[mirror_ubuntu-hirsute-kernel.git] / kernel / bpf / offload.c

/*
 * Copyright (C) 2017-2018 Netronome Systems, Inc.
 *
 * This software is licensed under the GNU General License Version 2,
 * June 1991 as shown in the file COPYING in the top-level directory of this
 * source tree.
 *
 * THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS"
 * WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING,
 * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
 * FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE
 * OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME
 * THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
 */

#include <linux/bpf.h>
#include <linux/bpf_verifier.h>
#include <linux/bug.h>
#include <linux/kdev_t.h>
#include <linux/list.h>
#include <linux/lockdep.h>
#include <linux/netdevice.h>
#include <linux/printk.h>
#include <linux/proc_ns.h>
#include <linux/rhashtable.h>
#include <linux/rtnetlink.h>
#include <linux/rwsem.h>

/* Protects offdevs, members of bpf_offload_netdev and offload members
 * of all progs.
 * RTNL lock cannot be taken when holding this lock.
 */
static DECLARE_RWSEM(bpf_devs_lock);

struct bpf_offload_dev {
	const struct bpf_prog_offload_ops *ops;
	struct list_head netdevs;
};

struct bpf_offload_netdev {
	struct rhash_head l;
	struct net_device *netdev;
	struct bpf_offload_dev *offdev;
	struct list_head progs;
	struct list_head maps;
	struct list_head offdev_netdevs;
};

static const struct rhashtable_params offdevs_params = {
	.nelem_hint		= 4,
	.key_len		= sizeof(struct net_device *),
	.key_offset		= offsetof(struct bpf_offload_netdev, netdev),
	.head_offset		= offsetof(struct bpf_offload_netdev, l),
	.automatic_shrinking	= true,
};

static struct rhashtable offdevs;
static bool offdevs_inited;

static int bpf_dev_offload_check(struct net_device *netdev)
{
	if (!netdev)
		return -EINVAL;
	if (!netdev->netdev_ops->ndo_bpf)
		return -EOPNOTSUPP;
	return 0;
}

static struct bpf_offload_netdev *
bpf_offload_find_netdev(struct net_device *netdev)
{
	lockdep_assert_held(&bpf_devs_lock);

	if (!offdevs_inited)
		return NULL;
	return rhashtable_lookup_fast(&offdevs, &netdev, offdevs_params);
}

int bpf_prog_offload_init(struct bpf_prog *prog, union bpf_attr *attr)
{
	struct bpf_offload_netdev *ondev;
	struct bpf_prog_offload *offload;
	int err;

	if (attr->prog_type != BPF_PROG_TYPE_SCHED_CLS &&
	    attr->prog_type != BPF_PROG_TYPE_XDP)
		return -EINVAL;

	if (attr->prog_flags)
		return -EINVAL;

	offload = kzalloc(sizeof(*offload), GFP_USER);
	if (!offload)
		return -ENOMEM;

	offload->prog = prog;

	offload->netdev = dev_get_by_index(current->nsproxy->net_ns,
					   attr->prog_ifindex);
	err = bpf_dev_offload_check(offload->netdev);
	if (err)
		goto err_maybe_put;

	down_write(&bpf_devs_lock);
	ondev = bpf_offload_find_netdev(offload->netdev);
	if (!ondev) {
		err = -EINVAL;
		goto err_unlock;
	}
	offload->offdev = ondev->offdev;
	prog->aux->offload = offload;
	list_add_tail(&offload->offloads, &ondev->progs);
	dev_put(offload->netdev);
	up_write(&bpf_devs_lock);

	return 0;
err_unlock:
	up_write(&bpf_devs_lock);
err_maybe_put:
	if (offload->netdev)
		dev_put(offload->netdev);
	kfree(offload);
	return err;
}

static int __bpf_offload_ndo(struct bpf_prog *prog, enum bpf_netdev_command cmd,
			     struct netdev_bpf *data)
{
	struct bpf_prog_offload *offload = prog->aux->offload;
	struct net_device *netdev;

	ASSERT_RTNL();

	if (!offload)
		return -ENODEV;
	netdev = offload->netdev;

	data->command = cmd;

	return netdev->netdev_ops->ndo_bpf(netdev, data);
}

int bpf_prog_offload_verifier_prep(struct bpf_verifier_env *env)
{
	struct bpf_prog_offload *offload;
	int ret = -ENODEV;

	down_read(&bpf_devs_lock);
	offload = env->prog->aux->offload;
	if (offload)
		ret = offload->offdev->ops->prepare(offload->netdev, env);
	offload->dev_state = !ret;
	up_read(&bpf_devs_lock);

	return ret;
}

int bpf_prog_offload_verify_insn(struct bpf_verifier_env *env,
				 int insn_idx, int prev_insn_idx)
{
	struct bpf_prog_offload *offload;
	int ret = -ENODEV;

	down_read(&bpf_devs_lock);
	offload = env->prog->aux->offload;
	if (offload)
		ret = offload->offdev->ops->insn_hook(env, insn_idx,
						      prev_insn_idx);
	up_read(&bpf_devs_lock);

	return ret;
}

int bpf_prog_offload_finalize(struct bpf_verifier_env *env)
{
	struct bpf_prog_offload *offload;
	int ret = -ENODEV;

	down_read(&bpf_devs_lock);
	offload = env->prog->aux->offload;
	if (offload) {
		if (offload->offdev->ops->finalize)
			ret = offload->offdev->ops->finalize(env);
		else
			ret = 0;
	}
	up_read(&bpf_devs_lock);

	return ret;
}

static void __bpf_prog_offload_destroy(struct bpf_prog *prog)
{
	struct bpf_prog_offload *offload = prog->aux->offload;
	struct netdev_bpf data = {};

	data.offload.prog = prog;

	if (offload->dev_state)
		WARN_ON(__bpf_offload_ndo(prog, BPF_OFFLOAD_DESTROY, &data));

