1/* SPDX-License-Identifier: GPL-2.0 */
2/*
3 * Resizable, Scalable, Concurrent Hash Table
4 *
5 * Copyright (c) 2015-2016 Herbert Xu <herbert@gondor.apana.org.au>
6 * Copyright (c) 2014-2015 Thomas Graf <tgraf@suug.ch>
7 * Copyright (c) 2008-2014 Patrick McHardy <kaber@trash.net>
8 *
9 * Code partially derived from nft_hash
10 * Rewritten with rehash code from br_multicast plus single list
11 * pointer as suggested by Josh Triplett
12 *
13 * This program is free software; you can redistribute it and/or modify
14 * it under the terms of the GNU General Public License version 2 as
15 * published by the Free Software Foundation.
16 */
17
18#ifndef _LINUX_RHASHTABLE_H
19#define _LINUX_RHASHTABLE_H
20
21#include <linux/err.h>
22#include <linux/errno.h>
23#include <linux/jhash.h>
24#include <linux/list_nulls.h>
25#include <linux/workqueue.h>
26#include <linux/rculist.h>
27
28#include <linux/rhashtable-types.h>
29/*
30 * The end of the chain is marked with a special nulls marks which has
31 * the least significant bit set.
32 */
33
34/* Maximum chain length before rehash
35 *
36 * The maximum (not average) chain length grows with the size of the hash
37 * table, at a rate of (log N)/(log log N).
38 *
39 * The value of 16 is selected so that even if the hash table grew to
40 * 2^32 you would not expect the maximum chain length to exceed it
41 * unless we are under attack (or extremely unlucky).
42 *
43 * As this limit is only to detect attacks, we don't need to set it to a
44 * lower value as you'd need the chain length to vastly exceed 16 to have
45 * any real effect on the system.
46 */
47#define RHT_ELASTICITY 16u
48
49/**
50 * struct bucket_table - Table of hash buckets
51 * @size: Number of hash buckets
52 * @nest: Number of bits of first-level nested table.
53 * @rehash: Current bucket being rehashed
54 * @hash_rnd: Random seed to fold into hash
55 * @locks_mask: Mask to apply before accessing locks[]
56 * @locks: Array of spinlocks protecting individual buckets
57 * @walkers: List of active walkers
58 * @rcu: RCU structure for freeing the table
59 * @future_tbl: Table under construction during rehashing
60 * @ntbl: Nested table used when out of memory.
61 * @buckets: size * hash buckets
62 */
63struct bucket_table {
64 unsigned int size;
65 unsigned int nest;
66 unsigned int rehash;
67 u32 hash_rnd;
68 unsigned int locks_mask;
69 spinlock_t *locks;
70 struct list_head walkers;
71 struct rcu_head rcu;
72
73 struct bucket_table __rcu *future_tbl;
74
75 struct rhash_head __rcu *buckets[] ____cacheline_aligned_in_smp;
76};
77
78/*
79 * NULLS_MARKER() expects a hash value with the low
80 * bits mostly likely to be significant, and it discards
81 * the msb.
82 * We git it an address, in which the bottom 2 bits are
83 * always 0, and the msb might be significant.
84 * So we shift the address down one bit to align with
85 * expectations and avoid losing a significant bit.
86 */
87#define RHT_NULLS_MARKER(ptr) \
88 ((void *)NULLS_MARKER(((unsigned long) (ptr)) >> 1))
89#define INIT_RHT_NULLS_HEAD(ptr) \
90 ((ptr) = RHT_NULLS_MARKER(&(ptr)))
91
92static inline bool rht_is_a_nulls(const struct rhash_head *ptr)
93{
94 return ((unsigned long) ptr & 1);
95}
96
97static inline void *rht_obj(const struct rhashtable *ht,
98 const struct rhash_head *he)
99{
100 return (char *)he - ht->p.head_offset;
101}
102
103static inline unsigned int rht_bucket_index(const struct bucket_table *tbl,
104 unsigned int hash)
105{
106 return hash & (tbl->size - 1);
107}
108
109static inline unsigned int rht_key_get_hash(struct rhashtable *ht,
110 const void *key, const struct rhashtable_params params,
111 unsigned int hash_rnd)
112{
113 unsigned int hash;
114
115 /* params must be equal to ht->p if it isn't constant. */
116 if (!__builtin_constant_p(params.key_len))
117 hash = ht->p.hashfn(key, ht->key_len, hash_rnd);
118 else if (params.key_len) {
119 unsigned int key_len = params.key_len;
120
121 if (params.hashfn)
122 hash = params.hashfn(key, key_len, hash_rnd);
123 else if (key_len & (sizeof(u32) - 1))
124 hash = jhash(key, key_len, hash_rnd);
125 else
126 hash = jhash2(key, key_len / sizeof(u32), hash_rnd);
127 } else {
128 unsigned int key_len = ht->p.key_len;
129
130 if (params.hashfn)
131 hash = params.hashfn(key, key_len, hash_rnd);
132 else
133 hash = jhash(key, key_len, hash_rnd);
134 }
135
136 return hash;
137}
138
139static inline unsigned int rht_key_hashfn(
140 struct rhashtable *ht, const struct bucket_table *tbl,
141 const void *key, const struct rhashtable_params params)
142{
143 unsigned int hash = rht_key_get_hash(ht, key, params, tbl->hash_rnd);
144
145 return rht_bucket_index(tbl, hash);
146}
147
148static inline unsigned int rht_head_hashfn(
149 struct rhashtable *ht, const struct bucket_table *tbl,
150 const struct rhash_head *he, const struct rhashtable_params params)
151{
152 const char *ptr = rht_obj(ht, he);
153
154 return likely(params.obj_hashfn) ?
