xref: /kernel/linux/linux-6.6/net/sched/act_ipt.c (revision 62306a36)
1// SPDX-License-Identifier: GPL-2.0-or-later
2/*
3 * net/sched/act_ipt.c		iptables target interface
4 *
5 *TODO: Add other tables. For now we only support the ipv4 table targets
6 *
7 * Copyright:	Jamal Hadi Salim (2002-13)
8 */
9
10#include <linux/types.h>
11#include <linux/kernel.h>
12#include <linux/string.h>
13#include <linux/errno.h>
14#include <linux/skbuff.h>
15#include <linux/rtnetlink.h>
16#include <linux/module.h>
17#include <linux/init.h>
18#include <linux/slab.h>
19#include <net/netlink.h>
20#include <net/pkt_sched.h>
21#include <linux/tc_act/tc_ipt.h>
22#include <net/tc_act/tc_ipt.h>
23#include <net/tc_wrapper.h>
24#include <net/ip.h>
25
26#include <linux/netfilter_ipv4/ip_tables.h>
27
28
29static struct tc_action_ops act_ipt_ops;
30static struct tc_action_ops act_xt_ops;
31
32static int ipt_init_target(struct net *net, struct xt_entry_target *t,
33			   char *table, unsigned int hook)
34{
35	struct xt_tgchk_param par;
36	struct xt_target *target;
37	struct ipt_entry e = {};
38	int ret = 0;
39
40	target = xt_request_find_target(AF_INET, t->u.user.name,
41					t->u.user.revision);
42	if (IS_ERR(target))
43		return PTR_ERR(target);
44
45	t->u.kernel.target = target;
46	memset(&par, 0, sizeof(par));
47	par.net       = net;
48	par.table     = table;
49	par.entryinfo = &e;
50	par.target    = target;
51	par.targinfo  = t->data;
52	par.hook_mask = 1 << hook;
53	par.family    = NFPROTO_IPV4;
54
55	ret = xt_check_target(&par, t->u.target_size - sizeof(*t), 0, false);
56	if (ret < 0) {
57		module_put(t->u.kernel.target->me);
58		return ret;
59	}
60	return 0;
61}
62
63static void ipt_destroy_target(struct xt_entry_target *t, struct net *net)
64{
65	struct xt_tgdtor_param par = {
66		.target   = t->u.kernel.target,
67		.targinfo = t->data,
68		.family   = NFPROTO_IPV4,
69		.net      = net,
70	};
71	if (par.target->destroy != NULL)
72		par.target->destroy(&par);
73	module_put(par.target->me);
74}
75
76static void tcf_ipt_release(struct tc_action *a)
77{
78	struct tcf_ipt *ipt = to_ipt(a);
79
80	if (ipt->tcfi_t) {
81		ipt_destroy_target(ipt->tcfi_t, a->idrinfo->net);
82		kfree(ipt->tcfi_t);
83	}
84	kfree(ipt->tcfi_tname);
85}
86
87static const struct nla_policy ipt_policy[TCA_IPT_MAX + 1] = {
88	[TCA_IPT_TABLE]	= { .type = NLA_STRING, .len = IFNAMSIZ },
89	[TCA_IPT_HOOK]	= NLA_POLICY_RANGE(NLA_U32, NF_INET_PRE_ROUTING,
90					   NF_INET_NUMHOOKS),
91	[TCA_IPT_INDEX]	= { .type = NLA_U32 },
92	[TCA_IPT_TARG]	= { .len = sizeof(struct xt_entry_target) },
93};
94
95static int __tcf_ipt_init(struct net *net, unsigned int id, struct nlattr *nla,
96			  struct nlattr *est, struct tc_action **a,
97			  const struct tc_action_ops *ops,
98			  struct tcf_proto *tp, u32 flags)
99{
100	struct tc_action_net *tn = net_generic(net, id);
