded.

We still need a way to ensure that a dead bucket_table is never
re-linked by rhashtable_walk_stop().  This can be achieved by calling
call_rcu() inside the locked region, and checking with
rcu_head_after_call_rcu() in rhashtable_walk_stop() to see if the
bucket table is empty and dead.
Acked-by: default avatarHerbert Xu <herbert@gondor.apana.org.au>
Reviewed-by: default avatarPaul E. McKenney <paulmck@linux.ibm.com>
Signed-off-by: default avatarNeilBrown <neilb@suse.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parent 83b038db
...@@ -63,7 +63,6 @@ ...@@ -63,7 +63,6 @@
struct bucket_table { struct bucket_table {
unsigned int size; unsigned int size;
unsigned int nest; unsigned int nest;
unsigned int rehash;
u32 hash_rnd; u32 hash_rnd;
unsigned int locks_mask; unsigned int locks_mask;
spinlock_t *locks; spinlock_t *locks;
...@@ -776,12 +775,6 @@ static inline int rhltable_insert( ...@@ -776,12 +775,6 @@ static inline int rhltable_insert(
* @obj: pointer to hash head inside object * @obj: pointer to hash head inside object
* @params: hash table parameters * @params: hash table parameters
* *
* Locks down the bucket chain in both the old and new table if a resize
* is in progress to ensure that writers can't remove from the old table
* and can't insert to the new table during the atomic operation of search
* and insertion. Searches for duplicates in both the old and new table if
* a resize is in progress.
*
* This lookup function may only be used for fixed key hash table (key_len * This lookup function may only be used for fixed key hash table (key_len
* parameter set). It will BUG() if used inappropriately. * parameter set). It will BUG() if used inappropriately.
* *
...@@ -837,12 +830,6 @@ static inline void *rhashtable_lookup_get_insert_fast( ...@@ -837,12 +830,6 @@ static inline void *rhashtable_lookup_get_insert_fast(
* @obj: pointer to hash head inside object * @obj: pointer to hash head inside object
* @params: hash table parameters * @params: hash table parameters
* *
* Locks down the bucket chain in both the old and new table if a resize
* is in progress to ensure that writers can't remove from the old table
* and can't insert to the new table during the atomic operation of search
* and insertion. Searches for duplicates in both the old and new table if
* a resize is in progress.
*
* Lookups may occur in parallel with hashtable mutations and resizing. * Lookups may occur in parallel with hashtable mutations and resizing.
* *
* Will trigger an automatic deferred table resizing if residency in the * Will trigger an automatic deferred table resizing if residency in the
......
...@@ -197,6 +197,7 @@ static struct bucket_table *bucket_table_alloc(struct rhashtable *ht, ...@@ -197,6 +197,7 @@ static struct bucket_table *bucket_table_alloc(struct rhashtable *ht,
return NULL; return NULL;
} }
rcu_head_init(&tbl->rcu);
INIT_LIST_HEAD(&tbl->walkers); INIT_LIST_HEAD(&tbl->walkers);
tbl->hash_rnd = get_random_u32(); tbl->hash_rnd = get_random_u32();
...@@ -280,10 +281,9 @@ static int rhashtable_rehash_chain(struct rhashtable *ht, ...@@ -280,10 +281,9 @@ static int rhashtable_rehash_chain(struct rhashtable *ht,
while (!(err = rhashtable_rehash_one(ht, old_hash))) while (!(err = rhashtable_rehash_one(ht, old_hash)))
; ;
if (err == -ENOENT) { if (err == -ENOENT)
old_tbl->rehash++;
err = 0; err = 0;
}
spin_unlock_bh(old_bucket_lock); spin_unlock_bh(old_bucket_lock);
return err; return err;
...@@ -330,13 +330,16 @@ static int rhashtable_rehash_table(struct rhashtable *ht) ...@@ -330,13 +330,16 @@ static int rhashtable_rehash_table(struct rhashtable *ht)
spin_lock(&ht->lock); spin_lock(&ht->lock);
list_for_each_entry(walker, &old_tbl->walkers, list) list_for_each_entry(walker, &old_tbl->walkers, list)
walker->tbl = NULL; walker->tbl = NULL;
spin_unlock(&ht->lock);
/* Wait for readers. All new readers will see the new /* Wait for readers. All new readers will see the new
* table, and thus no references to the old table will * table, and thus no references to the old table will
* remain. * remain.
