[people/ms/linux.git] / kernel / bpf / arraymap.c

/* Copyright (c) 2011-2014 PLUMgrid, http://plumgrid.com
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of version 2 of the GNU General Public
 * License as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
 * General Public License for more details.
 */
#include <linux/bpf.h>
#include <linux/err.h>
#include <linux/vmalloc.h>
#include <linux/slab.h>
#include <linux/mm.h>
#include <linux/filter.h>
#include <linux/perf_event.h>

/* Called from syscall */
static struct bpf_map *array_map_alloc(union bpf_attr *attr)
{
	struct bpf_array *array;
	u32 elem_size, array_size;

	/* check sanity of attributes */
	if (attr->max_entries == 0 || attr->key_size != 4 ||
	    attr->value_size == 0)
		return ERR_PTR(-EINVAL);

	if (attr->value_size >= 1 << (KMALLOC_SHIFT_MAX - 1))
		/* if value_size is bigger, the user space won't be able to
		 * access the elements.
		 */
		return ERR_PTR(-E2BIG);

	elem_size = round_up(attr->value_size, 8);

	/* check round_up into zero and u32 overflow */
	if (elem_size == 0 ||
	    attr->max_entries > (U32_MAX - PAGE_SIZE - sizeof(*array)) / elem_size)
		return ERR_PTR(-ENOMEM);

	array_size = sizeof(*array) + attr->max_entries * elem_size;

	/* allocate all map elements and zero-initialize them */
	array = kzalloc(array_size, GFP_USER | __GFP_NOWARN);
	if (!array) {
		array = vzalloc(array_size);
		if (!array)
			return ERR_PTR(-ENOMEM);
	}

	/* copy mandatory map attributes */
	array->map.key_size = attr->key_size;
	array->map.value_size = attr->value_size;
	array->map.max_entries = attr->max_entries;
	array->map.pages = round_up(array_size, PAGE_SIZE) >> PAGE_SHIFT;
	array->elem_size = elem_size;

	return &array->map;
}

/* Called from syscall or from eBPF program */
static void *array_map_lookup_elem(struct bpf_map *map, void *key)
{
	struct bpf_array *array = container_of(map, struct bpf_array, map);
	u32 index = *(u32 *)key;

	if (index >= array->map.max_entries)
		return NULL;

	return array->value + array->elem_size * index;
}

/* Called from syscall */
static int array_map_get_next_key(struct bpf_map *map, void *key, void *next_key)
{
	struct bpf_array *array = container_of(map, struct bpf_array, map);
	u32 index = *(u32 *)key;
	u32 *next = (u32 *)next_key;

	if (index >= array->map.max_entries) {
		*next = 0;
		return 0;
	}

	if (index == array->map.max_entries - 1)
		return -ENOENT;

	*next = index + 1;
	return 0;
}

/* Called from syscall or from eBPF program */
static int array_map_update_elem(struct bpf_map *map, void *key, void *value,
				 u64 map_flags)
{
	struct bpf_array *array = container_of(map, struct bpf_array, map);
	u32 index = *(u32 *)key;

	if (map_flags > BPF_EXIST)
		/* unknown flags */
		return -EINVAL;

	if (index >= array->map.max_entries)
		/* all elements were pre-allocated, cannot insert a new one */
		return -E2BIG;

	if (map_flags == BPF_NOEXIST)
		/* all elements already exist */
		return -EEXIST;

	memcpy(array->value + array->elem_size * index, value, map->value_size);
	return 0;
}

/* Called from syscall or from eBPF program */
static int array_map_delete_elem(struct bpf_map *map, void *key)
{
	return -EINVAL;
}

/* Called when map->refcnt goes to zero, either from workqueue or from syscall */
static void array_map_free(struct bpf_map *map)
{
	struct bpf_array *array = container_of(map, struct bpf_array, map);

