Commit ca26893f authored by Herbert Xu's avatar Herbert Xu Committed by David S. Miller

rhashtable: Add rhlist interface

The insecure_elasticity setting is an ugly wart brought out by
users who need to insert duplicate objects (that is, distinct
objects with identical keys) into the same table.

In fact, those users have a much bigger problem.  Once those
duplicate objects are inserted, they don't have an interface to
find them (unless you count the walker interface which walks
over the entire table).

Some users have resorted to doing a manual walk over the hash
table which is of course broken because they don't handle the
potential existence of multiple hash tables.  The result is that
they will break sporadically when they encounter a hash table
resize/rehash.

This patch provides a way out for those users, at the expense
of an extra pointer per object.  Essentially each object is now
a list of objects carrying the same key.  The hash table will
only see the lists so nothing changes as far as rhashtable is
concerned.

To use this new interface, you need to insert a struct rhlist_head
into your objects instead of struct rhash_head.  While the hash
table is unchanged, for type-safety you'll need to use struct
rhltable instead of struct rhashtable.  All the existing interfaces
have been duplicated for rhlist, including the hash table walker.

One missing feature is nulls marking because AFAIK the only potential
user of it does not need duplicate objects.  Should anyone need
this it shouldn't be too hard to add.
Signed-off-by: default avatarHerbert Xu <herbert@gondor.apana.org.au>
Acked-by: default avatarThomas Graf <tgraf@suug.ch>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parent fd07160b
/* /*
* Resizable, Scalable, Concurrent Hash Table * Resizable, Scalable, Concurrent Hash Table
* *
* Copyright (c) 2015 Herbert Xu <herbert@gondor.apana.org.au> * Copyright (c) 2015-2016 Herbert Xu <herbert@gondor.apana.org.au>
* Copyright (c) 2014-2015 Thomas Graf <tgraf@suug.ch> * Copyright (c) 2014-2015 Thomas Graf <tgraf@suug.ch>
* Copyright (c) 2008-2014 Patrick McHardy <kaber@trash.net> * Copyright (c) 2008-2014 Patrick McHardy <kaber@trash.net>
* *
...@@ -53,6 +53,11 @@ struct rhash_head { ...@@ -53,6 +53,11 @@ struct rhash_head {
struct rhash_head __rcu *next; struct rhash_head __rcu *next;
}; };
struct rhlist_head {
struct rhash_head rhead;
struct rhlist_head __rcu *next;
};
/** /**
* struct bucket_table - Table of hash buckets * struct bucket_table - Table of hash buckets
* @size: Number of hash buckets * @size: Number of hash buckets
...@@ -137,6 +142,7 @@ struct rhashtable_params { ...@@ -137,6 +142,7 @@ struct rhashtable_params {
* @key_len: Key length for hashfn * @key_len: Key length for hashfn
* @elasticity: Maximum chain length before rehash * @elasticity: Maximum chain length before rehash
* @p: Configuration parameters * @p: Configuration parameters
* @rhlist: True if this is an rhltable
* @run_work: Deferred worker to expand/shrink asynchronously * @run_work: Deferred worker to expand/shrink asynchronously
* @mutex: Mutex to protect current/future table swapping * @mutex: Mutex to protect current/future table swapping
* @lock: Spin lock to protect walker list * @lock: Spin lock to protect walker list
...@@ -147,11 +153,20 @@ struct rhashtable { ...@@ -147,11 +153,20 @@ struct rhashtable {
unsigned int key_len; unsigned int key_len;
unsigned int elasticity; unsigned int elasticity;
struct rhashtable_params p; struct rhashtable_params p;
bool rhlist;
struct work_struct run_work; struct work_struct run_work;
struct mutex mutex; struct mutex mutex;
spinlock_t lock; spinlock_t lock;
}; };
/**
* struct rhltable - Hash table with duplicate objects in a list
* @ht: Underlying rhtable
*/
struct rhltable {
struct rhashtable ht;
};
/** /**
* struct rhashtable_walker - Hash table walker * struct rhashtable_walker - Hash table walker
* @list: List entry on list of walkers * @list: List entry on list of walkers
...@@ -163,9 +178,10 @@ struct rhashtable_walker { ...@@ -163,9 +178,10 @@ struct rhashtable_walker {
}; };
/** /**
* struct rhashtable_iter - Hash table iterator, fits into netlink cb * struct rhashtable_iter - Hash table iterator
* @ht: Table to iterate through * @ht: Table to iterate through
* @p: Current pointer * @p: Current pointer
* @list: Current hash list pointer
* @walker: Associated rhashtable walker * @walker: Associated rhashtable walker
* @slot: Current slot * @slot: Current slot
* @skip: Number of entries to skip in slot * @skip: Number of entries to skip in slot
...@@ -173,6 +189,7 @@ struct rhashtable_walker { ...@@ -173,6 +189,7 @@ struct rhashtable_walker {
struct rhashtable_iter { struct rhashtable_iter {
struct rhashtable *ht; struct rhashtable *ht;
struct rhash_head *p; struct rhash_head *p;
struct rhlist_head *list;
struct rhashtable_walker walker; struct rhashtable_walker walker;
