Commit | Line | Data |
---|---|---|
0f8e4bd8 AS |
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 | 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 | htab->buckets = kmalloc_array(htab->n_buckets, sizeof(struct hlist_head), | |
69 | GFP_USER | __GFP_NOWARN); | |
70 | ||
71 | if (!htab->buckets) { | |
72 | htab->buckets = vmalloc(htab->n_buckets * sizeof(struct hlist_head)); | |
73 | if (!htab->buckets) | |
74 | goto free_htab; | |
75 | } | |
76 | ||
77 | for (i = 0; i < htab->n_buckets; i++) | |
78 | INIT_HLIST_HEAD(&htab->buckets[i]); | |
79 | ||
80 | spin_lock_init(&htab->lock); | |
81 | htab->count = 0; | |
82 | ||
83 | htab->elem_size = sizeof(struct htab_elem) + | |
84 | round_up(htab->map.key_size, 8) + | |
85 | htab->map.value_size; | |
86 | return &htab->map; | |
87 | ||
88 | free_htab: | |
89 | kfree(htab); | |
90 | return ERR_PTR(err); | |
91 | } | |
92 | ||
93 | static inline u32 htab_map_hash(const void *key, u32 key_len) | |
94 | { | |
95 | return jhash(key, key_len, 0); | |
96 | } | |
97 | ||
98 | static inline struct hlist_head *select_bucket(struct bpf_htab *htab, u32 hash) | |
99 | { | |
100 | return &htab->buckets[hash & (htab->n_buckets - 1)]; | |
101 | } | |
102 | ||
103 | static struct htab_elem *lookup_elem_raw(struct hlist_head *head, u32 hash, | |
104 | void *key, u32 key_size) | |
105 | { | |
106 | struct htab_elem *l; | |
107 | ||
108 | hlist_for_each_entry_rcu(l, head, hash_node) | |
109 | if (l->hash == hash && !memcmp(&l->key, key, key_size)) | |
110 | return l; | |
111 | ||
112 | return NULL; | |
113 | } | |
114 | ||
115 | /* Called from syscall or from eBPF program */ | |
116 | static void *htab_map_lookup_elem(struct bpf_map *map, void *key) | |
117 | { | |
118 | struct bpf_htab *htab = container_of(map, struct bpf_htab, map); | |
119 | struct hlist_head *head; | |
120 | struct htab_elem *l; | |
121 | u32 hash, key_size; | |
122 | ||
123 | /* Must be called with rcu_read_lock. */ | |
124 | WARN_ON_ONCE(!rcu_read_lock_held()); | |
125 | ||
126 | key_size = map->key_size; | |
127 | ||
128 | hash = htab_map_hash(key, key_size); | |
129 | ||
130 | head = select_bucket(htab, hash); | |
131 | ||
132 | l = lookup_elem_raw(head, hash, key, key_size); | |
133 | ||
134 | if (l) | |
135 | return l->key + round_up(map->key_size, 8); | |
136 | ||
137 | return NULL; | |
138 | } | |
139 | ||
140 | /* Called from syscall */ | |
141 | static int htab_map_get_next_key(struct bpf_map *map, void *key, void *next_key) | |
142 | { | |
143 | struct bpf_htab *htab = container_of(map, struct bpf_htab, map); | |
144 | struct hlist_head *head; | |
145 | struct htab_elem *l, *next_l; | |
146 | u32 hash, key_size; | |
147 | int i; | |
148 | ||
149 | WARN_ON_ONCE(!rcu_read_lock_held()); | |
150 | ||
151 | key_size = map->key_size; | |
152 | ||
153 | hash = htab_map_hash(key, key_size); | |
154 | ||
155 | head = select_bucket(htab, hash); | |
156 | ||
157 | /* lookup the key */ | |
158 | l = lookup_elem_raw(head, hash, key, key_size); | |
159 | ||
160 | if (!l) { | |
161 | i = 0; | |
162 | goto find_first_elem; | |
163 | } | |
164 | ||
165 | /* key was found, get next key in the same bucket */ | |
166 | next_l = hlist_entry_safe(rcu_dereference_raw(hlist_next_rcu(&l->hash_node)), | |
167 | struct htab_elem, hash_node); | |
168 | ||
169 | if (next_l) { | |
170 | /* if next elem in this hash list is non-zero, just return it */ | |
171 | memcpy(next_key, next_l->key, key_size); | |
172 | return 0; | |
173 | } | |
174 | ||
175 | /* no more elements in this hash list, go to the next bucket */ | |
176 | i = hash & (htab->n_buckets - 1); | |
177 | i++; | |
178 | ||
179 | find_first_elem: | |
180 | /* iterate over buckets */ | |
181 | for (; i < htab->n_buckets; i++) { | |
182 | head = select_bucket(htab, i); | |
183 | ||
184 | /* pick first element in the bucket */ | |
185 | next_l = hlist_entry_safe(rcu_dereference_raw(hlist_first_rcu(head)), | |
186 | struct htab_elem, hash_node); | |
187 | if (next_l) { | |
188 | /* if it's not empty, just return it */ | |
189 | memcpy(next_key, next_l->key, key_size); | |
190 | return 0; | |
191 | } | |
192 | } | |
193 | ||
194 | /* itereated over all buckets and all elements */ | |
195 | return -ENOENT; | |
196 | } | |
197 | ||
198 | /* Called from syscall or from eBPF program */ | |
199 | static int htab_map_update_elem(struct bpf_map *map, void *key, void *value, | |
200 | u64 map_flags) | |
201 | { | |
202 | struct bpf_htab *htab = container_of(map, struct bpf_htab, map); | |
203 | struct htab_elem *l_new, *l_old; | |
204 | struct hlist_head *head; | |
205 | unsigned long flags; | |
206 | u32 key_size; | |
207 | int ret; | |
208 | ||
209 | if (map_flags > BPF_EXIST) | |
210 | /* unknown flags */ | |
211 | return -EINVAL; | |
212 | ||
213 | WARN_ON_ONCE(!rcu_read_lock_held()); | |
214 | ||
215 | /* allocate new element outside of lock */ | |
216 | l_new = kmalloc(htab->elem_size, GFP_ATOMIC); | |
217 | if (!l_new) | |
218 | return -ENOMEM; | |
219 | ||
220 | key_size = map->key_size; | |
221 | ||
222 | memcpy(l_new->key, key, key_size); | |
223 | memcpy(l_new->key + round_up(key_size, 8), value, map->value_size); | |
224 | ||
225 | l_new->hash = htab_map_hash(l_new->key, key_size); | |
226 | ||
227 | /* bpf_map_update_elem() can be called in_irq() */ | |
228 | spin_lock_irqsave(&htab->lock, flags); | |
229 | ||
230 | head = select_bucket(htab, l_new->hash); | |
231 | ||
232 | l_old = lookup_elem_raw(head, l_new->hash, key, key_size); | |
233 | ||
234 | if (!l_old && unlikely(htab->count >= map->max_entries)) { | |
235 | /* if elem with this 'key' doesn't exist and we've reached | |
236 | * max_entries limit, fail insertion of new elem | |
237 | */ | |
238 | ret = -E2BIG; | |
239 | goto err; | |
240 | } | |
241 | ||
242 | if (l_old && map_flags == BPF_NOEXIST) { | |
243 | /* elem already exists */ | |
244 | ret = -EEXIST; | |
245 | goto err; | |
246 | } | |
247 | ||
248 | if (!l_old && map_flags == BPF_EXIST) { | |
249 | /* elem doesn't exist, cannot update it */ | |
250 | ret = -ENOENT; | |
251 | goto err; | |
252 | } | |
253 | ||
254 | /* add new element to the head of the list, so that concurrent | |
255 | * search will find it before old elem | |
256 | */ | |
257 | hlist_add_head_rcu(&l_new->hash_node, head); | |
258 | if (l_old) { | |
259 | hlist_del_rcu(&l_old->hash_node); | |
260 | kfree_rcu(l_old, rcu); | |
261 | } else { | |
262 | htab->count++; | |
263 | } | |
264 | spin_unlock_irqrestore(&htab->lock, flags); | |
265 | ||
266 | return 0; | |
267 | err: | |
268 | spin_unlock_irqrestore(&htab->lock, flags); | |
269 | kfree(l_new); | |
270 | return ret; | |
271 | } | |
272 | ||
273 | /* Called from syscall or from eBPF program */ | |
274 | static int htab_map_delete_elem(struct bpf_map *map, void *key) | |
275 | { | |
276 | struct bpf_htab *htab = container_of(map, struct bpf_htab, map); | |
277 | struct hlist_head *head; | |
278 | struct htab_elem *l; | |
279 | unsigned long flags; | |
280 | u32 hash, key_size; | |
281 | int ret = -ENOENT; | |
282 | ||
283 | WARN_ON_ONCE(!rcu_read_lock_held()); | |
284 | ||
285 | key_size = map->key_size; | |
286 | ||
287 | hash = htab_map_hash(key, key_size); | |
288 | ||
289 | spin_lock_irqsave(&htab->lock, flags); | |
290 | ||
291 | head = select_bucket(htab, hash); | |
292 | ||
293 | l = lookup_elem_raw(head, hash, key, key_size); | |
294 | ||
295 | if (l) { | |
296 | hlist_del_rcu(&l->hash_node); | |
297 | htab->count--; | |
298 | kfree_rcu(l, rcu); | |
299 | ret = 0; | |
300 | } | |
301 | ||
302 | spin_unlock_irqrestore(&htab->lock, flags); | |
303 | return ret; | |
304 | } | |
305 | ||
306 | static void delete_all_elements(struct bpf_htab *htab) | |
307 | { | |
308 | int i; | |
309 | ||
310 | for (i = 0; i < htab->n_buckets; i++) { | |
311 | struct hlist_head *head = select_bucket(htab, i); | |
312 | struct hlist_node *n; | |
313 | struct htab_elem *l; | |
314 | ||
315 | hlist_for_each_entry_safe(l, n, head, hash_node) { | |
316 | hlist_del_rcu(&l->hash_node); | |
317 | htab->count--; | |
318 | kfree(l); | |
319 | } | |
320 | } | |
321 | } | |
322 | ||
323 | /* Called when map->refcnt goes to zero, either from workqueue or from syscall */ | |
324 | static void htab_map_free(struct bpf_map *map) | |
325 | { | |
326 | struct bpf_htab *htab = container_of(map, struct bpf_htab, map); | |
327 | ||
328 | /* at this point bpf_prog->aux->refcnt == 0 and this map->refcnt == 0, | |
329 | * so the programs (can be more than one that used this map) were | |
330 | * disconnected from events. Wait for outstanding critical sections in | |
331 | * these programs to complete | |
332 | */ | |
333 | synchronize_rcu(); | |
334 | ||
335 | /* some of kfree_rcu() callbacks for elements of this map may not have | |
336 | * executed. It's ok. Proceed to free residual elements and map itself | |
337 | */ | |
338 | delete_all_elements(htab); | |
339 | kvfree(htab->buckets); | |
340 | kfree(htab); | |
341 | } | |
342 | ||
343 | static struct bpf_map_ops htab_ops = { | |
344 | .map_alloc = htab_map_alloc, | |
345 | .map_free = htab_map_free, | |
346 | .map_get_next_key = htab_map_get_next_key, | |
347 | .map_lookup_elem = htab_map_lookup_elem, | |
348 | .map_update_elem = htab_map_update_elem, | |
349 | .map_delete_elem = htab_map_delete_elem, | |
350 | }; | |
351 | ||
352 | static struct bpf_map_type_list tl = { | |
353 | .ops = &htab_ops, | |
354 | .type = BPF_MAP_TYPE_HASH, | |
355 | }; | |
356 | ||
357 | static int __init register_htab_map(void) | |
358 | { | |
359 | bpf_register_map_type(&tl); | |
360 | return 0; | |
361 | } | |
362 | late_initcall(register_htab_map); |