	/* at this point bpf_prog->aux->refcnt == 0 and this map->refcnt == 0,
	 * so the programs (can be more than one that used this map) were
	 * disconnected from events. Wait for outstanding programs to complete
	 * and free the array
	 */
	synchronize_rcu();

	kvfree(array);
}

static const struct bpf_map_ops array_ops = {
	.map_alloc = array_map_alloc,
	.map_free = array_map_free,
	.map_get_next_key = array_map_get_next_key,
	.map_lookup_elem = array_map_lookup_elem,
	.map_update_elem = array_map_update_elem,
	.map_delete_elem = array_map_delete_elem,
};

static struct bpf_map_type_list array_type __read_mostly = {
	.ops = &array_ops,
	.type = BPF_MAP_TYPE_ARRAY,
};

static int __init register_array_map(void)
{
	bpf_register_map_type(&array_type);
	return 0;
}
late_initcall(register_array_map);

static struct bpf_map *fd_array_map_alloc(union bpf_attr *attr)
{
	/* only file descriptors can be stored in this type of map */
	if (attr->value_size != sizeof(u32))
		return ERR_PTR(-EINVAL);
	return array_map_alloc(attr);
}

static void fd_array_map_free(struct bpf_map *map)
{
	struct bpf_array *array = container_of(map, struct bpf_array, map);
	int i;

	synchronize_rcu();

	/* make sure it's empty */
	for (i = 0; i < array->map.max_entries; i++)
		BUG_ON(array->ptrs[i] != NULL);
	kvfree(array);
}

static void *fd_array_map_lookup_elem(struct bpf_map *map, void *key)
{
	return NULL;
}

/* only called from syscall */
static int fd_array_map_update_elem(struct bpf_map *map, void *key,
				    void *value, u64 map_flags)
{
	struct bpf_array *array = container_of(map, struct bpf_array, map);
	void *new_ptr, *old_ptr;
	u32 index = *(u32 *)key, ufd;

	if (map_flags != BPF_ANY)
		return -EINVAL;

	if (index >= array->map.max_entries)
		return -E2BIG;

	ufd = *(u32 *)value;
	new_ptr = map->ops->map_fd_get_ptr(map, ufd);
	if (IS_ERR(new_ptr))
		return PTR_ERR(new_ptr);

	old_ptr = xchg(array->ptrs + index, new_ptr);
	if (old_ptr)
		map->ops->map_fd_put_ptr(old_ptr);

	return 0;
}

static int fd_array_map_delete_elem(struct bpf_map *map, void *key)
{
	struct bpf_array *array = container_of(map, struct bpf_array, map);
	void *old_ptr;
	u32 index = *(u32 *)key;

	if (index >= array->map.max_entries)
		return -E2BIG;

	old_ptr = xchg(array->ptrs + index, NULL);
	if (old_ptr) {
		map->ops->map_fd_put_ptr(old_ptr);
		return 0;
	} else {
		return -ENOENT;
	}
}

static void *prog_fd_array_get_ptr(struct bpf_map *map, int fd)
{
	struct bpf_array *array = container_of(map, struct bpf_array, map);
	struct bpf_prog *prog = bpf_prog_get(fd);
	if (IS_ERR(prog))
		return prog;

	if (!bpf_prog_array_compatible(array, prog)) {
		bpf_prog_put(prog);
		return ERR_PTR(-EINVAL);
	}
	return prog;
}

static void prog_fd_array_put_ptr(void *ptr)
{
	struct bpf_prog *prog = ptr;

	bpf_prog_put_rcu(prog);
}

/* decrement refcnt of all bpf_progs that are stored in this map */
void bpf_fd_array_map_clear(struct bpf_map *map)
{
	struct bpf_array *array = container_of(map, struct bpf_array, map);
	int i;

	for (i = 0; i < array->map.max_entries; i++)
		fd_array_map_delete_elem(map, &i);
}

static const struct bpf_map_ops prog_array_ops = {
	.map_alloc = fd_array_map_alloc,
	.map_free = fd_array_map_free,
	.map_get_next_key = array_map_get_next_key,
	.map_lookup_elem = fd_array_map_lookup_elem,
	.map_update_elem = fd_array_map_update_elem,
	.map_delete_elem = fd_array_map_delete_elem,
	.map_fd_get_ptr = prog_fd_array_get_ptr,
	.map_fd_put_ptr = prog_fd_array_put_ptr,
};