unsigned int slot; unsigned int slot;
unsigned int skip; unsigned int skip;
...@@ -339,13 +356,11 @@ static inline int lockdep_rht_bucket_is_held(const struct bucket_table *tbl, ...@@ -339,13 +356,11 @@ static inline int lockdep_rht_bucket_is_held(const struct bucket_table *tbl,
int rhashtable_init(struct rhashtable *ht, int rhashtable_init(struct rhashtable *ht,
const struct rhashtable_params *params); const struct rhashtable_params *params);
int rhltable_init(struct rhltable *hlt,
const struct rhashtable_params *params);
struct bucket_table *rhashtable_insert_slow(struct rhashtable *ht, void *rhashtable_insert_slow(struct rhashtable *ht, const void *key,
const void *key, struct rhash_head *obj);
struct rhash_head *obj,
struct bucket_table *old_tbl,
void **data);
int rhashtable_insert_rehash(struct rhashtable *ht, struct bucket_table *tbl);
void rhashtable_walk_enter(struct rhashtable *ht, void rhashtable_walk_enter(struct rhashtable *ht,
struct rhashtable_iter *iter); struct rhashtable_iter *iter);
...@@ -507,6 +522,31 @@ void rhashtable_destroy(struct rhashtable *ht); ...@@ -507,6 +522,31 @@ void rhashtable_destroy(struct rhashtable *ht);
rht_for_each_entry_rcu_continue(tpos, pos, (tbl)->buckets[hash],\ rht_for_each_entry_rcu_continue(tpos, pos, (tbl)->buckets[hash],\
tbl, hash, member) tbl, hash, member)
/**
* rhl_for_each_rcu - iterate over rcu hash table list
* @pos: the &struct rlist_head to use as a loop cursor.
* @list: the head of the list
*
* This hash chain list-traversal primitive should be used on the
* list returned by rhltable_lookup.
*/
#define rhl_for_each_rcu(pos, list) \
for (pos = list; pos; pos = rcu_dereference_raw(pos->next))
/**
* rhl_for_each_entry_rcu - iterate over rcu hash table list of given type
* @tpos: the type * to use as a loop cursor.
* @pos: the &struct rlist_head to use as a loop cursor.
* @list: the head of the list
* @member: name of the &struct rlist_head within the hashable struct.
*
* This hash chain list-traversal primitive should be used on the
* list returned by rhltable_lookup.
*/
#define rhl_for_each_entry_rcu(tpos, pos, list, member) \
for (pos = list; pos && rht_entry(tpos, pos, member); \
pos = rcu_dereference_raw(pos->next))
static inline int rhashtable_compare(struct rhashtable_compare_arg *arg, static inline int rhashtable_compare(struct rhashtable_compare_arg *arg,
const void *obj) const void *obj)
{ {
...@@ -516,18 +556,8 @@ static inline int rhashtable_compare(struct rhashtable_compare_arg *arg, ...@@ -516,18 +556,8 @@ static inline int rhashtable_compare(struct rhashtable_compare_arg *arg,
return memcmp(ptr + ht->p.key_offset, arg->key, ht->p.key_len); return memcmp(ptr + ht->p.key_offset, arg->key, ht->p.key_len);
} }
/** /* Internal function, do not use. */
* rhashtable_lookup_fast - search hash table, inlined version static inline struct rhash_head *__rhashtable_lookup(
* @ht: hash table
* @key: the pointer to the key
* @params: hash table parameters
*
* Computes the hash value for the key and traverses the bucket chain looking
* for a entry with an identical key. The first matching entry is returned.
*
* Returns the first entry on which the compare function returned true.
*/
static inline void *rhashtable_lookup_fast(
struct rhashtable *ht, const void *key, struct rhashtable *ht, const void *key,
const struct rhashtable_params params) const struct rhashtable_params params)
{ {
...@@ -539,8 +569,6 @@ static inline void *rhashtable_lookup_fast( ...@@ -539,8 +569,6 @@ static inline void *rhashtable_lookup_fast(
struct rhash_head *he; struct rhash_head *he;
unsigned int hash; unsigned int hash;
rcu_read_lock();
tbl = rht_dereference_rcu(ht->tbl, ht); tbl = rht_dereference_rcu(ht->tbl, ht);
restart: restart:
hash = rht_key_hashfn(ht, tbl, key, params); hash = rht_key_hashfn(ht, tbl, key, params);
...@@ -549,8 +577,7 @@ static inline void *rhashtable_lookup_fast( ...@@ -549,8 +577,7 @@ static inline void *rhashtable_lookup_fast(
params.obj_cmpfn(&arg, rht_obj(ht, he)) : params.obj_cmpfn(&arg, rht_obj(ht, he)) :
rhashtable_compare(&arg, rht_obj(ht, he))) rhashtable_compare(&arg, rht_obj(ht, he)))
continue; continue;
rcu_read_unlock(); return he;
return rht_obj(ht, he);
} }
/* Ensure we see any new tables. */ /* Ensure we see any new tables. */
...@@ -559,96 +586,165 @@ static inline void *rhashtable_lookup_fast( ...@@ -559,96 +586,165 @@ static inline void *rhashtable_lookup_fast(
tbl = rht_dereference_rcu(tbl->future_tbl, ht); tbl = rht_dereference_rcu(tbl->future_tbl, ht);
if (unlikely(tbl)) if (unlikely(tbl))
goto restart; goto restart;
rcu_read_unlock();
return NULL; return NULL;
} }
/**
* rhashtable_lookup - search hash table
* @ht: hash table
* @key: the pointer to the key
* @params: hash table parameters
*
* Computes the hash value for the key and traverses the bucket chain looking
* for a entry with an identical key. The first matching entry is returned.