	/* Make sure BPF_PROG_GET_NEXT_ID can't find this dead program */
	bpf_prog_free_id(prog, true);

	list_del_init(&offload->offloads);
	kfree(offload);
	prog->aux->offload = NULL;
}

void bpf_prog_offload_destroy(struct bpf_prog *prog)
{
	rtnl_lock();
	down_write(&bpf_devs_lock);
	if (prog->aux->offload)
		__bpf_prog_offload_destroy(prog);
	up_write(&bpf_devs_lock);
	rtnl_unlock();
}

static int bpf_prog_offload_translate(struct bpf_prog *prog)
{
	struct netdev_bpf data = {};
	int ret;

	data.offload.prog = prog;

	rtnl_lock();
	ret = __bpf_offload_ndo(prog, BPF_OFFLOAD_TRANSLATE, &data);
	rtnl_unlock();

	return ret;
}

static unsigned int bpf_prog_warn_on_exec(const void *ctx,
					  const struct bpf_insn *insn)
{
	WARN(1, "attempt to execute device eBPF program on the host!");
	return 0;
}

int bpf_prog_offload_compile(struct bpf_prog *prog)
{
	prog->bpf_func = bpf_prog_warn_on_exec;

	return bpf_prog_offload_translate(prog);
}

struct ns_get_path_bpf_prog_args {
	struct bpf_prog *prog;
	struct bpf_prog_info *info;
};

static struct ns_common *bpf_prog_offload_info_fill_ns(void *private_data)
{
	struct ns_get_path_bpf_prog_args *args = private_data;
	struct bpf_prog_aux *aux = args->prog->aux;
	struct ns_common *ns;
	struct net *net;

	rtnl_lock();
	down_read(&bpf_devs_lock);

	if (aux->offload) {
		args->info->ifindex = aux->offload->netdev->ifindex;
		net = dev_net(aux->offload->netdev);
		get_net(net);
		ns = &net->ns;
	} else {
		args->info->ifindex = 0;
		ns = NULL;
	}

	up_read(&bpf_devs_lock);
	rtnl_unlock();

	return ns;
}

int bpf_prog_offload_info_fill(struct bpf_prog_info *info,
			       struct bpf_prog *prog)
{
	struct ns_get_path_bpf_prog_args args = {
		.prog	= prog,
		.info	= info,
	};
	struct bpf_prog_aux *aux = prog->aux;
	struct inode *ns_inode;
	struct path ns_path;
	char __user *uinsns;
	void *res;
	u32 ulen;

	res = ns_get_path_cb(&ns_path, bpf_prog_offload_info_fill_ns, &args);
	if (IS_ERR(res)) {
		if (!info->ifindex)
			return -ENODEV;
		return PTR_ERR(res);
	}

	down_read(&bpf_devs_lock);

	if (!aux->offload) {
		up_read(&bpf_devs_lock);
		return -ENODEV;
	}

	ulen = info->jited_prog_len;
	info->jited_prog_len = aux->offload->jited_len;
	if (info->jited_prog_len & ulen) {
		uinsns = u64_to_user_ptr(info->jited_prog_insns);
		ulen = min_t(u32, info->jited_prog_len, ulen);
		if (copy_to_user(uinsns, aux->offload->jited_image, ulen)) {
			up_read(&bpf_devs_lock);
			return -EFAULT;
		}
	}

	up_read(&bpf_devs_lock);

	ns_inode = ns_path.dentry->d_inode;
	info->netns_dev = new_encode_dev(ns_inode->i_sb->s_dev);
	info->netns_ino = ns_inode->i_ino;
	path_put(&ns_path);

	return 0;
}

const struct bpf_prog_ops bpf_offload_prog_ops = {
};

static int bpf_map_offload_ndo(struct bpf_offloaded_map *offmap,
			       enum bpf_netdev_command cmd)
{
	struct netdev_bpf data = {};
	struct net_device *netdev;

	ASSERT_RTNL();

	data.command = cmd;
	data.offmap = offmap;
	/* Caller must make sure netdev is valid */
	netdev = offmap->netdev;

	return netdev->netdev_ops->ndo_bpf(netdev, &data);
}

struct bpf_map *bpf_map_offload_map_alloc(union bpf_attr *attr)
{
	struct net *net = current->nsproxy->net_ns;
	struct bpf_offload_netdev *ondev;
	struct bpf_offloaded_map *offmap;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return ERR_PTR(-EPERM);
	if (attr->map_type != BPF_MAP_TYPE_ARRAY &&
	    attr->map_type != BPF_MAP_TYPE_HASH)
		return ERR_PTR(-EINVAL);

	offmap = kzalloc(sizeof(*offmap), GFP_USER);
	if (!offmap)
		return ERR_PTR(-ENOMEM);

	bpf_map_init_from_attr(&offmap->map, attr);

	rtnl_lock();
	down_write(&bpf_devs_lock);
	offmap->netdev = __dev_get_by_index(net, attr->map_ifindex);
	err = bpf_dev_offload_check(offmap->netdev);
	if (err)
		goto err_unlock;

	ondev = bpf_offload_find_netdev(offmap->netdev);
	if (!ondev) {
		err = -EINVAL;
		goto err_unlock;
	}

	err = bpf_map_offload_ndo(offmap, BPF_OFFLOAD_MAP_ALLOC);
	if (err)
		goto err_unlock;

	list_add_tail(&offmap->offloads, &ondev->maps);
	up_write(&bpf_devs_lock);
	rtnl_unlock();

	return &offmap->map;

err_unlock:
	up_write(&bpf_devs_lock);
	rtnl_unlock();
	kfree(offmap);
	return ERR_PTR(err);
}

static void __bpf_map_offload_destroy(struct bpf_offloaded_map *offmap)
{
	WARN_ON(bpf_map_offload_ndo(offmap, BPF_OFFLOAD_MAP_FREE));
	/* Make sure BPF_MAP_GET_NEXT_ID can't find this dead map */
	bpf_map_free_id(&offmap->map, true);
	list_del_init(&offmap->offloads);
	offmap->netdev = NULL;
}

void bpf_map_offload_map_free(struct bpf_map *map)
{
	struct bpf_offloaded_map *offmap = map_to_offmap(map);

	rtnl_lock();
	down_write(&bpf_devs_lock);
	if (offmap->netdev)
		__bpf_map_offload_destroy(offmap);
	up_write(&bpf_devs_lock);
	rtnl_unlock();