static struct bpf_map_type_list prog_array_type __read_mostly = {
	.ops = &prog_array_ops,
	.type = BPF_MAP_TYPE_PROG_ARRAY,
};

static int __init register_prog_array_map(void)
{
	bpf_register_map_type(&prog_array_type);
	return 0;
}
late_initcall(register_prog_array_map);

static void perf_event_array_map_free(struct bpf_map *map)
{
	bpf_fd_array_map_clear(map);
	fd_array_map_free(map);
}

static void *perf_event_fd_array_get_ptr(struct bpf_map *map, int fd)
{
	struct perf_event *event;
	const struct perf_event_attr *attr;

	event = perf_event_get(fd);
	if (IS_ERR(event))
		return event;

	attr = perf_event_attrs(event);
	if (IS_ERR(attr))
		goto err;

	if (attr->inherit)
		goto err;

	if (attr->type == PERF_TYPE_RAW)
		return event;

	if (attr->type == PERF_TYPE_HARDWARE)
		return event;

	if (attr->type == PERF_TYPE_SOFTWARE &&
	    attr->config == PERF_COUNT_SW_BPF_OUTPUT)
		return event;
err:
	perf_event_release_kernel(event);
	return ERR_PTR(-EINVAL);
}

static void perf_event_fd_array_put_ptr(void *ptr)
{
	struct perf_event *event = ptr;

	perf_event_release_kernel(event);
}

static const struct bpf_map_ops perf_event_array_ops = {
	.map_alloc = fd_array_map_alloc,
	.map_free = perf_event_array_map_free,
	.map_get_next_key = array_map_get_next_key,
	.map_lookup_elem = fd_array_map_lookup_elem,
	.map_update_elem = fd_array_map_update_elem,
	.map_delete_elem = fd_array_map_delete_elem,
	.map_fd_get_ptr = perf_event_fd_array_get_ptr,
	.map_fd_put_ptr = perf_event_fd_array_put_ptr,
};

static struct bpf_map_type_list perf_event_array_type __read_mostly = {
	.ops = &perf_event_array_ops,
	.type = BPF_MAP_TYPE_PERF_EVENT_ARRAY,
};