*
* This must only be called under the RCU read lock.
*
* Returns the first entry on which the compare function returned true.
*/
static inline void *rhashtable_lookup(
struct rhashtable *ht, const void *key,
const struct rhashtable_params params)
{
struct rhash_head *he = __rhashtable_lookup(ht, key, params);
return he ? rht_obj(ht, he) : NULL;
}
/**
* rhashtable_lookup_fast - search hash table, without RCU read lock
* @ht: hash table
* @key: the pointer to the key
* @params: hash table parameters
*
* Computes the hash value for the key and traverses the bucket chain looking
* for a entry with an identical key. The first matching entry is returned.
*
* Only use this function when you have other mechanisms guaranteeing
* that the object won't go away after the RCU read lock is released.
*
* Returns the first entry on which the compare function returned true.
*/
static inline void *rhashtable_lookup_fast(
struct rhashtable *ht, const void *key,
const struct rhashtable_params params)
{
void *obj;
rcu_read_lock();
obj = rhashtable_lookup(ht, key, params);
rcu_read_unlock();
return obj;
}
/**
* rhltable_lookup - search hash list table
* @hlt: hash table
* @key: the pointer to the key
* @params: hash table parameters
*
* Computes the hash value for the key and traverses the bucket chain looking
* for a entry with an identical key. All matching entries are returned
* in a list.
*
* This must only be called under the RCU read lock.
*
* Returns the list of entries that match the given key.
*/
static inline struct rhlist_head *rhltable_lookup(
struct rhltable *hlt, const void *key,
const struct rhashtable_params params)
{
struct rhash_head *he = __rhashtable_lookup(&hlt->ht, key, params);
return he ? container_of(he, struct rhlist_head, rhead) : NULL;
}
/* Internal function, please use rhashtable_insert_fast() instead. This /* Internal function, please use rhashtable_insert_fast() instead. This
* function returns the existing element already in hashes in there is a clash, * function returns the existing element already in hashes in there is a clash,
* otherwise it returns an error via ERR_PTR(). * otherwise it returns an error via ERR_PTR().
*/ */
static inline void *__rhashtable_insert_fast( static inline void *__rhashtable_insert_fast(
struct rhashtable *ht, const void *key, struct rhash_head *obj, struct rhashtable *ht, const void *key, struct rhash_head *obj,
const struct rhashtable_params params) const struct rhashtable_params params, bool rhlist)
{ {
struct rhashtable_compare_arg arg = { struct rhashtable_compare_arg arg = {
.ht = ht, .ht = ht,
.key = key, .key = key,
}; };
struct bucket_table *tbl, *new_tbl; struct rhash_head __rcu **pprev;
struct bucket_table *tbl;
struct rhash_head *head; struct rhash_head *head;
spinlock_t *lock; spinlock_t *lock;
unsigned int elasticity;
unsigned int hash; unsigned int hash;
void *data = NULL; int elasticity;
int err; void *data;
restart:
rcu_read_lock(); rcu_read_lock();
tbl = rht_dereference_rcu(ht->tbl, ht); tbl = rht_dereference_rcu(ht->tbl, ht);
hash = rht_head_hashfn(ht, tbl, obj, params);
lock = rht_bucket_lock(tbl, hash);
spin_lock_bh(lock);
/* All insertions must grab the oldest table containing if (unlikely(rht_dereference_bucket(tbl->future_tbl, tbl, hash))) {
* the hashed bucket that is yet to be rehashed. slow_path:
*/
for (;;) {
hash = rht_head_hashfn(ht, tbl, obj, params);
lock = rht_bucket_lock(tbl, hash);
spin_lock_bh(lock);
if (tbl->rehash <= hash)
break;
spin_unlock_bh(lock); spin_unlock_bh(lock);
tbl = rht_dereference_rcu(tbl->future_tbl, ht); rcu_read_unlock();
return rhashtable_insert_slow(ht, key, obj);
} }
new_tbl = rht_dereference_rcu(tbl->future_tbl, ht); elasticity = ht->elasticity;
if (unlikely(new_tbl)) { pprev = &tbl->buckets[hash];
tbl = rhashtable_insert_slow(ht, key, obj, new_tbl, &data); rht_for_each(head, tbl, hash) {
if (!IS_ERR_OR_NULL(tbl)) struct rhlist_head *plist;
goto slow_path; struct rhlist_head *list;
elasticity--;
if (!key ||
(params.obj_cmpfn ?