155 rht_bucket_index(tbl, params.obj_hashfn(ptr, params.key_len ?:
156 ht->p.key_len,
157 tbl->hash_rnd)) :
158 rht_key_hashfn(ht, tbl, ptr + params.key_offset, params);
159}
160
161/**
162 * rht_grow_above_75 - returns true if nelems > 0.75 * table-size
163 * @ht: hash table
164 * @tbl: current table
165 */
166static inline bool rht_grow_above_75(const struct rhashtable *ht,
167 const struct bucket_table *tbl)
168{
169 /* Expand table when exceeding 75% load */
170 return atomic_read(&ht->nelems) > (tbl->size / 4 * 3) &&
171 (!ht->p.max_size || tbl->size < ht->p.max_size);
172}
173
174/**
175 * rht_shrink_below_30 - returns true if nelems < 0.3 * table-size
176 * @ht: hash table
177 * @tbl: current table
178 */
179static inline bool rht_shrink_below_30(const struct rhashtable *ht,
180 const struct bucket_table *tbl)
181{
182 /* Shrink table beneath 30% load */
183 return atomic_read(&ht->nelems) < (tbl->size * 3 / 10) &&
184 tbl->size > ht->p.min_size;
185}
186
187/**
188 * rht_grow_above_100 - returns true if nelems > table-size
189 * @ht: hash table
190 * @tbl: current table
191 */
192static inline bool rht_grow_above_100(const struct rhashtable *ht,
193 const struct bucket_table *tbl)
194{
195 return atomic_read(&ht->nelems) > tbl->size &&
196 (!ht->p.max_size || tbl->size < ht->p.max_size);
197}
198
199/**
200 * rht_grow_above_max - returns true if table is above maximum
201 * @ht: hash table
202 * @tbl: current table
203 */
204static inline bool rht_grow_above_max(const struct rhashtable *ht,
205 const struct bucket_table *tbl)
206{
207 return atomic_read(&ht->nelems) >= ht->max_elems;
208}
209
210/* The bucket lock is selected based on the hash and protects mutations
211 * on a group of hash buckets.
212 *
213 * A maximum of tbl->size/2 bucket locks is allocated. This ensures that
214 * a single lock always covers both buckets which may both contains
215 * entries which link to the same bucket of the old table during resizing.
216 * This allows to simplify the locking as locking the bucket in both
217 * tables during resize always guarantee protection.
218 *
219 * IMPORTANT: When holding the bucket lock of both the old and new table
220 * during expansions and shrinking, the old bucket lock must always be
221 * acquired first.
222 */
223static inline spinlock_t *rht_bucket_lock(const struct bucket_table *tbl,
224 unsigned int hash)
225{
226 return &tbl->locks[hash & tbl->locks_mask];
227}
228
229#ifdef CONFIG_PROVE_LOCKING
230int lockdep_rht_mutex_is_held(struct rhashtable *ht);
231int lockdep_rht_bucket_is_held(const struct bucket_table *tbl, u32 hash);
232#else
233static inline int lockdep_rht_mutex_is_held(struct rhashtable *ht)
234{
235 return 1;
236}
237
238static inline int lockdep_rht_bucket_is_held(const struct bucket_table *tbl,
239 u32 hash)
240{
241 return 1;
242}
243#endif /* CONFIG_PROVE_LOCKING */
244
245void *rhashtable_insert_slow(struct rhashtable *ht, const void *key,
246 struct rhash_head *obj);
247
248void rhashtable_walk_enter(struct rhashtable *ht,
249 struct rhashtable_iter *iter);
250void rhashtable_walk_exit(struct rhashtable_iter *iter);
251int rhashtable_walk_start_check(struct rhashtable_iter *iter) __acquires(RCU);
252
253static inline void rhashtable_walk_start(struct rhashtable_iter *iter)
254{
255 (void)rhashtable_walk_start_check(iter);
256}
257
258void *rhashtable_walk_next(struct rhashtable_iter *iter);
259void *rhashtable_walk_peek(struct rhashtable_iter *iter);
260void rhashtable_walk_stop(struct rhashtable_iter *iter) __releases(RCU);
261
262void rhashtable_free_and_destroy(struct rhashtable *ht,
263 void (*free_fn)(void *ptr, void *arg),
264 void *arg);
265void rhashtable_destroy(struct rhashtable *ht);
266
267struct rhash_head __rcu **rht_bucket_nested(const struct bucket_table *tbl,
268 unsigned int hash);
269struct rhash_head __rcu **rht_bucket_nested_insert(struct rhashtable *ht,
270 struct bucket_table *tbl,
271 unsigned int hash);
272
273#define rht_dereference(p, ht) \
274 rcu_dereference_protected(p, lockdep_rht_mutex_is_held(ht))
275
276#define rht_dereference_rcu(p, ht) \
277 rcu_dereference_check(p, lockdep_rht_mutex_is_held(ht))
278
279#define rht_dereference_bucket(p, tbl, hash) \
280 rcu_dereference_protected(p, lockdep_rht_bucket_is_held(tbl, hash))
281
282#define rht_dereference_bucket_rcu(p, tbl, hash) \
283 rcu_dereference_check(p, lockdep_rht_bucket_is_held(tbl, hash))
284
285#define rht_entry(tpos, pos, member) \
286 ({ tpos = container_of(pos, typeof(*tpos), member); 1; })
287
288static inline struct rhash_head __rcu *const *rht_bucket(
289 const struct bucket_table *tbl, unsigned int hash)
290{
291 return unlikely(tbl->nest) ? rht_bucket_nested(tbl, hash) :
292 &tbl->buckets[hash];
293}
294
295static inline struct rhash_head __rcu **rht_bucket_var(
296 struct bucket_table *tbl, unsigned int hash)
297{
298 return unlikely(tbl->nest) ? rht_bucket_nested(tbl, hash) :
299 &tbl->buckets[hash];
300}
301
302static inline struct rhash_head __rcu **rht_bucket_insert(
303 struct rhashtable *ht, struct bucket_table *tbl, unsigned int hash)
304{
305 return unlikely(tbl->nest) ? rht_bucket_nested_insert(ht, tbl, hash) :
306 &tbl->buckets[hash];
307}
308
309/**
310 * rht_for_each_continue - continue iterating over hash chain
311 * @pos: the &struct rhash_head to use as a loop cursor.