static int __init register_perf_event_array_map(void)
{
	bpf_register_map_type(&perf_event_array_type);
	return 0;
}
late_initcall(register_perf_event_array_map);
Commit	Line	Data
28fbcfa0 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/err.h>
	14	#include <linux/vmalloc.h>
	15	#include <linux/slab.h>
	16	#include <linux/mm.h>
04fd61ab	17	#include <linux/filter.h>
0cdf5640	18	#include <linux/perf_event.h>
28fbcfa0 AS	19
	20	/* Called from syscall */
	21	static struct bpf_map array_map_alloc(union bpf_attr attr)
	22	{
	23	struct bpf_array *array;
daaf427c	24	u32 elem_size, array_size;
28fbcfa0 AS	25
	26	/* check sanity of attributes */
	27	if (attr->max_entries == 0 \|\| attr->key_size != 4 \|\|
	28	attr->value_size == 0)
	29	return ERR_PTR(-EINVAL);
	30
01b3f521 AS	31	if (attr->value_size >= 1 << (KMALLOC_SHIFT_MAX - 1))
	32	/* if value_size is bigger, the user space won't be able to
	33	* access the elements.
	34	*/
	35	return ERR_PTR(-E2BIG);
	36
28fbcfa0 AS	37	elem_size = round_up(attr->value_size, 8);
28fbcfa0 AS	38
daaf427c AS	39	/* check round_up into zero and u32 overflow */
daaf427c AS	40	if (elem_size == 0 \|\|
01b3f521	41	attr->max_entries > (U32_MAX - PAGE_SIZE - sizeof(*array)) / elem_size)
daaf427c AS	42	return ERR_PTR(-ENOMEM);
	43
	44	array_size = sizeof(array) + attr->max_entries elem_size;
	45
28fbcfa0	46	/* allocate all map elements and zero-initialize them */
daaf427c	47	array = kzalloc(array_size, GFP_USER \| __GFP_NOWARN);
28fbcfa0	48	if (!array) {
daaf427c	49	array = vzalloc(array_size);
28fbcfa0 AS	50	if (!array)
	51	return ERR_PTR(-ENOMEM);
	52	}
	53
	54	/* copy mandatory map attributes */
	55	array->map.key_size = attr->key_size;
	56	array->map.value_size = attr->value_size;
	57	array->map.max_entries = attr->max_entries;
aaac3ba9	58	array->map.pages = round_up(array_size, PAGE_SIZE) >> PAGE_SHIFT;
28fbcfa0 AS	59	array->elem_size = elem_size;
	60
	61	return &array->map;
28fbcfa0 AS	62	}
	63
	64	/* Called from syscall or from eBPF program */
	65	static void array_map_lookup_elem(struct bpf_map map, void *key)
	66	{
	67	struct bpf_array *array = container_of(map, struct bpf_array, map);
	68	u32 index = (u32 )key;
	69
	70	if (index >= array->map.max_entries)
	71	return NULL;
	72
	73	return array->value + array->elem_size * index;
	74	}
	75
	76	/* Called from syscall */
	77	static int array_map_get_next_key(struct bpf_map map, void key, void *next_key)
	78	{
	79	struct bpf_array *array = container_of(map, struct bpf_array, map);
	80	u32 index = (u32 )key;
	81	u32 next = (u32 )next_key;
	82
	83	if (index >= array->map.max_entries) {
	84	*next = 0;
	85	return 0;
	86	}
	87
	88	if (index == array->map.max_entries - 1)
	89	return -ENOENT;
	90
	91	*next = index + 1;
	92	return 0;
	93	}
	94
	95	/* Called from syscall or from eBPF program */
	96	static int array_map_update_elem(struct bpf_map map, void key, void *value,
	97	u64 map_flags)
	98	{
	99	struct bpf_array *array = container_of(map, struct bpf_array, map);
	100	u32 index = (u32 )key;
	101
	102	if (map_flags > BPF_EXIST)
	103	/* unknown flags */
	104	return -EINVAL;
	105
	106	if (index >= array->map.max_entries)
	107	/* all elements were pre-allocated, cannot insert a new one */
	108	return -E2BIG;
	109
	110	if (map_flags == BPF_NOEXIST)
daaf427c	111	/* all elements already exist */
28fbcfa0 AS	112	return -EEXIST;
28fbcfa0 AS	113
fbca9d2d	114	memcpy(array->value + array->elem_size * index, value, map->value_size);
28fbcfa0 AS	115	return 0;
	116	}
	117
	118	/* Called from syscall or from eBPF program */