params.obj_cmpfn(&arg, rht_obj(ht, head)) :
rhashtable_compare(&arg, rht_obj(ht, head))))
continue;
data = rht_obj(ht, head);
err = PTR_ERR(tbl); if (!rhlist)
if (err == -EEXIST) goto out;
err = 0;
goto out;
}
err = -E2BIG; list = container_of(obj, struct rhlist_head, rhead);
if (unlikely(rht_grow_above_max(ht, tbl))) plist = container_of(head, struct rhlist_head, rhead);
goto out;
if (unlikely(rht_grow_above_100(ht, tbl))) { RCU_INIT_POINTER(list->next, plist);
slow_path: head = rht_dereference_bucket(head->next, tbl, hash);
spin_unlock_bh(lock); RCU_INIT_POINTER(list->rhead.next, head);
err = rhashtable_insert_rehash(ht, tbl); rcu_assign_pointer(*pprev, obj);
rcu_read_unlock();
if (err)
return ERR_PTR(err);
goto restart; goto good;
} }
err = 0; if (elasticity <= 0)
elasticity = ht->elasticity; goto slow_path;
rht_for_each(head, tbl, hash) {
if (key && data = ERR_PTR(-E2BIG);
unlikely(!(params.obj_cmpfn ? if (unlikely(rht_grow_above_max(ht, tbl)))
params.obj_cmpfn(&arg, rht_obj(ht, head)) : goto out;
rhashtable_compare(&arg, rht_obj(ht, head))))) {
data = rht_obj(ht, head); if (unlikely(rht_grow_above_100(ht, tbl)))
goto out; goto slow_path;
}
if (!--elasticity)
goto slow_path;
}
head = rht_dereference_bucket(tbl->buckets[hash], tbl, hash); head = rht_dereference_bucket(tbl->buckets[hash], tbl, hash);
RCU_INIT_POINTER(obj->next, head); RCU_INIT_POINTER(obj->next, head);
if (rhlist) {
struct rhlist_head *list;
list = container_of(obj, struct rhlist_head, rhead);
RCU_INIT_POINTER(list->next, NULL);
}
rcu_assign_pointer(tbl->buckets[hash], obj); rcu_assign_pointer(tbl->buckets[hash], obj);
...@@ -656,11 +752,14 @@ static inline void *__rhashtable_insert_fast( ...@@ -656,11 +752,14 @@ static inline void *__rhashtable_insert_fast(
if (rht_grow_above_75(ht, tbl)) if (rht_grow_above_75(ht, tbl))
schedule_work(&ht->run_work); schedule_work(&ht->run_work);
good:
data = NULL;
out: out:
spin_unlock_bh(lock); spin_unlock_bh(lock);
rcu_read_unlock(); rcu_read_unlock();
return err ? ERR_PTR(err) : data; return data;
} }
/** /**
...@@ -685,13 +784,65 @@ static inline int rhashtable_insert_fast( ...@@ -685,13 +784,65 @@ static inline int rhashtable_insert_fast(
{ {
void *ret; void *ret;
ret = __rhashtable_insert_fast(ht, NULL, obj, params); ret = __rhashtable_insert_fast(ht, NULL, obj, params, false);
if (IS_ERR(ret)) if (IS_ERR(ret))
return PTR_ERR(ret); return PTR_ERR(ret);
return ret == NULL ? 0 : -EEXIST; return ret == NULL ? 0 : -EEXIST;
} }
/**
* rhltable_insert_key - insert object into hash list table
* @hlt: hash list table
* @key: the pointer to the key
* @list: pointer to hash list head inside object
* @params: hash table parameters
*
* Will take a per bucket spinlock to protect against mutual mutations
* on the same bucket. Multiple insertions may occur in parallel unless
* they map to the same bucket lock.
*
* It is safe to call this function from atomic context.
*
* Will trigger an automatic deferred table resizing if the size grows
* beyond the watermark indicated by grow_decision() which can be passed
* to rhashtable_init().
*/
static inline int rhltable_insert_key(
struct rhltable *hlt, const void *key, struct rhlist_head *list,
const struct rhashtable_params params)
{
return PTR_ERR(__rhashtable_insert_fast(&hlt->ht, key, &list->rhead,
params, true));
}
/**
* rhltable_insert - insert object into hash list table
* @hlt: hash list table
* @list: pointer to hash list head inside object
* @params: hash table parameters
*
* Will take a per bucket spinlock to protect against mutual mutations
* on the same bucket. Multiple insertions may occur in parallel unless
* they map to the same bucket lock.
*
* It is safe to call this function from atomic context.
*
* Will trigger an automatic deferred table resizing if the size grows
* beyond the watermark indicated by grow_decision() which can be passed
* to rhashtable_init().