101	bool bind = flags & TCA_ACT_FLAGS_BIND;
102	struct nlattr *tb[TCA_IPT_MAX + 1];
103	struct tcf_ipt *ipt;
104	struct xt_entry_target *td, *t;
105	char *tname;
106	bool exists = false;
107	int ret = 0, err;
108	u32 hook = 0;
109	u32 index = 0;
110
111	if (nla == NULL)
112		return -EINVAL;
113
114	err = nla_parse_nested_deprecated(tb, TCA_IPT_MAX, nla, ipt_policy,
115					  NULL);
116	if (err < 0)
117		return err;
118
119	if (tb[TCA_IPT_INDEX] != NULL)
120		index = nla_get_u32(tb[TCA_IPT_INDEX]);
121
122	err = tcf_idr_check_alloc(tn, &index, a, bind);
123	if (err < 0)
124		return err;
125	exists = err;
126	if (exists && bind)
127		return 0;
128
129	if (tb[TCA_IPT_HOOK] == NULL || tb[TCA_IPT_TARG] == NULL) {
130		if (exists)
131			tcf_idr_release(*a, bind);
132		else
133			tcf_idr_cleanup(tn, index);
134		return -EINVAL;
135	}
136
137	td = (struct xt_entry_target *)nla_data(tb[TCA_IPT_TARG]);
138	if (nla_len(tb[TCA_IPT_TARG]) != td->u.target_size) {
139		if (exists)
140			tcf_idr_release(*a, bind);
141		else
142			tcf_idr_cleanup(tn, index);
143		return -EINVAL;
144	}
145
146	if (!exists) {
147		ret = tcf_idr_create(tn, index, est, a, ops, bind,
148				     false, flags);
149		if (ret) {
150			tcf_idr_cleanup(tn, index);
151			return ret;
152		}
153		ret = ACT_P_CREATED;
154	} else {
155		if (bind)/* dont override defaults */
156			return 0;
157
158		if (!(flags & TCA_ACT_FLAGS_REPLACE)) {
159			tcf_idr_release(*a, bind);
160			return -EEXIST;
161		}
162	}
163
164	err = -EINVAL;
165	hook = nla_get_u32(tb[TCA_IPT_HOOK]);
166	switch (hook) {
167	case NF_INET_PRE_ROUTING:
168		break;
169	case NF_INET_POST_ROUTING:
170		break;
171	default:
172		goto err1;
173	}
174
175	if (tb[TCA_IPT_TABLE]) {
176		/* mangle only for now */
177		if (nla_strcmp(tb[TCA_IPT_TABLE], "mangle"))
178			goto err1;
179	}
180
181	tname = kstrdup("mangle", GFP_KERNEL);
182	if (unlikely(!tname))
183		goto err1;
184
185	t = kmemdup(td, td->u.target_size, GFP_KERNEL);
186	if (unlikely(!t))
187		goto err2;
188
189	err = ipt_init_target(net, t, tname, hook);
190	if (err < 0)
191		goto err3;
192
193	ipt = to_ipt(*a);
194
195	spin_lock_bh(&ipt->tcf_lock);
196	if (ret != ACT_P_CREATED) {
197		ipt_destroy_target(ipt->tcfi_t, net);
198		kfree(ipt->tcfi_tname);
199		kfree(ipt->tcfi_t);
200	}
201	ipt->tcfi_tname = tname;
202	ipt->tcfi_t     = t;
203	ipt->tcfi_hook  = hook;
204	spin_unlock_bh(&ipt->tcf_lock);
205	return ret;
206
207err3:
208	kfree(t);
209err2:
210	kfree(tname);
211err1:
212	tcf_idr_release(*a, bind);
213	return err;
214}
215
216static int tcf_ipt_init(struct net *net, struct nlattr *nla,
217			struct nlattr *est, struct tc_action **a,
218			struct tcf_proto *tp,
219			u32 flags, struct netlink_ext_ack *extack)
220{
221	return __tcf_ipt_init(net, act_ipt_ops.net_id, nla, est,