	kfree(offmap);
}

int bpf_map_offload_lookup_elem(struct bpf_map *map, void *key, void *value)
{
	struct bpf_offloaded_map *offmap = map_to_offmap(map);
	int ret = -ENODEV;

	down_read(&bpf_devs_lock);
	if (offmap->netdev)
		ret = offmap->dev_ops->map_lookup_elem(offmap, key, value);
	up_read(&bpf_devs_lock);

	return ret;
}

int bpf_map_offload_update_elem(struct bpf_map *map,
				void *key, void *value, u64 flags)
{
	struct bpf_offloaded_map *offmap = map_to_offmap(map);
	int ret = -ENODEV;

	if (unlikely(flags > BPF_EXIST))
		return -EINVAL;

	down_read(&bpf_devs_lock);
	if (offmap->netdev)
		ret = offmap->dev_ops->map_update_elem(offmap, key, value,
						       flags);
	up_read(&bpf_devs_lock);

	return ret;
}

int bpf_map_offload_delete_elem(struct bpf_map *map, void *key)
{
	struct bpf_offloaded_map *offmap = map_to_offmap(map);
	int ret = -ENODEV;

	down_read(&bpf_devs_lock);
	if (offmap->netdev)
		ret = offmap->dev_ops->map_delete_elem(offmap, key);
	up_read(&bpf_devs_lock);

	return ret;
}

int bpf_map_offload_get_next_key(struct bpf_map *map, void *key, void *next_key)
{
	struct bpf_offloaded_map *offmap = map_to_offmap(map);
	int ret = -ENODEV;

	down_read(&bpf_devs_lock);
	if (offmap->netdev)
		ret = offmap->dev_ops->map_get_next_key(offmap, key, next_key);
	up_read(&bpf_devs_lock);

	return ret;
}

struct ns_get_path_bpf_map_args {
	struct bpf_offloaded_map *offmap;
	struct bpf_map_info *info;
};

static struct ns_common *bpf_map_offload_info_fill_ns(void *private_data)
{
	struct ns_get_path_bpf_map_args *args = private_data;
	struct ns_common *ns;
	struct net *net;

	rtnl_lock();
	down_read(&bpf_devs_lock);

	if (args->offmap->netdev) {
		args->info->ifindex = args->offmap->netdev->ifindex;
		net = dev_net(args->offmap->netdev);
		get_net(net);
		ns = &net->ns;
	} else {
		args->info->ifindex = 0;
		ns = NULL;
	}

	up_read(&bpf_devs_lock);
	rtnl_unlock();

	return ns;
}

int bpf_map_offload_info_fill(struct bpf_map_info *info, struct bpf_map *map)
{
	struct ns_get_path_bpf_map_args args = {
		.offmap	= map_to_offmap(map),
		.info	= info,
	};
	struct inode *ns_inode;
	struct path ns_path;
	void *res;

	res = ns_get_path_cb(&ns_path, bpf_map_offload_info_fill_ns, &args);
	if (IS_ERR(res)) {
		if (!info->ifindex)
			return -ENODEV;
		return PTR_ERR(res);
	}

	ns_inode = ns_path.dentry->d_inode;
	info->netns_dev = new_encode_dev(ns_inode->i_sb->s_dev);
	info->netns_ino = ns_inode->i_ino;
	path_put(&ns_path);

	return 0;
}

static bool __bpf_offload_dev_match(struct bpf_prog *prog,
				    struct net_device *netdev)
{
	struct bpf_offload_netdev *ondev1, *ondev2;
	struct bpf_prog_offload *offload;

	if (!bpf_prog_is_dev_bound(prog->aux))
		return false;

	offload = prog->aux->offload;
	if (!offload)
		return false;
	if (offload->netdev == netdev)
		return true;

	ondev1 = bpf_offload_find_netdev(offload->netdev);
	ondev2 = bpf_offload_find_netdev(netdev);

	return ondev1 && ondev2 && ondev1->offdev == ondev2->offdev;
}

bool bpf_offload_dev_match(struct bpf_prog *prog, struct net_device *netdev)
{
	bool ret;

	down_read(&bpf_devs_lock);
	ret = __bpf_offload_dev_match(prog, netdev);
	up_read(&bpf_devs_lock);

	return ret;
}
EXPORT_SYMBOL_GPL(bpf_offload_dev_match);

bool bpf_offload_prog_map_match(struct bpf_prog *prog, struct bpf_map *map)
{
	struct bpf_offloaded_map *offmap;
	bool ret;

	if (!bpf_map_is_dev_bound(map))
		return bpf_map_offload_neutral(map);
	offmap = map_to_offmap(map);

	down_read(&bpf_devs_lock);
	ret = __bpf_offload_dev_match(prog, offmap->netdev);
	up_read(&bpf_devs_lock);

	return ret;
}

int bpf_offload_dev_netdev_register(struct bpf_offload_dev *offdev,
				    struct net_device *netdev)
{
	struct bpf_offload_netdev *ondev;
	int err;

	ondev = kzalloc(sizeof(*ondev), GFP_KERNEL);
	if (!ondev)
		return -ENOMEM;

	ondev->netdev = netdev;
	ondev->offdev = offdev;
	INIT_LIST_HEAD(&ondev->progs);
	INIT_LIST_HEAD(&ondev->maps);

	down_write(&bpf_devs_lock);
	err = rhashtable_insert_fast(&offdevs, &ondev->l, offdevs_params);
	if (err) {
		netdev_warn(netdev, "failed to register for BPF offload\n");
		goto err_unlock_free;
	}

	list_add(&ondev->offdev_netdevs, &offdev->netdevs);
	up_write(&bpf_devs_lock);
	return 0;

err_unlock_free:
	up_write(&bpf_devs_lock);
	kfree(ondev);
	return err;
}
EXPORT_SYMBOL_GPL(bpf_offload_dev_netdev_register);

void bpf_offload_dev_netdev_unregister(struct bpf_offload_dev *offdev,
				       struct net_device *netdev)
{
	struct bpf_offload_netdev *ondev, *altdev;
	struct bpf_offloaded_map *offmap, *mtmp;
	struct bpf_prog_offload *offload, *ptmp;

