xref: /openbmc/linux/kernel/bpf/queue_stack_maps.c (revision fbb6b31a)
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[] __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 (!bpf_capable())
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 numa_node = bpf_map_attr_numa_node(attr);
70 	struct bpf_queue_stack *qs;
71 	u64 size, queue_size;
72 
73 	size = (u64) attr->max_entries + 1;
74 	queue_size = sizeof(*qs) + size * attr->value_size;
75 
76 	qs = bpf_map_area_alloc(queue_size, numa_node);
77 	if (!qs)
78 		return ERR_PTR(-ENOMEM);
79 
80 	memset(qs, 0, sizeof(*qs));
81 
82 	bpf_map_init_from_attr(&qs->map, attr);
83 
84 	qs->size = size;
85 
86 	raw_spin_lock_init(&qs->lock);
87 
88 	return &qs->map;
89 }
90 
91 /* Called when map->refcnt goes to zero, either from workqueue or from syscall */
92 static void queue_stack_map_free(struct bpf_map *map)
93 {
94 	struct bpf_queue_stack *qs = bpf_queue_stack(map);
95 
96 	bpf_map_area_free(qs);
97 }
98 
99 static int __queue_map_get(struct bpf_map *map, void *value, bool delete)
100 {
101 	struct bpf_queue_stack *qs = bpf_queue_stack(map);
102 	unsigned long flags;
103 	int err = 0;
104 	void *ptr;
105 
106 	raw_spin_lock_irqsave(&qs->lock, flags);
107 
108 	if (queue_stack_map_is_empty(qs)) {
109 		memset(value, 0, qs->map.value_size);
110 		err = -ENOENT;
111 		goto out;
112 	}
113 
114 	ptr = &qs->elements[qs->tail * qs->map.value_size];
115 	memcpy(value, ptr, qs->map.value_size);
116 
117 	if (delete) {
118 		if (unlikely(++qs->tail >= qs->size))
119 			qs->tail = 0;
120 	}
121 
122 out:
123 	raw_spin_unlock_irqrestore(&qs->lock, flags);
124 	return err;
125 }
126 
127 
128 static int __stack_map_get(struct bpf_map *map, void *value, bool delete)
129 {
130 	struct bpf_queue_stack *qs = bpf_queue_stack(map);
131 	unsigned long flags;
132 	int err = 0;
133 	void *ptr;
134 	u32 index;
135 
136 	raw_spin_lock_irqsave(&qs->lock, flags);
137 
138 	if (queue_stack_map_is_empty(qs)) {
139 		memset(value, 0, qs->map.value_size);
140 		err = -ENOENT;
141 		goto out;
142 	}
143 
144 	index = qs->head - 1;
145 	if (unlikely(index >= qs->size))
146 		index = qs->size - 1;
147 
148 	ptr = &qs->elements[index * qs->map.value_size];
149 	memcpy(value, ptr, qs->map.value_size);
150 
151 	if (delete)
152 		qs->head = index;
153 
154 out:
155 	raw_spin_unlock_irqrestore(&qs->lock, flags);
156 	return err;
157 }
158 
159 /* Called from syscall or from eBPF program */
160 static int queue_map_peek_elem(struct bpf_map *map, void *value)
161 {
162 	return __queue_map_get(map, value, false);
163 }
164 
165 /* Called from syscall or from eBPF program */
166 static int stack_map_peek_elem(struct bpf_map *map, void *value)
167 {
168 	return __stack_map_get(map, value, false);
169 }
170 
171 /* Called from syscall or from eBPF program */
172 static int queue_map_pop_elem(struct bpf_map *map, void *value)
173 {
174 	return __queue_map_get(map, value, true);
175 }
176 
177 /* Called from syscall or from eBPF program */
178 static int stack_map_pop_elem(struct bpf_map *map, void *value)
179 {
180 	return __stack_map_get(map, value, true);
181 }
182 
183 /* Called from syscall or from eBPF program */