312 * @head: the previous &struct rhash_head to continue from
313 * @tbl: the &struct bucket_table
314 * @hash: the hash value / bucket index
315 */
316#define rht_for_each_continue(pos, head, tbl, hash) \
317 for (pos = rht_dereference_bucket(head, tbl, hash); \
318 !rht_is_a_nulls(pos); \
319 pos = rht_dereference_bucket((pos)->next, tbl, hash))
320
321/**
322 * rht_for_each - iterate over hash chain
323 * @pos: the &struct rhash_head to use as a loop cursor.
324 * @tbl: the &struct bucket_table
325 * @hash: the hash value / bucket index
326 */
327#define rht_for_each(pos, tbl, hash) \
328 rht_for_each_continue(pos, *rht_bucket(tbl, hash), tbl, hash)
329
330/**
331 * rht_for_each_entry_continue - continue iterating over hash chain
332 * @tpos: the type * to use as a loop cursor.
333 * @pos: the &struct rhash_head to use as a loop cursor.
334 * @head: the previous &struct rhash_head to continue from
335 * @tbl: the &struct bucket_table
336 * @hash: the hash value / bucket index
337 * @member: name of the &struct rhash_head within the hashable struct.
338 */
339#define rht_for_each_entry_continue(tpos, pos, head, tbl, hash, member) \
340 for (pos = rht_dereference_bucket(head, tbl, hash); \
341 (!rht_is_a_nulls(pos)) && rht_entry(tpos, pos, member); \
342 pos = rht_dereference_bucket((pos)->next, tbl, hash))
343
344/**
345 * rht_for_each_entry - iterate over hash chain of given type
346 * @tpos: the type * to use as a loop cursor.
347 * @pos: the &struct rhash_head to use as a loop cursor.
348 * @tbl: the &struct bucket_table
349 * @hash: the hash value / bucket index
350 * @member: name of the &struct rhash_head within the hashable struct.
351 */
352#define rht_for_each_entry(tpos, pos, tbl, hash, member) \
353 rht_for_each_entry_continue(tpos, pos, *rht_bucket(tbl, hash), \
354 tbl, hash, member)
355
356/**
357 * rht_for_each_entry_safe - safely iterate over hash chain of given type
358 * @tpos: the type * to use as a loop cursor.
359 * @pos: the &struct rhash_head to use as a loop cursor.
360 * @next: the &struct rhash_head to use as next in loop cursor.
361 * @tbl: the &struct bucket_table
362 * @hash: the hash value / bucket index
363 * @member: name of the &struct rhash_head within the hashable struct.
364 *
365 * This hash chain list-traversal primitive allows for the looped code to
366 * remove the loop cursor from the list.
367 */
368#define rht_for_each_entry_safe(tpos, pos, next, tbl, hash, member) \
369 for (pos = rht_dereference_bucket(*rht_bucket(tbl, hash), tbl, hash), \
370 next = !rht_is_a_nulls(pos) ? \
371 rht_dereference_bucket(pos->next, tbl, hash) : NULL; \
372 (!rht_is_a_nulls(pos)) && rht_entry(tpos, pos, member); \
373 pos = next, \
374 next = !rht_is_a_nulls(pos) ? \
375 rht_dereference_bucket(pos->next, tbl, hash) : NULL)
376
377/**
378 * rht_for_each_rcu_continue - continue iterating over rcu hash chain
379 * @pos: the &struct rhash_head to use as a loop cursor.
380 * @head: the previous &struct rhash_head to continue from
381 * @tbl: the &struct bucket_table
382 * @hash: the hash value / bucket index
383 *
384 * This hash chain list-traversal primitive may safely run concurrently with
385 * the _rcu mutation primitives such as rhashtable_insert() as long as the
386 * traversal is guarded by rcu_read_lock().
387 */
388#define rht_for_each_rcu_continue(pos, head, tbl, hash) \
389 for (({barrier(); }), \
390 pos = rht_dereference_bucket_rcu(head, tbl, hash); \
391 !rht_is_a_nulls(pos); \
392 pos = rcu_dereference_raw(pos->next))
393
394/**
395 * rht_for_each_rcu - iterate over rcu hash chain
396 * @pos: the &struct rhash_head to use as a loop cursor.
397 * @tbl: the &struct bucket_table
398 * @hash: the hash value / bucket index
399 *
400 * This hash chain list-traversal primitive may safely run concurrently with
401 * the _rcu mutation primitives such as rhashtable_insert() as long as the
402 * traversal is guarded by rcu_read_lock().
403 */
404#define rht_for_each_rcu(pos, tbl, hash) \
405 rht_for_each_rcu_continue(pos, *rht_bucket(tbl, hash), tbl, hash)
406
407/**
408 * rht_for_each_entry_rcu_continue - continue iterating over rcu hash chain
409 * @tpos: the type * to use as a loop cursor.
410 * @pos: the &struct rhash_head to use as a loop cursor.
411 * @head: the previous &struct rhash_head to continue from
412 * @tbl: the &struct bucket_table
413 * @hash: the hash value / bucket index
414 * @member: name of the &struct rhash_head within the hashable struct.
415 *
416 * This hash chain list-traversal primitive may safely run concurrently with
417 * the _rcu mutation primitives such as rhashtable_insert() as long as the
418 * traversal is guarded by rcu_read_lock().
419 */
420#define rht_for_each_entry_rcu_continue(tpos, pos, head, tbl, hash, member) \
421 for (({barrier(); }), \
422 pos = rht_dereference_bucket_rcu(head, tbl, hash); \
423 (!rht_is_a_nulls(pos)) && rht_entry(tpos, pos, member); \
424 pos = rht_dereference_bucket_rcu(pos->next, tbl, hash))
425
426/**
427 * rht_for_each_entry_rcu - iterate over rcu hash chain of given type
428 * @tpos: the type * to use as a loop cursor.
429 * @pos: the &struct rhash_head to use as a loop cursor.
430 * @tbl: the &struct bucket_table
431 * @hash: the hash value / bucket index
432 * @member: name of the &struct rhash_head within the hashable struct.
433 *
434 * This hash chain list-traversal primitive may safely run concurrently with
435 * the _rcu mutation primitives such as rhashtable_insert() as long as the
436 * traversal is guarded by rcu_read_lock().
437 */
438#define rht_for_each_entry_rcu(tpos, pos, tbl, hash, member) \
439 rht_for_each_entry_rcu_continue(tpos, pos, *rht_bucket(tbl, hash), \
440 tbl, hash, member)
441
442/**
443 * rhl_for_each_rcu - iterate over rcu hash table list
444 * @pos: the &struct rlist_head to use as a loop cursor.
445 * @list: the head of the list
446 *
447 * This hash chain list-traversal primitive should be used on the
448 * list returned by rhltable_lookup.
449 */
450#define rhl_for_each_rcu(pos, list) \
451 for (pos = list; pos; pos = rcu_dereference_raw(pos->next))
452
453/**
454 * rhl_for_each_entry_rcu - iterate over rcu hash table list of given type
455 * @tpos: the type * to use as a loop cursor.
456 * @pos: the &struct rlist_head to use as a loop cursor.
457 * @list: the head of the list
458 * @member: name of the &struct rlist_head within the hashable struct.
459 *
460 * This hash chain list-traversal primitive should be used on the
461 * list returned by rhltable_lookup.
462 */
463#define rhl_for_each_entry_rcu(tpos, pos, list, member) \
464 for (pos = list; pos && rht_entry(tpos, pos, member); \
465 pos = rcu_dereference_raw(pos->next))
466
467static inline int rhashtable_compare(struct rhashtable_compare_arg *arg,
468 const void *obj)
469{
470 struct rhashtable *ht = arg->ht;
471 const char *ptr = obj;
472
473 return memcmp(ptr + ht->p.key_offset, arg->key, ht->p.key_len);
474}
475
476/* Internal function, do not use. */
477static inline struct rhash_head *__rhashtable_lookup(
478 struct rhashtable *ht, const void *key,
479 const struct rhashtable_params params)
480{
481 struct rhashtable_compare_arg arg = {
482 .ht = ht,
483 .key = key,
484 };
485 struct rhash_head __rcu * const *head;
486 struct bucket_table *tbl;
487 struct rhash_head *he;
488 unsigned int hash;
489
490 tbl = rht_dereference_rcu(ht->tbl, ht);
491restart:
492 hash = rht_key_hashfn(ht, tbl, key, params);
493 head = rht_bucket(tbl, hash);
494 do {
495 rht_for_each_rcu_continue(he, *head, tbl, hash) {
496 if (params.obj_cmpfn ?
497 params.obj_cmpfn(&arg, rht_obj(ht, he)) :
498 rhashtable_compare(&arg, rht_obj(ht, he)))
499 continue;
500 return he;
501 }
502 /* An object might have been moved to a different hash chain,
503 * while we walk along it - better check and retry.
504 */
505 } while (he != RHT_NULLS_MARKER(head));
506
507 /* Ensure we see any new tables. */
508 smp_rmb();
509
510 tbl = rht_dereference_rcu(tbl->future_tbl, ht);
511 if (unlikely(tbl))
512 goto restart;
513
514 return NULL;
515}
516
517/**
518 * rhashtable_lookup - search hash table
519 * @ht: hash table
520 * @key: the pointer to the key
521 * @params: hash table parameters
522 *
523 * Computes the hash value for the key and traverses the bucket chain looking
524 * for a entry with an identical key. The first matching entry is returned.
525 *
526 * This must only be called under the RCU read lock.
527 *
528 * Returns the first entry on which the compare function returned true.
529 */
530static inline void *rhashtable_lookup(
531 struct rhashtable *ht, const void *key,
532 const struct rhashtable_params params)
533{
534 struct rhash_head *he = __rhashtable_lookup(ht, key, params);
535
536 return he ? rht_obj(ht, he) : NULL;
537}
538
539/**
540 * rhashtable_lookup_fast - search hash table, without RCU read lock
541 * @ht: hash table
542 * @key: the pointer to the key
543 * @params: hash table parameters
544 *
545 * Computes the hash value for the key and traverses the bucket chain looking
546 * for a entry with an identical key. The first matching entry is returned.
547 *
548 * Only use this function when you have other mechanisms guaranteeing
549 * that the object won't go away after the RCU read lock is released.
550 *
551 * Returns the first entry on which the compare function returned true.
552 */
553static inline void *rhashtable_lookup_fast(
554 struct rhashtable *ht, const void *key,
555 const struct rhashtable_params params)
556{
557 void *obj;
558
559 rcu_read_lock();
560 obj = rhashtable_lookup(ht, key, params);
561 rcu_read_unlock();
562
563 return obj;
564}
565
566/**
567 * rhltable_lookup - search hash list table
568 * @hlt: hash table
569 * @key: the pointer to the key
570 * @params: hash table parameters
571 *
572 * Computes the hash value for the key and traverses the bucket chain looking
573 * for a entry with an identical key. All matching entries are returned
574 * in a list.
575 *
576 * This must only be called under the RCU read lock.
577 *
578 * Returns the list of entries that match the given key.
579 */
580static inline struct rhlist_head *rhltable_lookup(
581 struct rhltable *hlt, const void *key,
582 const struct rhashtable_params params)
583{
584 struct rhash_head *he = __rhashtable_lookup(&hlt->ht, key, params);
585
586 return he ? container_of(he, struct rhlist_head, rhead) : NULL;
587}
588
589/* Internal function, please use rhashtable_insert_fast() instead. This
590 * function returns the existing element already in hashes in there is a clash,
591 * otherwise it returns an error via ERR_PTR().
592 */
593static inline void *__rhashtable_insert_fast(
594 struct rhashtable *ht, const void *key, struct rhash_head *obj,
595 const struct rhashtable_params params, bool rhlist)
596{
597 struct rhashtable_compare_arg arg = {
598 .ht = ht,
599 .key = key,
600 };
601 struct rhash_head __rcu **pprev;
602 struct bucket_table *tbl;
603 struct rhash_head *head;
604 spinlock_t *lock;
605 unsigned int hash;
606 int elasticity;
607 void *data;
608
609 rcu_read_lock();
610
611 tbl = rht_dereference_rcu(ht->tbl, ht);
612 hash = rht_head_hashfn(ht, tbl, obj, params);
613 lock = rht_bucket_lock(tbl, hash);
614 spin_lock_bh(lock);
615
616 if (unlikely(rcu_access_pointer(tbl->future_tbl))) {
617slow_path:
618 spin_unlock_bh(lock);
619 rcu_read_unlock();
620 return rhashtable_insert_slow(ht, key, obj);
621 }
622
623 elasticity = RHT_ELASTICITY;
624 pprev = rht_bucket_insert(ht, tbl, hash);
625 data = ERR_PTR(-ENOMEM);
626 if (!pprev)
627 goto out;
628
629 rht_for_each_continue(head, *pprev, tbl, hash) {
630 struct rhlist_head *plist;
631 struct rhlist_head *list;
632
633 elasticity--;
634 if (!key ||
635 (params.obj_cmpfn ?
636 params.obj_cmpfn(&arg, rht_obj(ht, head)) :
637 rhashtable_compare(&arg, rht_obj(ht, head)))) {
638 pprev = &head->next;
639 continue;
640 }
641
642 data = rht_obj(ht, head);
643
644 if (!rhlist)
645 goto out;
646
647
648 list = container_of(obj, struct rhlist_head, rhead);
649 plist = container_of(head, struct rhlist_head, rhead);
650
651 RCU_INIT_POINTER(list->next, plist);
652 head = rht_dereference_bucket(head->next, tbl, hash);
653 RCU_INIT_POINTER(list->rhead.next, head);
654 rcu_assign_pointer(*pprev, obj);
655
656 goto good;
657 }
658
659 if (elasticity <= 0)
660 goto slow_path;
661
662 data = ERR_PTR(-E2BIG);
663 if (unlikely(rht_grow_above_max(ht, tbl)))
664 goto out;
665
666 if (unlikely(rht_grow_above_100(ht, tbl)))
667 goto slow_path;
668
669 head = rht_dereference_bucket(*pprev, tbl, hash);
670
671 RCU_INIT_POINTER(obj->next, head);
672 if (rhlist) {
673 struct rhlist_head *list;
674
675 list = container_of(obj, struct rhlist_head, rhead);
676 RCU_INIT_POINTER(list->next, NULL);
677 }
678
679 rcu_assign_pointer(*pprev, obj);
680
681 atomic_inc(&ht->nelems);
682 if (rht_grow_above_75(ht, tbl))
683 schedule_work(&ht->run_work);
684
685good:
686 data = NULL;
687
688out:
689 spin_unlock_bh(lock);
690 rcu_read_unlock();
691
692 return data;
693}
694
695/**
696 * rhashtable_insert_fast - insert object into hash table
697 * @ht: hash table
698 * @obj: pointer to hash head inside object
699 * @params: hash table parameters
700 *
701 * Will take a per bucket spinlock to protect against mutual mutations
702 * on the same bucket. Multiple insertions may occur in parallel unless
703 * they map to the same bucket lock.
704 *
705 * It is safe to call this function from atomic context.
706 *
707 * Will trigger an automatic deferred table resizing if residency in the
708 * table grows beyond 70%.
709 */
710static inline int rhashtable_insert_fast(
711 struct rhashtable *ht, struct rhash_head *obj,
712 const struct rhashtable_params params)
713{
714 void *ret;
715
716 ret = __rhashtable_insert_fast(ht, NULL, obj, params, false);
717 if (IS_ERR(ret))
718 return PTR_ERR(ret);
719
720 return ret == NULL ? 0 : -EEXIST;
721}
722
723/**
724 * rhltable_insert_key - insert object into hash list table
725 * @hlt: hash list table
726 * @key: the pointer to the key
727 * @list: pointer to hash list head inside object
728 * @params: hash table parameters
729 *
730 * Will take a per bucket spinlock to protect against mutual mutations
731 * on the same bucket. Multiple insertions may occur in parallel unless
732 * they map to the same bucket lock.
733 *
734 * It is safe to call this function from atomic context.
735 *
736 * Will trigger an automatic deferred table resizing if residency in the
737 * table grows beyond 70%.
738 */
739static inline int rhltable_insert_key(
740 struct rhltable *hlt, const void *key, struct rhlist_head *list,
741 const struct rhashtable_params params)
742{
743 return PTR_ERR(__rhashtable_insert_fast(&hlt->ht, key, &list->rhead,
744 params, true));
745}
746
747/**
748 * rhltable_insert - insert object into hash list table
749 * @hlt: hash list table
750 * @list: pointer to hash list head inside object
751 * @params: hash table parameters
752 *
753 * Will take a per bucket spinlock to protect against mutual mutations
754 * on the same bucket. Multiple insertions may occur in parallel unless
755 * they map to the same bucket lock.
756 *
757 * It is safe to call this function from atomic context.
758 *
759 * Will trigger an automatic deferred table resizing if residency in the
760 * table grows beyond 70%.
761 */
762static inline int rhltable_insert(
763 struct rhltable *hlt, struct rhlist_head *list,
764 const struct rhashtable_params params)
765{
766 const char *key = rht_obj(&hlt->ht, &list->rhead);
767
768 key += params.key_offset;
769
770 return rhltable_insert_key(hlt, key, list, params);
771}
772
773/**
774 * rhashtable_lookup_insert_fast - lookup and insert object into hash table
775 * @ht: hash table
776 * @obj: pointer to hash head inside object
777 * @params: hash table parameters
778 *
779 * Locks down the bucket chain in both the old and new table if a resize
780 * is in progress to ensure that writers can't remove from the old table
781 * and can't insert to the new table during the atomic operation of search
782 * and insertion. Searches for duplicates in both the old and new table if
783 * a resize is in progress.
784 *
785 * This lookup function may only be used for fixed key hash table (key_len
786 * parameter set). It will BUG() if used inappropriately.
787 *
788 * It is safe to call this function from atomic context.
789 *
790 * Will trigger an automatic deferred table resizing if residency in the
791 * table grows beyond 70%.
792 */
793static inline int rhashtable_lookup_insert_fast(
794 struct rhashtable *ht, struct rhash_head *obj,
795 const struct rhashtable_params params)
796{
797 const char *key = rht_obj(ht, obj);
798 void *ret;
799
800 BUG_ON(ht->p.obj_hashfn);
801
802 ret = __rhashtable_insert_fast(ht, key + ht->p.key_offset, obj, params,
803 false);
804 if (IS_ERR(ret))
805 return PTR_ERR(ret);
806
807 return ret == NULL ? 0 : -EEXIST;
808}
809
810/**
811 * rhashtable_lookup_get_insert_fast - lookup and insert object into hash table
812 * @ht: hash table
813 * @obj: pointer to hash head inside object
814 * @params: hash table parameters
815 *
816 * Just like rhashtable_lookup_insert_fast(), but this function returns the
817 * object if it exists, NULL if it did not and the insertion was successful,
818 * and an ERR_PTR otherwise.
819 */
820static inline void *rhashtable_lookup_get_insert_fast(
821 struct rhashtable *ht, struct rhash_head *obj,
822 const struct rhashtable_params params)
823{
824 const char *key = rht_obj(ht, obj);
825
826 BUG_ON(ht->p.obj_hashfn);
827
828 return __rhashtable_insert_fast(ht, key + ht->p.key_offset, obj, params,
829 false);
830}
831
832/**
833 * rhashtable_lookup_insert_key - search and insert object to hash table
834 * with explicit key
835 * @ht: hash table
836 * @key: key
837 * @obj: pointer to hash head inside object
838 * @params: hash table parameters
839 *
840 * Locks down the bucket chain in both the old and new table if a resize
841 * is in progress to ensure that writers can't remove from the old table
842 * and can't insert to the new table during the atomic operation of search
843 * and insertion. Searches for duplicates in both the old and new table if
844 * a resize is in progress.
845 *
846 * Lookups may occur in parallel with hashtable mutations and resizing.
847 *
848 * Will trigger an automatic deferred table resizing if residency in the
849 * table grows beyond 70%.
850 *
851 * Returns zero on success.
852 */
853static inline int rhashtable_lookup_insert_key(
854 struct rhashtable *ht, const void *key, struct rhash_head *obj,
855 const struct rhashtable_params params)
856{
857 void *ret;
858
859 BUG_ON(!ht->p.obj_hashfn || !key);
860
861 ret = __rhashtable_insert_fast(ht, key, obj, params, false);
862 if (IS_ERR(ret))
863 return PTR_ERR(ret);
864
865 return ret == NULL ? 0 : -EEXIST;
866}
867
868/**
869 * rhashtable_lookup_get_insert_key - lookup and insert object into hash table
870 * @ht: hash table
871 * @obj: pointer to hash head inside object
872 * @params: hash table parameters
873 * @data: pointer to element data already in hashes
874 *
875 * Just like rhashtable_lookup_insert_key(), but this function returns the
876 * object if it exists, NULL if it does not and the insertion was successful,
877 * and an ERR_PTR otherwise.
878 */
879static inline void *rhashtable_lookup_get_insert_key(
880 struct rhashtable *ht, const void *key, struct rhash_head *obj,
881 const struct rhashtable_params params)
882{
883 BUG_ON(!ht->p.obj_hashfn || !key);
884
885 return __rhashtable_insert_fast(ht, key, obj, params, false);
886}
887
888/* Internal function, please use rhashtable_remove_fast() instead */
889static inline int __rhashtable_remove_fast_one(
890 struct rhashtable *ht, struct bucket_table *tbl,
891 struct rhash_head *obj, const struct rhashtable_params params,
892 bool rhlist)
893{
894 struct rhash_head __rcu **pprev;
895 struct rhash_head *he;
896 spinlock_t * lock;
897 unsigned int hash;
898 int err = -ENOENT;
899
900 hash = rht_head_hashfn(ht, tbl, obj, params);
901 lock = rht_bucket_lock(tbl, hash);
902
903 spin_lock_bh(lock);
904
905 pprev = rht_bucket_var(tbl, hash);
906 rht_for_each_continue(he, *pprev, tbl, hash) {
907 struct rhlist_head *list;
908
909 list = container_of(he, struct rhlist_head, rhead);
910
911 if (he != obj) {
912 struct rhlist_head __rcu **lpprev;
913
914 pprev = &he->next;
915
916 if (!rhlist)
917 continue;
918
919 do {
920 lpprev = &list->next;
921 list = rht_dereference_bucket(list->next,
922 tbl, hash);
923 } while (list && obj != &list->rhead);
924
925 if (!list)
926 continue;
927
928 list = rht_dereference_bucket(list->next, tbl, hash);
929 RCU_INIT_POINTER(*lpprev, list);
930 err = 0;
931 break;
932 }
933
934 obj = rht_dereference_bucket(obj->next, tbl, hash);
935 err = 1;
936
937 if (rhlist) {
938 list = rht_dereference_bucket(list->next, tbl, hash);
939 if (list) {
940 RCU_INIT_POINTER(list->rhead.next, obj);
941 obj = &list->rhead;
942 err = 0;
943 }
944 }
945
946 rcu_assign_pointer(*pprev, obj);
947 break;
948 }
949
950 spin_unlock_bh(lock);
951
952 if (err > 0) {
953 atomic_dec(&ht->nelems);
954 if (unlikely(ht->p.automatic_shrinking &&
955 rht_shrink_below_30(ht, tbl)))
956 schedule_work(&ht->run_work);
957 err = 0;
958 }
959
960 return err;
961}
962
963/* Internal function, please use rhashtable_remove_fast() instead */
964static inline int __rhashtable_remove_fast(
965 struct rhashtable *ht, struct rhash_head *obj,
966 const struct rhashtable_params params, bool rhlist)
967{
968 struct bucket_table *tbl;
969 int err;
970
971 rcu_read_lock();
972
973 tbl = rht_dereference_rcu(ht->tbl, ht);
974
975 /* Because we have already taken (and released) the bucket
976 * lock in old_tbl, if we find that future_tbl is not yet
977 * visible then that guarantees the entry to still be in
978 * the old tbl if it exists.
979 */
980 while ((err = __rhashtable_remove_fast_one(ht, tbl, obj, params,
981 rhlist)) &&
982 (tbl = rht_dereference_rcu(tbl->future_tbl, ht)))
983 ;
984
985 rcu_read_unlock();
986
987 return err;
988}
989
990/**
991 * rhashtable_remove_fast - remove object from hash table
992 * @ht: hash table
993 * @obj: pointer to hash head inside object
994 * @params: hash table parameters
995 *
996 * Since the hash chain is single linked, the removal operation needs to
997 * walk the bucket chain upon removal. The removal operation is thus
998 * considerable slow if the hash table is not correctly sized.
999 *
1000 * Will automatically shrink the table if permitted when residency drops
1001 * below 30%.
1002 *
1003 * Returns zero on success, -ENOENT if the entry could not be found.
1004 */
1005static inline int rhashtable_remove_fast(
1006 struct rhashtable *ht, struct rhash_head *obj,
1007 const struct rhashtable_params params)
1008{
1009 return __rhashtable_remove_fast(ht, obj, params, false);
1010}
1011
1012/**
1013 * rhltable_remove - remove object from hash list table
1014 * @hlt: hash list table
1015 * @list: pointer to hash list head inside object
1016 * @params: hash table parameters
1017 *
1018 * Since the hash chain is single linked, the removal operation needs to
1019 * walk the bucket chain upon removal. The removal operation is thus
1020 * considerable slow if the hash table is not correctly sized.
1021 *
1022 * Will automatically shrink the table if permitted when residency drops
1023 * below 30%
1024 *
1025 * Returns zero on success, -ENOENT if the entry could not be found.
1026 */
1027static inline int rhltable_remove(
1028 struct rhltable *hlt, struct rhlist_head *list,
1029 const struct rhashtable_params params)
1030{
1031 return __rhashtable_remove_fast(&hlt->ht, &list->rhead, params, true);
1032}
1033
1034/* Internal function, please use rhashtable_replace_fast() instead */
1035static inline int __rhashtable_replace_fast(
1036 struct rhashtable *ht, struct bucket_table *tbl,
1037 struct rhash_head *obj_old, struct rhash_head *obj_new,
1038 const struct rhashtable_params params)
1039{
1040 struct rhash_head __rcu **pprev;
1041 struct rhash_head *he;
1042 spinlock_t *lock;
1043 unsigned int hash;
1044 int err = -ENOENT;
1045
1046 /* Minimally, the old and new objects must have same hash
1047 * (which should mean identifiers are the same).
1048 */
1049 hash = rht_head_hashfn(ht, tbl, obj_old, params);
1050 if (hash != rht_head_hashfn(ht, tbl, obj_new, params))
1051 return -EINVAL;
1052
1053 lock = rht_bucket_lock(tbl, hash);
1054
1055 spin_lock_bh(lock);
1056
1057 pprev = rht_bucket_var(tbl, hash);
1058 rht_for_each_continue(he, *pprev, tbl, hash) {
1059 if (he != obj_old) {
1060 pprev = &he->next;
1061 continue;
1062 }
1063
1064 rcu_assign_pointer(obj_new->next, obj_old->next);
1065 rcu_assign_pointer(*pprev, obj_new);
1066 err = 0;
1067 break;
1068 }
1069
1070 spin_unlock_bh(lock);
1071
1072 return err;
1073}
1074
1075/**
1076 * rhashtable_replace_fast - replace an object in hash table
1077 * @ht: hash table
1078 * @obj_old: pointer to hash head inside object being replaced
1079 * @obj_new: pointer to hash head inside object which is new
1080 * @params: hash table parameters
1081 *
1082 * Replacing an object doesn't affect the number of elements in the hash table
1083 * or bucket, so we don't need to worry about shrinking or expanding the
1084 * table here.
1085 *
1086 * Returns zero on success, -ENOENT if the entry could not be found,
1087 * -EINVAL if hash is not the same for the old and new objects.
1088 */
1089static inline int rhashtable_replace_fast(
1090 struct rhashtable *ht, struct rhash_head *obj_old,
1091 struct rhash_head *obj_new,
1092 const struct rhashtable_params params)
1093{
1094 struct bucket_table *tbl;
1095 int err;
1096
1097 rcu_read_lock();
1098
1099 tbl = rht_dereference_rcu(ht->tbl, ht);
1100
1101 /* Because we have already taken (and released) the bucket
1102 * lock in old_tbl, if we find that future_tbl is not yet
1103 * visible then that guarantees the entry to still be in
1104 * the old tbl if it exists.
1105 */
1106 while ((err = __rhashtable_replace_fast(ht, tbl, obj_old,
1107 obj_new, params)) &&
1108 (tbl = rht_dereference_rcu(tbl->future_tbl, ht)))
1109 ;
1110
1111 rcu_read_unlock();
1112
1113 return err;
1114}
1115
1116/**
1117 * rhltable_walk_enter - Initialise an iterator
1118 * @hlt: Table to walk over
1119 * @iter: Hash table Iterator
1120 *
1121 * This function prepares a hash table walk.
1122 *
1123 * Note that if you restart a walk after rhashtable_walk_stop you
1124 * may see the same object twice. Also, you may miss objects if
1125 * there are removals in between rhashtable_walk_stop and the next
1126 * call to rhashtable_walk_start.
1127 *
1128 * For a completely stable walk you should construct your own data
1129 * structure outside the hash table.
1130 *
1131 * This function may be called from any process context, including
1132 * non-preemptable context, but cannot be called from softirq or
1133 * hardirq context.
1134 *
1135 * You must call rhashtable_walk_exit after this function returns.
1136 */
1137static inline void rhltable_walk_enter(struct rhltable *hlt,
1138 struct rhashtable_iter *iter)
1139{
1140 return rhashtable_walk_enter(&hlt->ht, iter);
1141}
1142
1143/**
1144 * rhltable_free_and_destroy - free elements and destroy hash list table
1145 * @hlt: the hash list table to destroy
1146 * @free_fn: callback to release resources of element
1147 * @arg: pointer passed to free_fn
1148 *
1149 * See documentation for rhashtable_free_and_destroy.
1150 */
1151static inline void rhltable_free_and_destroy(struct rhltable *hlt,
1152 void (*free_fn)(void *ptr,
1153 void *arg),
1154 void *arg)
1155{
1156 return rhashtable_free_and_destroy(&hlt->ht, free_fn, arg);
1157}
1158
1159static inline void rhltable_destroy(struct rhltable *hlt)
1160{
1161 return rhltable_free_and_destroy(hlt, NULL, NULL);
1162}
1163
1164#endif /* _LINUX_RHASHTABLE_H */
1165