	ASSERT_RTNL();

	down_write(&bpf_devs_lock);
	ondev = rhashtable_lookup_fast(&offdevs, &netdev, offdevs_params);
	if (WARN_ON(!ondev))
		goto unlock;

	WARN_ON(rhashtable_remove_fast(&offdevs, &ondev->l, offdevs_params));
	list_del(&ondev->offdev_netdevs);

	/* Try to move the objects to another netdev of the device */
	altdev = list_first_entry_or_null(&offdev->netdevs,
					  struct bpf_offload_netdev,
					  offdev_netdevs);
	if (altdev) {
		list_for_each_entry(offload, &ondev->progs, offloads)
			offload->netdev = altdev->netdev;
		list_splice_init(&ondev->progs, &altdev->progs);

		list_for_each_entry(offmap, &ondev->maps, offloads)
			offmap->netdev = altdev->netdev;
		list_splice_init(&ondev->maps, &altdev->maps);
	} else {
		list_for_each_entry_safe(offload, ptmp, &ondev->progs, offloads)
			__bpf_prog_offload_destroy(offload->prog);
		list_for_each_entry_safe(offmap, mtmp, &ondev->maps, offloads)
			__bpf_map_offload_destroy(offmap);
	}

	WARN_ON(!list_empty(&ondev->progs));
	WARN_ON(!list_empty(&ondev->maps));
	kfree(ondev);
unlock:
	up_write(&bpf_devs_lock);
}
EXPORT_SYMBOL_GPL(bpf_offload_dev_netdev_unregister);

struct bpf_offload_dev *
bpf_offload_dev_create(const struct bpf_prog_offload_ops *ops)
{
	struct bpf_offload_dev *offdev;
	int err;

	down_write(&bpf_devs_lock);
	if (!offdevs_inited) {
		err = rhashtable_init(&offdevs, &offdevs_params);
		if (err)
			return ERR_PTR(err);
		offdevs_inited = true;
	}
	up_write(&bpf_devs_lock);

	offdev = kzalloc(sizeof(*offdev), GFP_KERNEL);
	if (!offdev)
		return ERR_PTR(-ENOMEM);

	offdev->ops = ops;
	INIT_LIST_HEAD(&offdev->netdevs);

	return offdev;
}
EXPORT_SYMBOL_GPL(bpf_offload_dev_create);