* We do this inside the locked region so that
* rhashtable_walk_stop() can use rcu_head_after_call_rcu()
* to check if it should not re-link the table.
*/ */
call_rcu(&old_tbl->rcu, bucket_table_free_rcu); call_rcu(&old_tbl->rcu, bucket_table_free_rcu);
spin_unlock(&ht->lock);
return rht_dereference(new_tbl->future_tbl, ht) ? -EAGAIN : 0; return rht_dereference(new_tbl->future_tbl, ht) ? -EAGAIN : 0;
} }
...@@ -578,46 +581,22 @@ static void *rhashtable_try_insert(struct rhashtable *ht, const void *key, ...@@ -578,46 +581,22 @@ static void *rhashtable_try_insert(struct rhashtable *ht, const void *key,
struct bucket_table *new_tbl; struct bucket_table *new_tbl;
struct bucket_table *tbl; struct bucket_table *tbl;
unsigned int hash; unsigned int hash;
spinlock_t *lock;
void *data; void *data;
tbl = rcu_dereference(ht->tbl); new_tbl = rcu_dereference(ht->tbl);
/* All insertions must grab the oldest table containing
* the hashed bucket that is yet to be rehashed.
*/
for (;;) {
hash = rht_head_hashfn(ht, tbl, obj, ht->p);
lock = rht_bucket_lock(tbl, hash);
spin_lock_bh(lock);
if (tbl->rehash <= hash)
break;
spin_unlock_bh(lock);
tbl = rht_dereference_rcu(tbl->future_tbl, ht);
}
data = rhashtable_lookup_one(ht, tbl, hash, key, obj);
new_tbl = rhashtable_insert_one(ht, tbl, hash, obj, data);
if (PTR_ERR(new_tbl) != -EEXIST)
data = ERR_CAST(new_tbl);
while (!IS_ERR_OR_NULL(new_tbl)) { do {
tbl = new_tbl; tbl = new_tbl;
hash = rht_head_hashfn(ht, tbl, obj, ht->p); hash = rht_head_hashfn(ht, tbl, obj, ht->p);
spin_lock_nested(rht_bucket_lock(tbl, hash), spin_lock_bh(rht_bucket_lock(tbl, hash));
SINGLE_DEPTH_NESTING);
data = rhashtable_lookup_one(ht, tbl, hash, key, obj); data = rhashtable_lookup_one(ht, tbl, hash, key, obj);
new_tbl = rhashtable_insert_one(ht, tbl, hash, obj, data); new_tbl = rhashtable_insert_one(ht, tbl, hash, obj, data);
if (PTR_ERR(new_tbl) != -EEXIST) if (PTR_ERR(new_tbl) != -EEXIST)
data = ERR_CAST(new_tbl); data = ERR_CAST(new_tbl);
spin_unlock(rht_bucket_lock(tbl, hash)); spin_unlock_bh(rht_bucket_lock(tbl, hash));
} } while (!IS_ERR_OR_NULL(new_tbl));
spin_unlock_bh(lock);
if (PTR_ERR(data) == -EAGAIN) if (PTR_ERR(data) == -EAGAIN)
data = ERR_PTR(rhashtable_insert_rehash(ht, tbl) ?: data = ERR_PTR(rhashtable_insert_rehash(ht, tbl) ?:
...@@ -939,10 +918,11 @@ void rhashtable_walk_stop(struct rhashtable_iter *iter) ...@@ -939,10 +918,11 @@ void rhashtable_walk_stop(struct rhashtable_iter *iter)
ht = iter->ht; ht = iter->ht;
spin_lock(&ht->lock); spin_lock(&ht->lock);
if (tbl->rehash < tbl->size) if (rcu_head_after_call_rcu(&tbl->rcu, bucket_table_free_rcu))
list_add(&iter->walker.list, &tbl->walkers); /* This bucket table is being freed, don't re-link it. */
else
iter->walker.tbl = NULL; iter->walker.tbl = NULL;
else
list_add(&iter->walker.list, &tbl->walkers);
spin_unlock(&ht->lock); spin_unlock(&ht->lock);
out: out:
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment