xref: /openbmc/linux/kernel/bpf/hashtab.c (revision 9cfc5c90)
1 /* Copyright (c) 2011-2014 PLUMgrid, http://plumgrid.com
2  *
3  * This program is free software; you can redistribute it and/or
4  * modify it under the terms of version 2 of the GNU General Public
5  * License as published by the Free Software Foundation.
6  *
7  * This program is distributed in the hope that it will be useful, but
8  * WITHOUT ANY WARRANTY; without even the implied warranty of
9  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
10  * General Public License for more details.
11  */
12 #include <linux/bpf.h>
13 #include <linux/jhash.h>
14 #include <linux/filter.h>
15 #include <linux/vmalloc.h>
16 
17 struct bpf_htab {
18 	struct bpf_map map;
19 	struct hlist_head *buckets;
20 	raw_spinlock_t lock;
21 	u32 count;	/* number of elements in this hashtable */
22 	u32 n_buckets;	/* number of hash buckets */
23 	u32 elem_size;	/* size of each element in bytes */
24 };
25 
26 /* each htab element is struct htab_elem + key + value */
27 struct htab_elem {
28 	struct hlist_node hash_node;
29 	struct rcu_head rcu;
30 	u32 hash;
31 	char key[0] __aligned(8);
32 };
33 
34 /* Called from syscall */
35 static struct bpf_map *htab_map_alloc(union bpf_attr *attr)
36 {
37 	struct bpf_htab *htab;
38 	int err, i;
39 
40 	htab = kzalloc(sizeof(*htab), GFP_USER);
41 	if (!htab)
42 		return ERR_PTR(-ENOMEM);
43 
44 	/* mandatory map attributes */
45 	htab->map.key_size = attr->key_size;
46 	htab->map.value_size = attr->value_size;
47 	htab->map.max_entries = attr->max_entries;
48 
49 	/* check sanity of attributes.
50 	 * value_size == 0 may be allowed in the future to use map as a set
51 	 */
52 	err = -EINVAL;
53 	if (htab->map.max_entries == 0 || htab->map.key_size == 0 ||
54 	    htab->map.value_size == 0)
55 		goto free_htab;
56 
57 	/* hash table size must be power of 2 */
58 	htab->n_buckets = roundup_pow_of_two(htab->map.max_entries);
59 
60 	err = -E2BIG;
61 	if (htab->map.key_size > MAX_BPF_STACK)
62 		/* eBPF programs initialize keys on stack, so they cannot be
63 		 * larger than max stack size
64 		 */
65 		goto free_htab;
66 
67 	err = -ENOMEM;
68 	/* prevent zero size kmalloc and check for u32 overflow */
69 	if (htab->n_buckets == 0 ||
70 	    htab->n_buckets > U32_MAX / sizeof(struct hlist_head))
71 		goto free_htab;
72 
73 	htab->buckets = kmalloc_array(htab->n_buckets, sizeof(struct hlist_head),
74 				      GFP_USER | __GFP_NOWARN);
75 
76 	if (!htab->buckets) {
77 		htab->buckets = vmalloc(htab->n_buckets * sizeof(struct hlist_head));
78 		if (!htab->buckets)
79 			goto free_htab;
80 	}
81 
82 	for (i = 0; i < htab->n_buckets; i++)
83 		INIT_HLIST_HEAD(&htab->buckets[i]);
84 
85 	raw_spin_lock_init(&htab->lock);
86 	htab->count = 0;
87 
88 	htab->elem_size = sizeof(struct htab_elem) +
89 			  round_up(htab->map.key_size, 8) +
90 			  htab->map.value_size;
91 
92 	htab->map.pages = round_up(htab->n_buckets * sizeof(struct hlist_head) +
93 				   htab->elem_size * htab->map.max_entries,
94 				   PAGE_SIZE) >> PAGE_SHIFT;
95 	return &htab->map;
96 
97 free_htab:
98 	kfree(htab);
99 	return ERR_PTR(err);
100 }
101 
102 static inline u32 htab_map_hash(const void *key, u32 key_len)
103 {
104 	return jhash(key, key_len, 0);
105 }
106 
107 static inline struct hlist_head *select_bucket(struct bpf_htab *htab, u32 hash)
108 {
109 	return &htab->buckets[hash & (htab->n_buckets - 1)];
110 }
111 
112 static struct htab_elem *lookup_elem_raw(struct hlist_head *head, u32 hash,
113 					 void *key, u32 key_size)
114 {
115 	struct htab_elem *l;
116 
117 	hlist_for_each_entry_rcu(l, head, hash_node)
118 		if (l->hash == hash && !memcmp(&l->key, key, key_size))
119 			return l;
120 
121 	return NULL;
122 }
123 
124 /* Called from syscall or from eBPF program */
125 static void *htab_map_lookup_elem(struct bpf_map *map, void *key)
126 {
127 	struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
128 	struct hlist_head *head;
129 	struct htab_elem *l;
130 	u32 hash, key_size;
131 
132 	/* Must be called with rcu_read_lock. */
133 	WARN_ON_ONCE(!rcu_read_lock_held());
134 
135 	key_size = map->key_size;
136 
137 	hash = htab_map_hash(key, key_size);
138 
139 	head = select_bucket(htab, hash);
140 
141 	l = lookup_elem_raw(head, hash, key, key_size);
142 
143 	if (l)
144 		return l->key + round_up(map->key_size, 8);
145 
146 	return NULL;
147 }
148 
149 /* Called from syscall */
150 static int htab_map_get_next_key(struct bpf_map *map, void *key, void *next_key)
151 {
152 	struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
153 	struct hlist_head *head;
154 	struct htab_elem *l, *next_l;
155 	u32 hash, key_size;
156 	int i;
157 
158 	WARN_ON_ONCE(!rcu_read_lock_held());
159 
160 	key_size = map->key_size;
161 
162 	hash = htab_map_hash(key, key_size);
163 
164 	head = select_bucket(htab, hash);
165 
166 	/* lookup the key */
167 	l = lookup_elem_raw(head, hash, key, key_size);
168 
169 	if (!l) {
170 		i = 0;
171 		goto find_first_elem;
172 	}
173 
174 	/* key was found, get next key in the same bucket */
175 	next_l = hlist_entry_safe(rcu_dereference_raw(hlist_next_rcu(&l->hash_node)),
176 				  struct htab_elem, hash_node);
177 
178 	if (next_l) {
179 		/* if next elem in this hash list is non-zero, just return it */
180 		memcpy(next_key, next_l->key, key_size);
181 		return 0;
182 	}
183 
184 	/* no more elements in this hash list, go to the next bucket */
185 	i = hash & (htab->n_buckets - 1);
186 	i++;
187 
188 find_first_elem:
189 	/* iterate over buckets */
190 	for (; i < htab->n_buckets; i++) {
191 		head = select_bucket(htab, i);
192 
193 		/* pick first element in the bucket */
194 		next_l = hlist_entry_safe(rcu_dereference_raw(hlist_first_rcu(head)),
195 					  struct htab_elem, hash_node);
196 		if (next_l) {
197 			/* if it's not empty, just return it */
198 			memcpy(next_key, next_l->key, key_size);
199 			return 0;
200 		}
201 	}
202 
203 	/* itereated over all buckets and all elements */
204 	return -ENOENT;
205 }
206 
207 /* Called from syscall or from eBPF program */
208 static int htab_map_update_elem(struct bpf_map *map, void *key, void *value,
209 				u64 map_flags)
210 {
211 	struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
212 	struct htab_elem *l_new, *l_old;
213 	struct hlist_head *head;
214 	unsigned long flags;
215 	u32 key_size;
216 	int ret;
217 
218 	if (map_flags > BPF_EXIST)
219 		/* unknown flags */
220 		return -EINVAL;
221 
222 	WARN_ON_ONCE(!rcu_read_lock_held());
223 
224 	/* allocate new element outside of lock */
225 	l_new = kmalloc(htab->elem_size, GFP_ATOMIC);
226 	if (!l_new)
227 		return -ENOMEM;
228 
229 	key_size = map->key_size;
230 
231 	memcpy(l_new->key, key, key_size);
232 	memcpy(l_new->key + round_up(key_size, 8), value, map->value_size);
233 
234 	l_new->hash = htab_map_hash(l_new->key, key_size);
235 
236 	/* bpf_map_update_elem() can be called in_irq() */
237 	raw_spin_lock_irqsave(&htab->lock, flags);
238 
239 	head = select_bucket(htab, l_new->hash);
240 
241 	l_old = lookup_elem_raw(head, l_new->hash, key, key_size);
242 
243 	if (!l_old && unlikely(htab->count >= map->max_entries)) {
244 		/* if elem with this 'key' doesn't exist and we've reached
245 		 * max_entries limit, fail insertion of new elem
246 		 */
247 		ret = -E2BIG;
248 		goto err;
249 	}
250 
251 	if (l_old && map_flags == BPF_NOEXIST) {
252 		/* elem already exists */
253 		ret = -EEXIST;
254 		goto err;
255 	}
256 
257 	if (!l_old && map_flags == BPF_EXIST) {
258 		/* elem doesn't exist, cannot update it */
259 		ret = -ENOENT;
260 		goto err;
261 	}
262 
263 	/* add new element to the head of the list, so that concurrent
264 	 * search will find it before old elem
265 	 */
266 	hlist_add_head_rcu(&l_new->hash_node, head);
267 	if (l_old) {
268 		hlist_del_rcu(&l_old->hash_node);
269 		kfree_rcu(l_old, rcu);
270 	} else {
271 		htab->count++;
272 	}
273 	raw_spin_unlock_irqrestore(&htab->lock, flags);
274 
275 	return 0;
276 err:
277 	raw_spin_unlock_irqrestore(&htab->lock, flags);
278 	kfree(l_new);
279 	return ret;
280 }
281 
282 /* Called from syscall or from eBPF program */
283 static int htab_map_delete_elem(struct bpf_map *map, void *key)
284 {
285 	struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
286 	struct hlist_head *head;
287 	struct htab_elem *l;
288 	unsigned long flags;
289 	u32 hash, key_size;
290 	int ret = -ENOENT;
291 
292 	WARN_ON_ONCE(!rcu_read_lock_held());
293 
294 	key_size = map->key_size;
295 
296 	hash = htab_map_hash(key, key_size);
297 
298 	raw_spin_lock_irqsave(&htab->lock, flags);
299 
300 	head = select_bucket(htab, hash);
301 
302 	l = lookup_elem_raw(head, hash, key, key_size);
303 
304 	if (l) {
305 		hlist_del_rcu(&l->hash_node);
306 		htab->count--;
307 		kfree_rcu(l, rcu);
308 		ret = 0;
309 	}
310 
311 	raw_spin_unlock_irqrestore(&htab->lock, flags);
312 	return ret;
313 }
314 
315 static void delete_all_elements(struct bpf_htab *htab)
316 {
317 	int i;
318 
319 	for (i = 0; i < htab->n_buckets; i++) {
320 		struct hlist_head *head = select_bucket(htab, i);
321 		struct hlist_node *n;
322 		struct htab_elem *l;
323 
324 		hlist_for_each_entry_safe(l, n, head, hash_node) {
325 			hlist_del_rcu(&l->hash_node);
326 			htab->count--;
327 			kfree(l);
328 		}
329 	}
330 }
331 
332 /* Called when map->refcnt goes to zero, either from workqueue or from syscall */
333 static void htab_map_free(struct bpf_map *map)
334 {
335 	struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
336 
337 	/* at this point bpf_prog->aux->refcnt == 0 and this map->refcnt == 0,
338 	 * so the programs (can be more than one that used this map) were
339 	 * disconnected from events. Wait for outstanding critical sections in
340 	 * these programs to complete
341 	 */
342 	synchronize_rcu();
343 
344 	/* some of kfree_rcu() callbacks for elements of this map may not have
345 	 * executed. It's ok. Proceed to free residual elements and map itself
346 	 */
347 	delete_all_elements(htab);
348 	kvfree(htab->buckets);
349 	kfree(htab);
350 }
351 
352 static const struct bpf_map_ops htab_ops = {
353 	.map_alloc = htab_map_alloc,
354 	.map_free = htab_map_free,
355 	.map_get_next_key = htab_map_get_next_key,
356 	.map_lookup_elem = htab_map_lookup_elem,
357 	.map_update_elem = htab_map_update_elem,
358 	.map_delete_elem = htab_map_delete_elem,
359 };
360 
361 static struct bpf_map_type_list htab_type __read_mostly = {
362 	.ops = &htab_ops,
363 	.type = BPF_MAP_TYPE_HASH,
364 };
365 
366 static int __init register_htab_map(void)
367 {
368 	bpf_register_map_type(&htab_type);
369 	return 0;
370 }
371 late_initcall(register_htab_map);
372