void bpf_offload_dev_destroy(struct bpf_offload_dev *offdev)
{
	WARN_ON(!list_empty(&offdev->netdevs));
	kfree(offdev);
}
EXPORT_SYMBOL_GPL(bpf_offload_dev_destroy);
Commit	Line	Data
a39e17b2	1	/*
0cd3cbed	2	* Copyright (C) 2017-2018 Netronome Systems, Inc.
a39e17b2 JK	3	*
	4	* This software is licensed under the GNU General License Version 2,
	5	* June 1991 as shown in the file COPYING in the top-level directory of this
	6	* source tree.
	7	*
	8	* THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS"
	9	* WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING,
	10	* BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
	11	* FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE
	12	* OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME
	13	* THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
	14	*/
	15
ab3f0063 JK	16	#include <linux/bpf.h>
	17	#include <linux/bpf_verifier.h>
	18	#include <linux/bug.h>
675fc275	19	#include <linux/kdev_t.h>
ab3f0063	20	#include <linux/list.h>
9fd7c555	21	#include <linux/lockdep.h>
ab3f0063 JK	22	#include <linux/netdevice.h>
ab3f0063 JK	23	#include <linux/printk.h>
675fc275	24	#include <linux/proc_ns.h>
9fd7c555	25	#include <linux/rhashtable.h>
ab3f0063	26	#include <linux/rtnetlink.h>
e0d3974a	27	#include <linux/rwsem.h>
ab3f0063	28
9fd7c555	29	/* Protects offdevs, members of bpf_offload_netdev and offload members
a3884572	30	* of all progs.
e0d3974a JK	31	* RTNL lock cannot be taken when holding this lock.
	32	*/
	33	static DECLARE_RWSEM(bpf_devs_lock);
9fd7c555	34
602144c2	35	struct bpf_offload_dev {
1385d755	36	const struct bpf_prog_offload_ops *ops;
602144c2 JK	37	struct list_head netdevs;
	38	};
	39
9fd7c555 JK	40	struct bpf_offload_netdev {
	41	struct rhash_head l;
	42	struct net_device *netdev;
602144c2	43	struct bpf_offload_dev *offdev;
9fd7c555 JK	44	struct list_head progs;
9fd7c555 JK	45	struct list_head maps;
602144c2	46	struct list_head offdev_netdevs;
9fd7c555 JK	47	};
	48
	49	static const struct rhashtable_params offdevs_params = {
	50	.nelem_hint = 4,
	51	.key_len = sizeof(struct net_device *),
	52	.key_offset = offsetof(struct bpf_offload_netdev, netdev),
	53	.head_offset = offsetof(struct bpf_offload_netdev, l),
	54	.automatic_shrinking = true,
	55	};
	56
	57	static struct rhashtable offdevs;
	58	static bool offdevs_inited;
ab3f0063	59
5bc2d55c JK	60	static int bpf_dev_offload_check(struct net_device *netdev)
	61	{
	62	if (!netdev)
	63	return -EINVAL;
	64	if (!netdev->netdev_ops->ndo_bpf)
	65	return -EOPNOTSUPP;
	66	return 0;
	67	}
	68
9fd7c555 JK	69	static struct bpf_offload_netdev *
	70	bpf_offload_find_netdev(struct net_device *netdev)
	71	{
	72	lockdep_assert_held(&bpf_devs_lock);
	73
	74	if (!offdevs_inited)
	75	return NULL;
	76	return rhashtable_lookup_fast(&offdevs, &netdev, offdevs_params);
	77	}
	78
ab3f0063 JK	79	int bpf_prog_offload_init(struct bpf_prog prog, union bpf_attr attr)
ab3f0063 JK	80	{
9fd7c555	81	struct bpf_offload_netdev *ondev;
0a9c1991	82	struct bpf_prog_offload *offload;
5bc2d55c	83	int err;
ab3f0063	84
649f11dc JK	85	if (attr->prog_type != BPF_PROG_TYPE_SCHED_CLS &&
	86	attr->prog_type != BPF_PROG_TYPE_XDP)
	87	return -EINVAL;
ab3f0063 JK	88
	89	if (attr->prog_flags)
	90	return -EINVAL;
	91
	92	offload = kzalloc(sizeof(*offload), GFP_USER);
	93	if (!offload)
	94	return -ENOMEM;
	95
	96	offload->prog = prog;
ab3f0063	97
e0d3974a JK	98	offload->netdev = dev_get_by_index(current->nsproxy->net_ns,
e0d3974a JK	99	attr->prog_ifindex);
5bc2d55c JK	100	err = bpf_dev_offload_check(offload->netdev);
	101	if (err)
	102	goto err_maybe_put;
ab3f0063	103
e0d3974a	104	down_write(&bpf_devs_lock);
9fd7c555 JK	105	ondev = bpf_offload_find_netdev(offload->netdev);
9fd7c555 JK	106	if (!ondev) {
5bc2d55c	107	err = -EINVAL;
e0d3974a	108	goto err_unlock;
5bc2d55c	109	}
341b3e7b	110	offload->offdev = ondev->offdev;
ab3f0063	111	prog->aux->offload = offload;
9fd7c555	112	list_add_tail(&offload->offloads, &ondev->progs);
e0d3974a JK	113	dev_put(offload->netdev);
e0d3974a JK	114	up_write(&bpf_devs_lock);
ab3f0063 JK	115
ab3f0063 JK	116	return 0;
e0d3974a JK	117	err_unlock:
e0d3974a JK	118	up_write(&bpf_devs_lock);
5bc2d55c JK	119	err_maybe_put:
	120	if (offload->netdev)
	121	dev_put(offload->netdev);
e0d3974a	122	kfree(offload);
5bc2d55c	123	return err;
ab3f0063 JK	124	}
	125
	126	static int __bpf_offload_ndo(struct bpf_prog *prog, enum bpf_netdev_command cmd,
	127	struct netdev_bpf *data)
	128	{
0a9c1991	129	struct bpf_prog_offload *offload = prog->aux->offload;
ce3b9db4	130	struct net_device *netdev;
ab3f0063 JK	131
	132	ASSERT_RTNL();
	133
ce3b9db4	134	if (!offload)
ab3f0063	135	return -ENODEV;
ce3b9db4	136	netdev = offload->netdev;
ab3f0063 JK	137
	138	data->command = cmd;
	139
	140	return netdev->netdev_ops->ndo_bpf(netdev, data);
	141	}
	142
	143	int bpf_prog_offload_verifier_prep(struct bpf_verifier_env *env)
	144	{
00db12c3 QM	145	struct bpf_prog_offload *offload;
00db12c3 QM	146	int ret = -ENODEV;
ab3f0063	147
00db12c3 QM	148	down_read(&bpf_devs_lock);
	149	offload = env->prog->aux->offload;
	150	if (offload)