	119	static int array_map_delete_elem(struct bpf_map map, void key)
	120	{
	121	return -EINVAL;
	122	}
	123
	124	/* Called when map->refcnt goes to zero, either from workqueue or from syscall */
	125	static void array_map_free(struct bpf_map *map)
	126	{
	127	struct bpf_array *array = container_of(map, struct bpf_array, map);
	128
	129	/* at this point bpf_prog->aux->refcnt == 0 and this map->refcnt == 0,
	130	* so the programs (can be more than one that used this map) were
	131	* disconnected from events. Wait for outstanding programs to complete
	132	* and free the array
	133	*/
	134	synchronize_rcu();
	135
	136	kvfree(array);
	137	}
	138
a2c83fff	139	static const struct bpf_map_ops array_ops = {
28fbcfa0 AS	140	.map_alloc = array_map_alloc,
	141	.map_free = array_map_free,
	142	.map_get_next_key = array_map_get_next_key,
	143	.map_lookup_elem = array_map_lookup_elem,
	144	.map_update_elem = array_map_update_elem,
	145	.map_delete_elem = array_map_delete_elem,
	146	};
	147
a2c83fff	148	static struct bpf_map_type_list array_type __read_mostly = {
28fbcfa0 AS	149	.ops = &array_ops,
	150	.type = BPF_MAP_TYPE_ARRAY,
	151	};
	152
	153	static int __init register_array_map(void)
	154	{
a2c83fff	155	bpf_register_map_type(&array_type);
28fbcfa0 AS	156	return 0;
	157	}
	158	late_initcall(register_array_map);
04fd61ab	159
2a36f0b9	160	static struct bpf_map fd_array_map_alloc(union bpf_attr attr)
04fd61ab	161	{
2a36f0b9	162	/* only file descriptors can be stored in this type of map */
04fd61ab AS	163	if (attr->value_size != sizeof(u32))
	164	return ERR_PTR(-EINVAL);
	165	return array_map_alloc(attr);
	166	}
	167
2a36f0b9	168	static void fd_array_map_free(struct bpf_map *map)
04fd61ab AS	169	{
	170	struct bpf_array *array = container_of(map, struct bpf_array, map);
	171	int i;
	172
	173	synchronize_rcu();
	174
	175	/* make sure it's empty */
	176	for (i = 0; i < array->map.max_entries; i++)
2a36f0b9	177	BUG_ON(array->ptrs[i] != NULL);
04fd61ab AS	178	kvfree(array);
	179	}
	180
2a36f0b9	181	static void fd_array_map_lookup_elem(struct bpf_map map, void *key)
04fd61ab AS	182	{
	183	return NULL;
	184	}
	185
	186	/* only called from syscall */
2a36f0b9 WN	187	static int fd_array_map_update_elem(struct bpf_map map, void key,
2a36f0b9 WN	188	void *value, u64 map_flags)
04fd61ab AS	189	{
04fd61ab AS	190	struct bpf_array *array = container_of(map, struct bpf_array, map);
2a36f0b9	191	void new_ptr, old_ptr;
04fd61ab AS	192	u32 index = (u32 )key, ufd;
	193
	194	if (map_flags != BPF_ANY)
	195	return -EINVAL;
	196
	197	if (index >= array->map.max_entries)
	198	return -E2BIG;
	199
	200	ufd = (u32 )value;
2a36f0b9 WN	201	new_ptr = map->ops->map_fd_get_ptr(map, ufd);
	202	if (IS_ERR(new_ptr))
	203	return PTR_ERR(new_ptr);
04fd61ab	204
2a36f0b9 WN	205	old_ptr = xchg(array->ptrs + index, new_ptr);
	206	if (old_ptr)
	207	map->ops->map_fd_put_ptr(old_ptr);
04fd61ab AS	208
	209	return 0;
	210	}
	211
2a36f0b9	212	static int fd_array_map_delete_elem(struct bpf_map map, void key)
04fd61ab AS	213	{
04fd61ab AS	214	struct bpf_array *array = container_of(map, struct bpf_array, map);
2a36f0b9	215	void *old_ptr;
04fd61ab AS	216	u32 index = (u32 )key;
	217
	218	if (index >= array->map.max_entries)
	219	return -E2BIG;
	220
2a36f0b9 WN	221	old_ptr = xchg(array->ptrs + index, NULL);
	222	if (old_ptr) {
	223	map->ops->map_fd_put_ptr(old_ptr);
04fd61ab AS	224	return 0;
	225	} else {
	226	return -ENOENT;
	227	}
	228	}
	229
2a36f0b9 WN	230	static void prog_fd_array_get_ptr(struct bpf_map map, int fd)