*/
static inline int rhltable_insert(
struct rhltable *hlt, struct rhlist_head *list,
const struct rhashtable_params params)
{
const char *key = rht_obj(&hlt->ht, &list->rhead);
key += params.key_offset;
return rhltable_insert_key(hlt, key, list, params);
}
/** /**
* rhashtable_lookup_insert_fast - lookup and insert object into hash table * rhashtable_lookup_insert_fast - lookup and insert object into hash table
* @ht: hash table * @ht: hash table
...@@ -722,7 +873,8 @@ static inline int rhashtable_lookup_insert_fast( ...@@ -722,7 +873,8 @@ static inline int rhashtable_lookup_insert_fast(
BUG_ON(ht->p.obj_hashfn); BUG_ON(ht->p.obj_hashfn);
ret = __rhashtable_insert_fast(ht, key + ht->p.key_offset, obj, params); ret = __rhashtable_insert_fast(ht, key + ht->p.key_offset, obj, params,
false);
if (IS_ERR(ret)) if (IS_ERR(ret))
return PTR_ERR(ret); return PTR_ERR(ret);
...@@ -759,7 +911,7 @@ static inline int rhashtable_lookup_insert_key( ...@@ -759,7 +911,7 @@ static inline int rhashtable_lookup_insert_key(
BUG_ON(!ht->p.obj_hashfn || !key); BUG_ON(!ht->p.obj_hashfn || !key);
ret = __rhashtable_insert_fast(ht, key, obj, params); ret = __rhashtable_insert_fast(ht, key, obj, params, false);
if (IS_ERR(ret)) if (IS_ERR(ret))
return PTR_ERR(ret); return PTR_ERR(ret);
...@@ -783,13 +935,14 @@ static inline void *rhashtable_lookup_get_insert_key( ...@@ -783,13 +935,14 @@ static inline void *rhashtable_lookup_get_insert_key(
{ {
BUG_ON(!ht->p.obj_hashfn || !key); BUG_ON(!ht->p.obj_hashfn || !key);
return __rhashtable_insert_fast(ht, key, obj, params); return __rhashtable_insert_fast(ht, key, obj, params, false);
} }
/* Internal function, please use rhashtable_remove_fast() instead */ /* Internal function, please use rhashtable_remove_fast() instead */
static inline int __rhashtable_remove_fast( static inline int __rhashtable_remove_fast_one(
struct rhashtable *ht, struct bucket_table *tbl, struct rhashtable *ht, struct bucket_table *tbl,
struct rhash_head *obj, const struct rhashtable_params params) struct rhash_head *obj, const struct rhashtable_params params,
bool rhlist)
{ {
struct rhash_head __rcu **pprev; struct rhash_head __rcu **pprev;
struct rhash_head *he; struct rhash_head *he;
...@@ -804,39 +957,66 @@ static inline int __rhashtable_remove_fast( ...@@ -804,39 +957,66 @@ static inline int __rhashtable_remove_fast(
pprev = &tbl->buckets[hash]; pprev = &tbl->buckets[hash];
rht_for_each(he, tbl, hash) { rht_for_each(he, tbl, hash) {
struct rhlist_head *list;
list = container_of(he, struct rhlist_head, rhead);
if (he != obj) { if (he != obj) {
struct rhlist_head __rcu **lpprev;
pprev = &he->next; pprev = &he->next;
continue;
if (!rhlist)
continue;
do {
lpprev = &list->next;
list = rht_dereference_bucket(list->next,
tbl, hash);
} while (list && obj != &list->rhead);
if (!list)
continue;
list = rht_dereference_bucket(list->next, tbl, hash);
RCU_INIT_POINTER(*lpprev, list);
err = 0;
break;
} }
rcu_assign_pointer(*pprev, obj->next); obj = rht_dereference_bucket(obj->next, tbl, hash);
err = 0; err = 1;
if (rhlist) {
list = rht_dereference_bucket(list->next, tbl, hash);
if (list) {
RCU_INIT_POINTER(list->rhead.next, obj);
obj = &list->rhead;
err = 0;
}
}
rcu_assign_pointer(*pprev, obj);
break; break;
} }
spin_unlock_bh(lock); spin_unlock_bh(lock);
if (err > 0) {
atomic_dec(&ht->nelems);
if (unlikely(ht->p.automatic_shrinking &&
rht_shrink_below_30(ht, tbl)))
schedule_work(&ht->run_work);
err = 0;
}
return err; return err;
} }
/** /* Internal function, please use rhashtable_remove_fast() instead */
* rhashtable_remove_fast - remove object from hash table static inline int __rhashtable_remove_fast(
* @ht: hash table
* @obj: pointer to hash head inside object
* @params: hash table parameters
*
* Since the hash chain is single linked, the removal operation needs to
* walk the bucket chain upon removal. The removal operation is thus
* considerable slow if the hash table is not correctly sized.
*
* Will automatically shrink the table via rhashtable_expand() if the
* shrink_decision function specified at rhashtable_init() returns true.
*
* Returns zero on success, -ENOENT if the entry could not be found.
*/
static inline int rhashtable_remove_fast(
struct rhashtable *ht, struct rhash_head *obj, struct rhashtable *ht, struct rhash_head *obj,
const struct rhashtable_params params) const struct rhashtable_params params, bool rhlist)
{ {
struct bucket_table *tbl; struct bucket_table *tbl;
int err; int err;
...@@ -850,24 +1030,60 @@ static inline int rhashtable_remove_fast( ...@@ -850,24 +1030,60 @@ static inline int rhashtable_remove_fast(
* visible then that guarantees the entry to still be in * visible then that guarantees the entry to still be in
* the old tbl if it exists. * the old tbl if it exists.