	151	ret = offload->offdev->ops->prepare(offload->netdev, env);
	152	offload->dev_state = !ret;
	153	up_read(&bpf_devs_lock);
ab3f0063	154
00db12c3	155	return ret;
ab3f0063 JK	156	}
ab3f0063 JK	157
cae1927c JK	158	int bpf_prog_offload_verify_insn(struct bpf_verifier_env *env,
	159	int insn_idx, int prev_insn_idx)
	160	{
0a9c1991	161	struct bpf_prog_offload *offload;
cae1927c JK	162	int ret = -ENODEV;
	163
	164	down_read(&bpf_devs_lock);
	165	offload = env->prog->aux->offload;
ce3b9db4	166	if (offload)
341b3e7b QM	167	ret = offload->offdev->ops->insn_hook(env, insn_idx,
341b3e7b QM	168	prev_insn_idx);
cae1927c JK	169	up_read(&bpf_devs_lock);
	170
	171	return ret;
	172	}
	173
c941ce9c QM	174	int bpf_prog_offload_finalize(struct bpf_verifier_env *env)
	175	{
	176	struct bpf_prog_offload *offload;
	177	int ret = -ENODEV;
	178
	179	down_read(&bpf_devs_lock);
	180	offload = env->prog->aux->offload;
	181	if (offload) {
6dc18fa6 QM	182	if (offload->offdev->ops->finalize)
6dc18fa6 QM	183	ret = offload->offdev->ops->finalize(env);
c941ce9c QM	184	else
	185	ret = 0;
	186	}
	187	up_read(&bpf_devs_lock);
	188
	189	return ret;
	190	}
	191
ab3f0063 JK	192	static void __bpf_prog_offload_destroy(struct bpf_prog *prog)
ab3f0063 JK	193	{
0a9c1991	194	struct bpf_prog_offload *offload = prog->aux->offload;
ab3f0063 JK	195	struct netdev_bpf data = {};
	196
	197	data.offload.prog = prog;
	198
ab3f0063 JK	199	if (offload->dev_state)
	200	WARN_ON(__bpf_offload_ndo(prog, BPF_OFFLOAD_DESTROY, &data));
	201
ad8ad79f JK	202	/* Make sure BPF_PROG_GET_NEXT_ID can't find this dead program */
	203	bpf_prog_free_id(prog, true);
	204
ab3f0063	205	list_del_init(&offload->offloads);
ce3b9db4 JK	206	kfree(offload);
ce3b9db4 JK	207	prog->aux->offload = NULL;
ab3f0063 JK	208	}
	209
	210	void bpf_prog_offload_destroy(struct bpf_prog *prog)
	211	{
ab3f0063	212	rtnl_lock();
e0d3974a	213	down_write(&bpf_devs_lock);
ce3b9db4 JK	214	if (prog->aux->offload)
ce3b9db4 JK	215	__bpf_prog_offload_destroy(prog);
e0d3974a	216	up_write(&bpf_devs_lock);
ab3f0063	217	rtnl_unlock();
ab3f0063 JK	218	}
	219
	220	static int bpf_prog_offload_translate(struct bpf_prog *prog)
	221	{
ab3f0063 JK	222	struct netdev_bpf data = {};
	223	int ret;
	224
	225	data.offload.prog = prog;
	226
ab3f0063 JK	227	rtnl_lock();
	228	ret = __bpf_offload_ndo(prog, BPF_OFFLOAD_TRANSLATE, &data);
	229	rtnl_unlock();
	230
	231	return ret;
	232	}
	233
	234	static unsigned int bpf_prog_warn_on_exec(const void *ctx,
	235	const struct bpf_insn *insn)
	236	{
	237	WARN(1, "attempt to execute device eBPF program on the host!");
	238	return 0;
	239	}
	240
	241	int bpf_prog_offload_compile(struct bpf_prog *prog)
	242	{
	243	prog->bpf_func = bpf_prog_warn_on_exec;
	244
	245	return bpf_prog_offload_translate(prog);
	246	}
	247
675fc275 JK	248	struct ns_get_path_bpf_prog_args {
	249	struct bpf_prog *prog;
	250	struct bpf_prog_info *info;
	251	};
	252
	253	static struct ns_common bpf_prog_offload_info_fill_ns(void private_data)
	254	{
	255	struct ns_get_path_bpf_prog_args *args = private_data;
	256	struct bpf_prog_aux *aux = args->prog->aux;
	257	struct ns_common *ns;
	258	struct net *net;
	259
	260	rtnl_lock();
	261	down_read(&bpf_devs_lock);
	262
	263	if (aux->offload) {
	264	args->info->ifindex = aux->offload->netdev->ifindex;
	265	net = dev_net(aux->offload->netdev);
	266	get_net(net);
	267	ns = &net->ns;
	268	} else {
	269	args->info->ifindex = 0;
	270	ns = NULL;
	271	}
	272
	273	up_read(&bpf_devs_lock);
	274	rtnl_unlock();
	275
	276	return ns;
	277	}
	278
	279	int bpf_prog_offload_info_fill(struct bpf_prog_info *info,
	280	struct bpf_prog *prog)
	281	{
	282	struct ns_get_path_bpf_prog_args args = {
	283	.prog = prog,
	284	.info = info,
	285	};
fcfb126d	286	struct bpf_prog_aux *aux = prog->aux;
675fc275 JK	287	struct inode *ns_inode;
675fc275 JK	288	struct path ns_path;
fcfb126d	289	char __user *uinsns;
675fc275	290	void *res;
fcfb126d	291	u32 ulen;
675fc275 JK	292
	293	res = ns_get_path_cb(&ns_path, bpf_prog_offload_info_fill_ns, &args);
	294	if (IS_ERR(res)) {
	295	if (!info->ifindex)
	296	return -ENODEV;
	297	return PTR_ERR(res);
	298	}
	299
fcfb126d JW	300	down_read(&bpf_devs_lock);
	301
	302	if (!aux->offload) {
	303	up_read(&bpf_devs_lock);
	304	return -ENODEV;
	305	}
	306
	307	ulen = info->jited_prog_len;
	308	info->jited_prog_len = aux->offload->jited_len;
	309	if (info->jited_prog_len & ulen) {
	310	uinsns = u64_to_user_ptr(info->jited_prog_insns);
	311	ulen = min_t(u32, info->jited_prog_len, ulen);
	312	if (copy_to_user(uinsns, aux->offload->jited_image, ulen)) {
	313	up_read(&bpf_devs_lock);
	314	return -EFAULT;
	315	}
	316	}
	317
	318	up_read(&bpf_devs_lock);
	319
675fc275 JK	320	ns_inode = ns_path.dentry->d_inode;
	321	info->netns_dev = new_encode_dev(ns_inode->i_sb->s_dev);
	322	info->netns_ino = ns_inode->i_ino;
	323	path_put(&ns_path);
	324
	325	return 0;
	326	}
	327
ab3f0063 JK	328	const struct bpf_prog_ops bpf_offload_prog_ops = {
	329	};
	330
a3884572 JK	331	static int bpf_map_offload_ndo(struct bpf_offloaded_map *offmap,
	332	enum bpf_netdev_command cmd)
	333	{
	334	struct netdev_bpf data = {};
