xref: /linux/kernel/bpf/queue_stack_maps.c (revision 37744feebc086908fd89760650f458ab19071750)
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * queue_stack_maps.c: BPF queue and stack maps
4  *
5  * Copyright (c) 2018 Politecnico di Torino
6  */
7 #include <linux/bpf.h>
8 #include <linux/list.h>
9 #include <linux/slab.h>
10 #include <linux/capability.h>
11 #include "percpu_freelist.h"
12 
13 #define QUEUE_STACK_CREATE_FLAG_MASK \
14 	(BPF_F_NUMA_NODE | BPF_F_ACCESS_MASK)
15 
16 struct bpf_queue_stack {
17 	struct bpf_map map;
18 	raw_spinlock_t lock;
19 	u32 head, tail;
20 	u32 size; /* max_entries + 1 */
21 
22 	char elements[0] __aligned(8);
23 };
24 
25 static struct bpf_queue_stack *bpf_queue_stack(struct bpf_map *map)
26 {
27 	return container_of(map, struct bpf_queue_stack, map);
28 }
29 
30 static bool queue_stack_map_is_empty(struct bpf_queue_stack *qs)
31 {
32 	return qs->head == qs->tail;
33 }
34 
35 static bool queue_stack_map_is_full(struct bpf_queue_stack *qs)
36 {
37 	u32 head = qs->head + 1;
38 
39 	if (unlikely(head >= qs->size))
40 		head = 0;
41 
42 	return head == qs->tail;
43 }
44 
45 /* Called from syscall */
46 static int queue_stack_map_alloc_check(union bpf_attr *attr)
47 {
48 	if (!capable(CAP_SYS_ADMIN))
49 		return -EPERM;
50 
51 	/* check sanity of attributes */
52 	if (attr->max_entries == 0 || attr->key_size != 0 ||
53 	    attr->value_size == 0 ||
54 	    attr->map_flags & ~QUEUE_STACK_CREATE_FLAG_MASK ||
55 	    !bpf_map_flags_access_ok(attr->map_flags))
56 		return -EINVAL;
57 
58 	if (attr->value_size > KMALLOC_MAX_SIZE)
59 		/* if value_size is bigger, the user space won't be able to
60 		 * access the elements.
61 		 */
62 		return -E2BIG;
63 
64 	return 0;
65 }
66 
67 static struct bpf_map *queue_stack_map_alloc(union bpf_attr *attr)
68 {
69 	int ret, numa_node = bpf_map_attr_numa_node(attr);
70 	struct bpf_map_memory mem = {0};
71 	struct bpf_queue_stack *qs;
72 	u64 size, queue_size, cost;
73 
74 	size = (u64) attr->max_entries + 1;
75 	cost = queue_size = sizeof(*qs) + size * attr->value_size;
76 
77 	ret = bpf_map_charge_init(&mem, cost);
78 	if (ret < 0)
79 		return ERR_PTR(ret);
80 
81 	qs = bpf_map_area_alloc(queue_size, numa_node);
82 	if (!qs) {
83 		bpf_map_charge_finish(&mem);
84 		return ERR_PTR(-ENOMEM);
85 	}
86 
87 	memset(qs, 0, sizeof(*qs));
88 
89 	bpf_map_init_from_attr(&qs->map, attr);
90 
91 	bpf_map_charge_move(&qs->map.memory, &mem);
92 	qs->size = size;
93 
94 	raw_spin_lock_init(&qs->lock);
95 
96 	return &qs->map;
97 }
98 
99 /* Called when map->refcnt goes to zero, either from workqueue or from syscall */
100 static void queue_stack_map_free(struct bpf_map *map)
101 {
102 	struct bpf_queue_stack *qs = bpf_queue_stack(map);
103 
104 	/* at this point bpf_prog->aux->refcnt == 0 and this map->refcnt == 0,
105 	 * so the programs (can be more than one that used this map) were
106 	 * disconnected from events. Wait for outstanding critical sections in
107 	 * these programs to complete
108 	 */
109 	synchronize_rcu();
110 
111 	bpf_map_area_free(qs);
112 }
113 
114 static int __queue_map_get(struct bpf_map *map, void *value, bool delete)
115 {
116 	struct bpf_queue_stack *qs = bpf_queue_stack(map);
117 	unsigned long flags;
118 	int err = 0;
119 	void *ptr;
120 
121 	raw_spin_lock_irqsave(&qs->lock, flags);
122 
123 	if (queue_stack_map_is_empty(qs)) {
124 		memset(value, 0, qs->map.value_size);
125 		err = -ENOENT;
126 		goto out;
127 	}
128 
129 	ptr = &qs->elements[qs->tail * qs->map.value_size];
130 	memcpy(value, ptr, qs->map.value_size);
131 
132 	if (delete) {
133 		if (unlikely(++qs->tail >= qs->size))
134 			qs->tail = 0;
135 	}
136 
137 out:
138 	raw_spin_unlock_irqrestore(&qs->lock, flags);
139 	return err;
140 }
141 
142 
143 static int __stack_map_get(struct bpf_map *map, void *value, bool delete)
144 {
145 	struct bpf_queue_stack *qs = bpf_queue_stack(map);
146 	unsigned long flags;
147 	int err = 0;
148 	void *ptr;
149 	u32 index;
150 
151 	raw_spin_lock_irqsave(&qs->lock, flags);
152 
153 	if (queue_stack_map_is_empty(qs)) {