*/ */
while ((err = __rhashtable_remove_fast(ht, tbl, obj, params)) && while ((err = __rhashtable_remove_fast_one(ht, tbl, obj, params,
rhlist)) &&
(tbl = rht_dereference_rcu(tbl->future_tbl, ht))) (tbl = rht_dereference_rcu(tbl->future_tbl, ht)))
; ;
if (err)
goto out;
atomic_dec(&ht->nelems);
if (unlikely(ht->p.automatic_shrinking &&
rht_shrink_below_30(ht, tbl)))
schedule_work(&ht->run_work);
out:
rcu_read_unlock(); rcu_read_unlock();
return err; return err;
} }
/**
* rhashtable_remove_fast - remove object from hash table
* @ht: hash table
* @obj: pointer to hash head inside object
* @params: hash table parameters
*
* Since the hash chain is single linked, the removal operation needs to
* walk the bucket chain upon removal. The removal operation is thus
* considerable slow if the hash table is not correctly sized.
*
* Will automatically shrink the table via rhashtable_expand() if the
* shrink_decision function specified at rhashtable_init() returns true.
*
* Returns zero on success, -ENOENT if the entry could not be found.
*/
static inline int rhashtable_remove_fast(
struct rhashtable *ht, struct rhash_head *obj,
const struct rhashtable_params params)
{
return __rhashtable_remove_fast(ht, obj, params, false);
}
/**
* rhltable_remove - remove object from hash list table
* @hlt: hash list table
* @list: pointer to hash list head inside object
* @params: hash table parameters
*
* Since the hash chain is single linked, the removal operation needs to
* walk the bucket chain upon removal. The removal operation is thus
* considerable slow if the hash table is not correctly sized.
*
* Will automatically shrink the table via rhashtable_expand() if the
* shrink_decision function specified at rhashtable_init() returns true.
*
* Returns zero on success, -ENOENT if the entry could not be found.
*/
static inline int rhltable_remove(
struct rhltable *hlt, struct rhlist_head *list,
const struct rhashtable_params params)
{
return __rhashtable_remove_fast(&hlt->ht, &list->rhead, params, true);
}
/* Internal function, please use rhashtable_replace_fast() instead */ /* Internal function, please use rhashtable_replace_fast() instead */
static inline int __rhashtable_replace_fast( static inline int __rhashtable_replace_fast(
struct rhashtable *ht, struct bucket_table *tbl, struct rhashtable *ht, struct bucket_table *tbl,
...@@ -958,4 +1174,51 @@ static inline int rhashtable_walk_init(struct rhashtable *ht, ...@@ -958,4 +1174,51 @@ static inline int rhashtable_walk_init(struct rhashtable *ht,
return 0; return 0;
} }
/**
* rhltable_walk_enter - Initialise an iterator
* @hlt: Table to walk over
* @iter: Hash table Iterator
*
* This function prepares a hash table walk.
*
* Note that if you restart a walk after rhashtable_walk_stop you
* may see the same object twice. Also, you may miss objects if
* there are removals in between rhashtable_walk_stop and the next
* call to rhashtable_walk_start.
*
* For a completely stable walk you should construct your own data
* structure outside the hash table.
*
* This function may sleep so you must not call it from interrupt
* context or with spin locks held.
*
* You must call rhashtable_walk_exit after this function returns.
*/
static inline void rhltable_walk_enter(struct rhltable *hlt,
struct rhashtable_iter *iter)
{
return rhashtable_walk_enter(&hlt->ht, iter);
}
/**
* rhltable_free_and_destroy - free elements and destroy hash list table
* @hlt: the hash list table to destroy
* @free_fn: callback to release resources of element
* @arg: pointer passed to free_fn
*
* See documentation for rhashtable_free_and_destroy.