	335	struct net_device *netdev;
	336
	337	ASSERT_RTNL();
	338
	339	data.command = cmd;
	340	data.offmap = offmap;
	341	/* Caller must make sure netdev is valid */
	342	netdev = offmap->netdev;
	343
	344	return netdev->netdev_ops->ndo_bpf(netdev, &data);
	345	}
	346
	347	struct bpf_map bpf_map_offload_map_alloc(union bpf_attr attr)
	348	{
	349	struct net *net = current->nsproxy->net_ns;
9fd7c555	350	struct bpf_offload_netdev *ondev;
a3884572 JK	351	struct bpf_offloaded_map *offmap;
	352	int err;
	353
	354	if (!capable(CAP_SYS_ADMIN))
	355	return ERR_PTR(-EPERM);
7a0ef693 JK	356	if (attr->map_type != BPF_MAP_TYPE_ARRAY &&
7a0ef693 JK	357	attr->map_type != BPF_MAP_TYPE_HASH)
a3884572 JK	358	return ERR_PTR(-EINVAL);
	359
	360	offmap = kzalloc(sizeof(*offmap), GFP_USER);
	361	if (!offmap)
	362	return ERR_PTR(-ENOMEM);
	363
	364	bpf_map_init_from_attr(&offmap->map, attr);
	365
	366	rtnl_lock();
	367	down_write(&bpf_devs_lock);
	368	offmap->netdev = __dev_get_by_index(net, attr->map_ifindex);
	369	err = bpf_dev_offload_check(offmap->netdev);
	370	if (err)
	371	goto err_unlock;
	372
9fd7c555 JK	373	ondev = bpf_offload_find_netdev(offmap->netdev);
	374	if (!ondev) {
	375	err = -EINVAL;
	376	goto err_unlock;
	377	}
	378
a3884572 JK	379	err = bpf_map_offload_ndo(offmap, BPF_OFFLOAD_MAP_ALLOC);
	380	if (err)
	381	goto err_unlock;
	382
9fd7c555	383	list_add_tail(&offmap->offloads, &ondev->maps);
a3884572 JK	384	up_write(&bpf_devs_lock);
	385	rtnl_unlock();
	386
	387	return &offmap->map;
	388
	389	err_unlock:
	390	up_write(&bpf_devs_lock);
	391	rtnl_unlock();
	392	kfree(offmap);
	393	return ERR_PTR(err);
	394	}
	395
	396	static void __bpf_map_offload_destroy(struct bpf_offloaded_map *offmap)
	397	{
	398	WARN_ON(bpf_map_offload_ndo(offmap, BPF_OFFLOAD_MAP_FREE));
	399	/* Make sure BPF_MAP_GET_NEXT_ID can't find this dead map */
	400	bpf_map_free_id(&offmap->map, true);
	401	list_del_init(&offmap->offloads);
	402	offmap->netdev = NULL;
	403	}
	404
	405	void bpf_map_offload_map_free(struct bpf_map *map)
	406	{
	407	struct bpf_offloaded_map *offmap = map_to_offmap(map);
	408
	409	rtnl_lock();
	410	down_write(&bpf_devs_lock);
	411	if (offmap->netdev)
	412	__bpf_map_offload_destroy(offmap);
	413	up_write(&bpf_devs_lock);
	414	rtnl_unlock();
	415
	416	kfree(offmap);
	417	}
	418
	419	int bpf_map_offload_lookup_elem(struct bpf_map map, void key, void *value)
	420	{
	421	struct bpf_offloaded_map *offmap = map_to_offmap(map);
	422	int ret = -ENODEV;
	423
	424	down_read(&bpf_devs_lock);
	425	if (offmap->netdev)
	426	ret = offmap->dev_ops->map_lookup_elem(offmap, key, value);
	427	up_read(&bpf_devs_lock);
	428
	429	return ret;
	430	}
	431
	432	int bpf_map_offload_update_elem(struct bpf_map *map,
	433	void key, void value, u64 flags)
	434	{
	435	struct bpf_offloaded_map *offmap = map_to_offmap(map);
	436	int ret = -ENODEV;
	437
	438	if (unlikely(flags > BPF_EXIST))
	439	return -EINVAL;
	440
	441	down_read(&bpf_devs_lock);
	442	if (offmap->netdev)
	443	ret = offmap->dev_ops->map_update_elem(offmap, key, value,
	444	flags);
	445	up_read(&bpf_devs_lock);
	446
	447	return ret;
448	}
449
450	int bpf_map_offload_delete_elem(struct bpf_map map, void key)
451	{
452	struct bpf_offloaded_map *offmap = map_to_offmap(map);
453	int ret = -ENODEV;
454
455	down_read(&bpf_devs_lock);
456	if (offmap->netdev)
457	ret = offmap->dev_ops->map_delete_elem(offmap, key);
458	up_read(&bpf_devs_lock);
459
460	return ret;
461	}
462
463	int bpf_map_offload_get_next_key(struct bpf_map map, void key, void *next_key)
464	{
465	struct bpf_offloaded_map *offmap = map_to_offmap(map);
466	int ret = -ENODEV;
467
468	down_read(&bpf_devs_lock);
469	if (offmap->netdev)
470	ret = offmap->dev_ops->map_get_next_key(offmap, key, next_key);
471	up_read(&bpf_devs_lock);
472
473	return ret;
474	}
475
52775b33 JK	476	struct ns_get_path_bpf_map_args {
	477	struct bpf_offloaded_map *offmap;
	478	struct bpf_map_info *info;
	479	};
	480
	481	static struct ns_common bpf_map_offload_info_fill_ns(void private_data)
	482	{
	483	struct ns_get_path_bpf_map_args *args = private_data;
	484	struct ns_common *ns;
	485	struct net *net;
	486
	487	rtnl_lock();
	488	down_read(&bpf_devs_lock);
	489
	490	if (args->offmap->netdev) {
	491	args->info->ifindex = args->offmap->netdev->ifindex;
	492	net = dev_net(args->offmap->netdev);
	493	get_net(net);
	494	ns = &net->ns;
	495	} else {
	496	args->info->ifindex = 0;
	497	ns = NULL;
	498	}
	499
	500	up_read(&bpf_devs_lock);
	501	rtnl_unlock();
	502
	503	return ns;
	504	}
	505
	506	int bpf_map_offload_info_fill(struct bpf_map_info info, struct bpf_map map)
	507	{
	508	struct ns_get_path_bpf_map_args args = {
	509	.offmap = map_to_offmap(map),
	510	.info = info,
	511	};
	512	struct inode *ns_inode;
	513	struct path ns_path;
	514	void *res;
	515
	516	res = ns_get_path_cb(&ns_path, bpf_map_offload_info_fill_ns, &args);
	517	if (IS_ERR(res)) {
	518	if (!info->ifindex)
	519	return -ENODEV;
	520	return PTR_ERR(res);
	521	}
	522
	523	ns_inode = ns_path.dentry->d_inode;
	524	info->netns_dev = new_encode_dev(ns_inode->i_sb->s_dev);
	525	info->netns_ino = ns_inode->i_ino;