	231	{
	232	struct bpf_array *array = container_of(map, struct bpf_array, map);
	233	struct bpf_prog *prog = bpf_prog_get(fd);
	234	if (IS_ERR(prog))
	235	return prog;
	236
	237	if (!bpf_prog_array_compatible(array, prog)) {
	238	bpf_prog_put(prog);
	239	return ERR_PTR(-EINVAL);
	240	}
	241	return prog;
	242	}
	243
	244	static void prog_fd_array_put_ptr(void *ptr)
	245	{
	246	struct bpf_prog *prog = ptr;
	247
	248	bpf_prog_put_rcu(prog);
	249	}
	250
04fd61ab	251	/* decrement refcnt of all bpf_progs that are stored in this map */
2a36f0b9	252	void bpf_fd_array_map_clear(struct bpf_map *map)
04fd61ab AS	253	{
	254	struct bpf_array *array = container_of(map, struct bpf_array, map);
	255	int i;
	256
	257	for (i = 0; i < array->map.max_entries; i++)
2a36f0b9	258	fd_array_map_delete_elem(map, &i);
04fd61ab AS	259	}
	260
	261	static const struct bpf_map_ops prog_array_ops = {
2a36f0b9 WN	262	.map_alloc = fd_array_map_alloc,
2a36f0b9 WN	263	.map_free = fd_array_map_free,
04fd61ab	264	.map_get_next_key = array_map_get_next_key,
2a36f0b9 WN	265	.map_lookup_elem = fd_array_map_lookup_elem,
	266	.map_update_elem = fd_array_map_update_elem,
	267	.map_delete_elem = fd_array_map_delete_elem,
	268	.map_fd_get_ptr = prog_fd_array_get_ptr,
	269	.map_fd_put_ptr = prog_fd_array_put_ptr,
04fd61ab AS	270	};
	271
	272	static struct bpf_map_type_list prog_array_type __read_mostly = {
	273	.ops = &prog_array_ops,
	274	.type = BPF_MAP_TYPE_PROG_ARRAY,
	275	};
	276
	277	static int __init register_prog_array_map(void)
	278	{
	279	bpf_register_map_type(&prog_array_type);
	280	return 0;
	281	}
	282	late_initcall(register_prog_array_map);
ea317b26 KX	283
	284	static void perf_event_array_map_free(struct bpf_map *map)
	285	{
	286	bpf_fd_array_map_clear(map);
	287	fd_array_map_free(map);
	288	}
	289
	290	static void perf_event_fd_array_get_ptr(struct bpf_map map, int fd)
	291	{
	292	struct perf_event *event;
	293	const struct perf_event_attr *attr;
	294
	295	event = perf_event_get(fd);
	296	if (IS_ERR(event))
	297	return event;
	298
	299	attr = perf_event_attrs(event);
	300	if (IS_ERR(attr))
62544ce8	301	goto err;
ea317b26	302
62544ce8 AS	303	if (attr->inherit)
	304	goto err;
	305
	306	if (attr->type == PERF_TYPE_RAW)
	307	return event;
	308
	309	if (attr->type == PERF_TYPE_HARDWARE)
	310	return event;
	311
	312	if (attr->type == PERF_TYPE_SOFTWARE &&
	313	attr->config == PERF_COUNT_SW_BPF_OUTPUT)
	314	return event;
	315	err:
	316	perf_event_release_kernel(event);
	317	return ERR_PTR(-EINVAL);
ea317b26 KX	318	}
	319
	320	static void perf_event_fd_array_put_ptr(void *ptr)
	321	{
	322	struct perf_event *event = ptr;
	323
	324	perf_event_release_kernel(event);
	325	}
	326
	327	static const struct bpf_map_ops perf_event_array_ops = {
	328	.map_alloc = fd_array_map_alloc,
	329	.map_free = perf_event_array_map_free,
	330	.map_get_next_key = array_map_get_next_key,
	331	.map_lookup_elem = fd_array_map_lookup_elem,
	332	.map_update_elem = fd_array_map_update_elem,
	333	.map_delete_elem = fd_array_map_delete_elem,
	334	.map_fd_get_ptr = perf_event_fd_array_get_ptr,
	335	.map_fd_put_ptr = perf_event_fd_array_put_ptr,
	336	};
	337
	338	static struct bpf_map_type_list perf_event_array_type __read_mostly = {
	339	.ops = &perf_event_array_ops,
	340	.type = BPF_MAP_TYPE_PERF_EVENT_ARRAY,
	341	};
	342
	343	static int __init register_perf_event_array_map(void)
	344	{
	345	bpf_register_map_type(&perf_event_array_type);
	346	return 0;
	347	}
	348	late_initcall(register_perf_event_array_map);