*/
static inline void rhltable_free_and_destroy(struct rhltable *hlt,
void (*free_fn)(void *ptr,
void *arg),
void *arg)
{
return rhashtable_free_and_destroy(&hlt->ht, free_fn, arg);
}
static inline void rhltable_destroy(struct rhltable *hlt)
{
return rhltable_free_and_destroy(hlt, NULL, NULL);
}
#endif /* _LINUX_RHASHTABLE_H */ #endif /* _LINUX_RHASHTABLE_H */
...@@ -378,22 +378,8 @@ static void rht_deferred_worker(struct work_struct *work) ...@@ -378,22 +378,8 @@ static void rht_deferred_worker(struct work_struct *work)
schedule_work(&ht->run_work); schedule_work(&ht->run_work);
} }
static bool rhashtable_check_elasticity(struct rhashtable *ht, static int rhashtable_insert_rehash(struct rhashtable *ht,
struct bucket_table *tbl, struct bucket_table *tbl)
unsigned int hash)
{
unsigned int elasticity = ht->elasticity;
struct rhash_head *head;
rht_for_each(head, tbl, hash)
if (!--elasticity)
return true;
return false;
}
int rhashtable_insert_rehash(struct rhashtable *ht,
struct bucket_table *tbl)
{ {
struct bucket_table *old_tbl; struct bucket_table *old_tbl;
struct bucket_table *new_tbl; struct bucket_table *new_tbl;
...@@ -439,57 +425,165 @@ int rhashtable_insert_rehash(struct rhashtable *ht, ...@@ -439,57 +425,165 @@ int rhashtable_insert_rehash(struct rhashtable *ht,
return err; return err;
} }
EXPORT_SYMBOL_GPL(rhashtable_insert_rehash);
struct bucket_table *rhashtable_insert_slow(struct rhashtable *ht, static void *rhashtable_lookup_one(struct rhashtable *ht,
const void *key, struct bucket_table *tbl, unsigned int hash,
struct rhash_head *obj, const void *key, struct rhash_head *obj)
struct bucket_table *tbl,
void **data)
{ {
struct rhashtable_compare_arg arg = {
.ht = ht,
.key = key,
};
struct rhash_head __rcu **pprev;
struct rhash_head *head; struct rhash_head *head;
unsigned int hash; int elasticity;
int err;
tbl = rhashtable_last_table(ht, tbl); elasticity = ht->elasticity;
hash = head_hashfn(ht, tbl, obj); pprev = &tbl->buckets[hash];
spin_lock_nested(rht_bucket_lock(tbl, hash), SINGLE_DEPTH_NESTING); rht_for_each(head, tbl, hash) {
struct rhlist_head *list;
err = -EEXIST; struct rhlist_head *plist;
if (key) {
*data = rhashtable_lookup_fast(ht, key, ht->p); elasticity--;
if (*data) if (!key ||
goto exit; (ht->p.obj_cmpfn ?
ht->p.obj_cmpfn(&arg, rht_obj(ht, head)) :
rhashtable_compare(&arg, rht_obj(ht, head))))
continue;
if (!ht->rhlist)
return rht_obj(ht, head);
list = container_of(obj, struct rhlist_head, rhead);
plist = container_of(head, struct rhlist_head, rhead);
RCU_INIT_POINTER(list->next, plist);
head = rht_dereference_bucket(head->next, tbl, hash);
RCU_INIT_POINTER(list->rhead.next, head);
rcu_assign_pointer(*pprev, obj);
return NULL;
} }
err = -E2BIG; if (elasticity <= 0)
if (unlikely(rht_grow_above_max(ht, tbl))) return ERR_PTR(-EAGAIN);
goto exit;
return ERR_PTR(-ENOENT);
}
static struct bucket_table *rhashtable_insert_one(struct rhashtable *ht,
struct bucket_table *tbl,
unsigned int hash,
struct rhash_head *obj,
void *data)
{
struct bucket_table *new_tbl;
struct rhash_head *head;
if (!IS_ERR_OR_NULL(data))
return ERR_PTR(-EEXIST);
err = -EAGAIN; if (PTR_ERR(data) != -EAGAIN && PTR_ERR(data) != -ENOENT)
if (rhashtable_check_elasticity(ht, tbl, hash) || return ERR_CAST(data);
rht_grow_above_100(ht, tbl))
goto exit;
err = 0; new_tbl = rcu_dereference(tbl->future_tbl);
if (new_tbl)
return new_tbl;
if (PTR_ERR(data) != -ENOENT)
return ERR_CAST(data);
if (unlikely(rht_grow_above_max(ht, tbl)))
return ERR_PTR(-E2BIG);
if (unlikely(rht_grow_above_100(ht, tbl)))
return ERR_PTR(-EAGAIN);
head = rht_dereference_bucket(tbl->buckets[hash], tbl, hash); head = rht_dereference_bucket(tbl->buckets[hash], tbl, hash);
RCU_INIT_POINTER(obj->next, head); RCU_INIT_POINTER(obj->next, head);
if (ht->rhlist) {
struct rhlist_head *list;
list = container_of(obj, struct rhlist_head, rhead);
RCU_INIT_POINTER(list->next, NULL);
}
rcu_assign_pointer(tbl->buckets[hash], obj); rcu_assign_pointer(tbl->buckets[hash], obj);
atomic_inc(&ht->nelems); atomic_inc(&ht->nelems);
if (rht_grow_above_75(ht, tbl))
schedule_work(&ht->run_work);
exit: return NULL;
spin_unlock(rht_bucket_lock(tbl, hash)); }
if (err == 0) static void *rhashtable_try_insert(struct rhashtable *ht, const void *key,
return NULL; struct rhash_head *obj)
else if (err == -EAGAIN) {
return tbl; struct bucket_table *new_tbl;