184 static int queue_stack_map_push_elem(struct bpf_map *map, void *value,
185 				     u64 flags)
186 {
187 	struct bpf_queue_stack *qs = bpf_queue_stack(map);
188 	unsigned long irq_flags;
189 	int err = 0;
190 	void *dst;
191 
192 	/* BPF_EXIST is used to force making room for a new element in case the
193 	 * map is full
194 	 */
195 	bool replace = (flags & BPF_EXIST);
196 
197 	/* Check supported flags for queue and stack maps */
198 	if (flags & BPF_NOEXIST || flags > BPF_EXIST)
199 		return -EINVAL;
200 
201 	raw_spin_lock_irqsave(&qs->lock, irq_flags);
202 
203 	if (queue_stack_map_is_full(qs)) {
204 		if (!replace) {
205 			err = -E2BIG;
206 			goto out;
207 		}
208 		/* advance tail pointer to overwrite oldest element */
209 		if (unlikely(++qs->tail >= qs->size))
210 			qs->tail = 0;
211 	}
212 
213 	dst = &qs->elements[qs->head * qs->map.value_size];
214 	memcpy(dst, value, qs->map.value_size);
215 
216 	if (unlikely(++qs->head >= qs->size))
217 		qs->head = 0;
218 
219 out:
220 	raw_spin_unlock_irqrestore(&qs->lock, irq_flags);
221 	return err;
222 }
223 
224 /* Called from syscall or from eBPF program */
225 static void *queue_stack_map_lookup_elem(struct bpf_map *map, void *key)
226 {
227 	return NULL;
228 }
229 
230 /* Called from syscall or from eBPF program */
231 static int queue_stack_map_update_elem(struct bpf_map *map, void *key,
232 				       void *value, u64 flags)
233 {
234 	return -EINVAL;
235 }
236 
237 /* Called from syscall or from eBPF program */
238 static int queue_stack_map_delete_elem(struct bpf_map *map, void *key)
239 {
240 	return -EINVAL;
241 }
242 
243 /* Called from syscall */
244 static int queue_stack_map_get_next_key(struct bpf_map *map, void *key,
245 					void *next_key)
246 {
247 	return -EINVAL;
248 }
249 
250 static int queue_map_btf_id;
251 const struct bpf_map_ops queue_map_ops = {
252 	.map_meta_equal = bpf_map_meta_equal,
253 	.map_alloc_check = queue_stack_map_alloc_check,
254 	.map_alloc = queue_stack_map_alloc,
255 	.map_free = queue_stack_map_free,
256 	.map_lookup_elem = queue_stack_map_lookup_elem,
257 	.map_update_elem = queue_stack_map_update_elem,
258 	.map_delete_elem = queue_stack_map_delete_elem,
259 	.map_push_elem = queue_stack_map_push_elem,
260 	.map_pop_elem = queue_map_pop_elem,
261 	.map_peek_elem = queue_map_peek_elem,
262 	.map_get_next_key = queue_stack_map_get_next_key,
263 	.map_btf_name = "bpf_queue_stack",
264 	.map_btf_id = &queue_map_btf_id,
265 };
266 
267 static int stack_map_btf_id;
268 const struct bpf_map_ops stack_map_ops = {
269 	.map_meta_equal = bpf_map_meta_equal,
270 	.map_alloc_check = queue_stack_map_alloc_check,
271 	.map_alloc = queue_stack_map_alloc,
272 	.map_free = queue_stack_map_free,
273 	.map_lookup_elem = queue_stack_map_lookup_elem,
274 	.map_update_elem = queue_stack_map_update_elem,
275 	.map_delete_elem = queue_stack_map_delete_elem,
276 	.map_push_elem = queue_stack_map_push_elem,
277 	.map_pop_elem = stack_map_pop_elem,
278 	.map_peek_elem = stack_map_peek_elem,
279 	.map_get_next_key = queue_stack_map_get_next_key,
280 	.map_btf_name = "bpf_queue_stack",
281 	.map_btf_id = &stack_map_btf_id,
282 };
283