	526	path_put(&ns_path);
	527
	528	return 0;
	529	}
	530
fd4f227d JK	531	static bool __bpf_offload_dev_match(struct bpf_prog *prog,
fd4f227d JK	532	struct net_device *netdev)
a3884572	533	{
fd4f227d	534	struct bpf_offload_netdev ondev1, ondev2;
a3884572	535	struct bpf_prog_offload *offload;
a3884572	536
0cd3cbed	537	if (!bpf_prog_is_dev_bound(prog->aux))
a3884572	538	return false;
a3884572	539
a3884572	540	offload = prog->aux->offload;
fd4f227d JK	541	if (!offload)
	542	return false;
	543	if (offload->netdev == netdev)
	544	return true;
	545
	546	ondev1 = bpf_offload_find_netdev(offload->netdev);
	547	ondev2 = bpf_offload_find_netdev(netdev);
	548
	549	return ondev1 && ondev2 && ondev1->offdev == ondev2->offdev;
	550	}
	551
	552	bool bpf_offload_dev_match(struct bpf_prog prog, struct net_device netdev)
	553	{
	554	bool ret;
	555
	556	down_read(&bpf_devs_lock);
	557	ret = __bpf_offload_dev_match(prog, netdev);
	558	up_read(&bpf_devs_lock);
	559
	560	return ret;
	561	}
	562	EXPORT_SYMBOL_GPL(bpf_offload_dev_match);
	563
	564	bool bpf_offload_prog_map_match(struct bpf_prog prog, struct bpf_map map)
	565	{
	566	struct bpf_offloaded_map *offmap;
	567	bool ret;
	568
	569	if (!bpf_map_is_dev_bound(map))
	570	return bpf_map_offload_neutral(map);
a3884572 JK	571	offmap = map_to_offmap(map);
a3884572 JK	572
fd4f227d JK	573	down_read(&bpf_devs_lock);
fd4f227d JK	574	ret = __bpf_offload_dev_match(prog, offmap->netdev);
a3884572 JK	575	up_read(&bpf_devs_lock);
	576
	577	return ret;
	578	}
	579
602144c2 JK	580	int bpf_offload_dev_netdev_register(struct bpf_offload_dev *offdev,
602144c2 JK	581	struct net_device *netdev)
a3884572	582	{
9fd7c555 JK	583	struct bpf_offload_netdev *ondev;
9fd7c555 JK	584	int err;
a3884572	585
9fd7c555 JK	586	ondev = kzalloc(sizeof(*ondev), GFP_KERNEL);
	587	if (!ondev)
	588	return -ENOMEM;
	589
	590	ondev->netdev = netdev;
602144c2	591	ondev->offdev = offdev;
9fd7c555 JK	592	INIT_LIST_HEAD(&ondev->progs);
	593	INIT_LIST_HEAD(&ondev->maps);
	594
	595	down_write(&bpf_devs_lock);
	596	err = rhashtable_insert_fast(&offdevs, &ondev->l, offdevs_params);
	597	if (err) {
	598	netdev_warn(netdev, "failed to register for BPF offload\n");
	599	goto err_unlock_free;
	600	}
a3884572	601
602144c2	602	list_add(&ondev->offdev_netdevs, &offdev->netdevs);
9fd7c555 JK	603	up_write(&bpf_devs_lock);
	604	return 0;
	605
	606	err_unlock_free:
	607	up_write(&bpf_devs_lock);
	608	kfree(ondev);
	609	return err;
a3884572	610	}
9fd7c555	611	EXPORT_SYMBOL_GPL(bpf_offload_dev_netdev_register);
a3884572	612
602144c2 JK	613	void bpf_offload_dev_netdev_unregister(struct bpf_offload_dev *offdev,
602144c2 JK	614	struct net_device *netdev)
ab3f0063	615	{
602144c2	616	struct bpf_offload_netdev ondev, altdev;
9fd7c555 JK	617	struct bpf_offloaded_map offmap, mtmp;
9fd7c555 JK	618	struct bpf_prog_offload offload, ptmp;
ab3f0063 JK	619
	620	ASSERT_RTNL();
	621
9fd7c555 JK	622	down_write(&bpf_devs_lock);
	623	ondev = rhashtable_lookup_fast(&offdevs, &netdev, offdevs_params);
	624	if (WARN_ON(!ondev))
	625	goto unlock;
ab3f0063	626
9fd7c555	627	WARN_ON(rhashtable_remove_fast(&offdevs, &ondev->l, offdevs_params));
602144c2 JK	628	list_del(&ondev->offdev_netdevs);
	629
	630	/* Try to move the objects to another netdev of the device */
	631	altdev = list_first_entry_or_null(&offdev->netdevs,
	632	struct bpf_offload_netdev,
	633	offdev_netdevs);
	634	if (altdev) {
	635	list_for_each_entry(offload, &ondev->progs, offloads)
	636	offload->netdev = altdev->netdev;
	637	list_splice_init(&ondev->progs, &altdev->progs);
	638
	639	list_for_each_entry(offmap, &ondev->maps, offloads)
	640	offmap->netdev = altdev->netdev;
	641	list_splice_init(&ondev->maps, &altdev->maps);
	642	} else {
	643	list_for_each_entry_safe(offload, ptmp, &ondev->progs, offloads)
	644	__bpf_prog_offload_destroy(offload->prog);
	645	list_for_each_entry_safe(offmap, mtmp, &ondev->maps, offloads)
	646	__bpf_map_offload_destroy(offmap);
	647	}
ab3f0063	648
9fd7c555 JK	649	WARN_ON(!list_empty(&ondev->progs));
	650	WARN_ON(!list_empty(&ondev->maps));
	651	kfree(ondev);
	652	unlock:
	653	up_write(&bpf_devs_lock);
	654	}
	655	EXPORT_SYMBOL_GPL(bpf_offload_dev_netdev_unregister);
602144c2	656
1385d755 QM	657	struct bpf_offload_dev *
1385d755 QM	658	bpf_offload_dev_create(const struct bpf_prog_offload_ops *ops)
602144c2 JK	659	{
	660	struct bpf_offload_dev *offdev;
	661	int err;
	662
	663	down_write(&bpf_devs_lock);
	664	if (!offdevs_inited) {
	665	err = rhashtable_init(&offdevs, &offdevs_params);
	666	if (err)
	667	return ERR_PTR(err);
	668	offdevs_inited = true;
	669	}
	670	up_write(&bpf_devs_lock);
	671
	672	offdev = kzalloc(sizeof(*offdev), GFP_KERNEL);
	673	if (!offdev)
	674	return ERR_PTR(-ENOMEM);
	675
1385d755	676	offdev->ops = ops;
602144c2 JK	677	INIT_LIST_HEAD(&offdev->netdevs);
	678
	679	return offdev;
	680	}
	681	EXPORT_SYMBOL_GPL(bpf_offload_dev_create);
	682
	683	void bpf_offload_dev_destroy(struct bpf_offload_dev *offdev)
	684	{
	685	WARN_ON(!list_empty(&offdev->netdevs));
	686	kfree(offdev);
	687	}
	688	EXPORT_SYMBOL_GPL(bpf_offload_dev_destroy);