else struct bucket_table *tbl;
return ERR_PTR(err); unsigned int hash;
spinlock_t *lock;
void *data;
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 = rcu_dereference(tbl->future_tbl);
}
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)) {
tbl = new_tbl;
hash = rht_head_hashfn(ht, tbl, obj, ht->p);
spin_lock_nested(rht_bucket_lock(tbl, hash),
SINGLE_DEPTH_NESTING);
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);
spin_unlock(rht_bucket_lock(tbl, hash));
}
spin_unlock_bh(lock);
if (PTR_ERR(data) == -EAGAIN)
data = ERR_PTR(rhashtable_insert_rehash(ht, tbl) ?:
-EAGAIN);
return data;
}
void *rhashtable_insert_slow(struct rhashtable *ht, const void *key,
struct rhash_head *obj)
{
void *data;
do {
rcu_read_lock();
data = rhashtable_try_insert(ht, key, obj);
rcu_read_unlock();
} while (PTR_ERR(data) == -EAGAIN);
return data;
} }
EXPORT_SYMBOL_GPL(rhashtable_insert_slow); EXPORT_SYMBOL_GPL(rhashtable_insert_slow);
...@@ -593,11 +687,16 @@ EXPORT_SYMBOL_GPL(rhashtable_walk_start); ...@@ -593,11 +687,16 @@ EXPORT_SYMBOL_GPL(rhashtable_walk_start);
void *rhashtable_walk_next(struct rhashtable_iter *iter) void *rhashtable_walk_next(struct rhashtable_iter *iter)
{ {
struct bucket_table *tbl = iter->walker.tbl; struct bucket_table *tbl = iter->walker.tbl;
struct rhlist_head *list = iter->list;
struct rhashtable *ht = iter->ht; struct rhashtable *ht = iter->ht;
struct rhash_head *p = iter->p; struct rhash_head *p = iter->p;
bool rhlist = ht->rhlist;
if (p) { if (p) {
p = rht_dereference_bucket_rcu(p->next, tbl, iter->slot); if (!rhlist || !(list = rcu_dereference(list->next))) {
p = rcu_dereference(p->next);
list = container_of(p, struct rhlist_head, rhead);
}
goto next; goto next;
} }
...@@ -605,6 +704,18 @@ void *rhashtable_walk_next(struct rhashtable_iter *iter) ...@@ -605,6 +704,18 @@ void *rhashtable_walk_next(struct rhashtable_iter *iter)
int skip = iter->skip; int skip = iter->skip;
rht_for_each_rcu(p, tbl, iter->slot) { rht_for_each_rcu(p, tbl, iter->slot) {
if (rhlist) {
list = container_of(p, struct rhlist_head,
rhead);
do {
if (!skip)
goto next;
skip--;
list = rcu_dereference(list->next);
} while (list);
continue;
}
if (!skip) if (!skip)
break; break;
skip--; skip--;
...@@ -614,7 +725,8 @@ void *rhashtable_walk_next(struct rhashtable_iter *iter) ...@@ -614,7 +725,8 @@ void *rhashtable_walk_next(struct rhashtable_iter *iter)
if (!rht_is_a_nulls(p)) { if (!rht_is_a_nulls(p)) {
iter->skip++; iter->skip++;
iter->p = p; iter->p = p;
return rht_obj(ht, p); iter->list = list;
return rht_obj(ht, rhlist ? &list->rhead : p);
} }
iter->skip = 0; iter->skip = 0;
...@@ -802,6 +914,48 @@ int rhashtable_init(struct rhashtable *ht, ...@@ -802,6 +914,48 @@ int rhashtable_init(struct rhashtable *ht,
} }
EXPORT_SYMBOL_GPL(rhashtable_init); EXPORT_SYMBOL_GPL(rhashtable_init);
/**
* rhltable_init - initialize a new hash list table
* @hlt: hash list table to be initialized
* @params: configuration parameters
*
* Initializes a new hash list table.
*
* See documentation for rhashtable_init.
*/
int rhltable_init(struct rhltable *hlt, const struct rhashtable_params *params)
{
int err;
/* No rhlist NULLs marking for now. */
if (params->nulls_base)
return -EINVAL;
err = rhashtable_init(&hlt->ht, params);
hlt->ht.rhlist = true;
return err;
}
EXPORT_SYMBOL_GPL(rhltable_init);
static void rhashtable_free_one(struct rhashtable *ht, struct rhash_head *obj,
void (*free_fn)(void *ptr, void *arg),
void *arg)
{
struct rhlist_head *list;
if (!ht->rhlist) {
free_fn(rht_obj(ht, obj), arg);
return;
}
list = container_of(obj, struct rhlist_head, rhead);
do {
obj = &list->rhead;
list = rht_dereference(list->next, ht);
free_fn(rht_obj(ht, obj), arg);
} while (list);
}
/** /**
* rhashtable_free_and_destroy - free elements and destroy hash table * rhashtable_free_and_destroy - free elements and destroy hash table
* @ht: the hash table to destroy * @ht: the hash table to destroy
...@@ -839,7 +993,7 @@ void rhashtable_free_and_destroy(struct rhashtable *ht, ...@@ -839,7 +993,7 @@ void rhashtable_free_and_destroy(struct rhashtable *ht,
pos = next, pos = next,
next = !rht_is_a_nulls(pos) ? next = !rht_is_a_nulls(pos) ?
rht_dereference(pos->next, ht) : NULL) rht_dereference(pos->next, ht) : NULL)
free_fn(rht_obj(ht, pos), arg); rhashtable_free_one(ht, pos, free_fn, arg);
} }
} }
......
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