154 		memset(value, 0, qs->map.value_size);
155 		err = -ENOENT;
156 		goto out;
157 	}
158 
159 	index = qs->head - 1;
160 	if (unlikely(index >= qs->size))
161 		index = qs->size - 1;
162 
163 	ptr = &qs->elements[index * qs->map.value_size];
164 	memcpy(value, ptr, qs->map.value_size);
165 
166 	if (delete)
167 		qs->head = index;
168 
169 out:
170 	raw_spin_unlock_irqrestore(&qs->lock, flags);
171 	return err;
172 }
173 
174 /* Called from syscall or from eBPF program */
175 static int queue_map_peek_elem(struct bpf_map *map, void *value)
176 {
177 	return __queue_map_get(map, value, false);
178 }
179 
180 /* Called from syscall or from eBPF program */
181 static int stack_map_peek_elem(struct bpf_map *map, void *value)
182 {
183 	return __stack_map_get(map, value, false);
184 }
185 
186 /* Called from syscall or from eBPF program */
187 static int queue_map_pop_elem(struct bpf_map *map, void *value)
188 {
189 	return __queue_map_get(map, value, true);
190 }
191 
192 /* Called from syscall or from eBPF program */
193 static int stack_map_pop_elem(struct bpf_map *map, void *value)
194 {
195 	return __stack_map_get(map, value, true);
196 }
197 
198 /* Called from syscall or from eBPF program */
199 static int queue_stack_map_push_elem(struct bpf_map *map, void *value,
200 				     u64 flags)
201 {
202 	struct bpf_queue_stack *qs = bpf_queue_stack(map);
203 	unsigned long irq_flags;
204 	int err = 0;
205 	void *dst;
206 
207 	/* BPF_EXIST is used to force making room for a new element in case the
208 	 * map is full
209 	 */
210 	bool replace = (flags & BPF_EXIST);
211 
212 	/* Check supported flags for queue and stack maps */
213 	if (flags & BPF_NOEXIST || flags > BPF_EXIST)
214 		return -EINVAL;
215 
216 	raw_spin_lock_irqsave(&qs->lock, irq_flags);
217 
218 	if (queue_stack_map_is_full(qs)) {
219 		if (!replace) {
220 			err = -E2BIG;
221 			goto out;
222 		}
223 		/* advance tail pointer to overwrite oldest element */
224 		if (unlikely(++qs->tail >= qs->size))
225 			qs->tail = 0;
226 	}
227 
228 	dst = &qs->elements[qs->head * qs->map.value_size];
229 	memcpy(dst, value, qs->map.value_size);
230 
231 	if (unlikely(++qs->head >= qs->size))
232 		qs->head = 0;
233 
234 out:
235 	raw_spin_unlock_irqrestore(&qs->lock, irq_flags);
236 	return err;
237 }
238 
239 /* Called from syscall or from eBPF program */
240 static void *queue_stack_map_lookup_elem(struct bpf_map *map, void *key)
241 {
242 	return NULL;
243 }
244 
245 /* Called from syscall or from eBPF program */
246 static int queue_stack_map_update_elem(struct bpf_map *map, void *key,
247 				       void *value, u64 flags)
248 {
249 	return -EINVAL;
250 }
251 
252 /* Called from syscall or from eBPF program */
253 static int queue_stack_map_delete_elem(struct bpf_map *map, void *key)
254 {
255 	return -EINVAL;
256 }
257 
258 /* Called from syscall */
259 static int queue_stack_map_get_next_key(struct bpf_map *map, void *key,
260 					void *next_key)
261 {
262 	return -EINVAL;
263 }
264 
265 const struct bpf_map_ops queue_map_ops = {
266 	.map_alloc_check = queue_stack_map_alloc_check,
267 	.map_alloc = queue_stack_map_alloc,
268 	.map_free = queue_stack_map_free,
269 	.map_lookup_elem = queue_stack_map_lookup_elem,
270 	.map_update_elem = queue_stack_map_update_elem,
271 	.map_delete_elem = queue_stack_map_delete_elem,
272 	.map_push_elem = queue_stack_map_push_elem,
273 	.map_pop_elem = queue_map_pop_elem,
274 	.map_peek_elem = queue_map_peek_elem,
275 	.map_get_next_key = queue_stack_map_get_next_key,
276 };
277 
278 const struct bpf_map_ops stack_map_ops = {
279 	.map_alloc_check = queue_stack_map_alloc_check,
280 	.map_alloc = queue_stack_map_alloc,
281 	.map_free = queue_stack_map_free,
282 	.map_lookup_elem = queue_stack_map_lookup_elem,
283 	.map_update_elem = queue_stack_map_update_elem,
284 	.map_delete_elem = queue_stack_map_delete_elem,
285 	.map_push_elem = queue_stack_map_push_elem,
286 	.map_pop_elem = stack_map_pop_elem,
287 	.map_peek_elem = stack_map_peek_elem,
288 	.map_get_next_key = queue_stack_map_get_next_key,
289 };
290