222			      a, &act_ipt_ops, tp, flags);
223}
224
225static int tcf_xt_init(struct net *net, struct nlattr *nla,
226		       struct nlattr *est, struct tc_action **a,
227		       struct tcf_proto *tp,
228		       u32 flags, struct netlink_ext_ack *extack)
229{
230	return __tcf_ipt_init(net, act_xt_ops.net_id, nla, est,
231			      a, &act_xt_ops, tp, flags);
232}
233
234static bool tcf_ipt_act_check(struct sk_buff *skb)
235{
236	const struct iphdr *iph;
237	unsigned int nhoff, len;
238
239	if (!pskb_may_pull(skb, sizeof(struct iphdr)))
240		return false;
241
242	nhoff = skb_network_offset(skb);
243	iph = ip_hdr(skb);
244	if (iph->ihl < 5 || iph->version != 4)
245		return false;
246
247	len = skb_ip_totlen(skb);
248	if (skb->len < nhoff + len || len < (iph->ihl * 4u))
249		return false;
250
251	return pskb_may_pull(skb, iph->ihl * 4u);
252}
253
254TC_INDIRECT_SCOPE int tcf_ipt_act(struct sk_buff *skb,
255				  const struct tc_action *a,
256				  struct tcf_result *res)
257{
258	char saved_cb[sizeof_field(struct sk_buff, cb)];
259	int ret = 0, result = 0;
260	struct tcf_ipt *ipt = to_ipt(a);
261	struct xt_action_param par;
262	struct nf_hook_state state = {
263		.net	= dev_net(skb->dev),
264		.in	= skb->dev,
265		.hook	= ipt->tcfi_hook,
266		.pf	= NFPROTO_IPV4,
267	};
268
269	if (skb_protocol(skb, false) != htons(ETH_P_IP))
270		return TC_ACT_UNSPEC;
271
272	if (skb_unclone(skb, GFP_ATOMIC))
273		return TC_ACT_UNSPEC;
274
275	if (!tcf_ipt_act_check(skb))
276		return TC_ACT_UNSPEC;
277
278	if (state.hook == NF_INET_POST_ROUTING) {
279		if (!skb_dst(skb))
280			return TC_ACT_UNSPEC;
281
282		state.out = skb->dev;
283	}
284
285	memcpy(saved_cb, skb->cb, sizeof(saved_cb));
286
287	spin_lock(&ipt->tcf_lock);
288
289	tcf_lastuse_update(&ipt->tcf_tm);
290	bstats_update(&ipt->tcf_bstats, skb);
291
292	/* yes, we have to worry about both in and out dev
293	 * worry later - danger - this API seems to have changed
294	 * from earlier kernels
295	 */
296	par.state    = &state;
297	par.target   = ipt->tcfi_t->u.kernel.target;
298	par.targinfo = ipt->tcfi_t->data;
299
300	memset(IPCB(skb), 0, sizeof(struct inet_skb_parm));
301
302	ret = par.target->target(skb, &par);
303
304	switch (ret) {
305	case NF_ACCEPT:
306		result = TC_ACT_OK;
307		break;
308	case NF_DROP:
309		result = TC_ACT_SHOT;
310		ipt->tcf_qstats.drops++;
311		break;
312	case XT_CONTINUE:
313		result = TC_ACT_PIPE;
314		break;
315	default:
316		net_notice_ratelimited("tc filter: Bogus netfilter code %d assume ACCEPT\n",
317				       ret);
318		result = TC_ACT_OK;
319		break;
320	}
321	spin_unlock(&ipt->tcf_lock);
322
323	memcpy(skb->cb, saved_cb, sizeof(skb->cb));
324
325	return result;
326
327}
328
329static int tcf_ipt_dump(struct sk_buff *skb, struct tc_action *a, int bind,
330			int ref)
331{
332	unsigned char *b = skb_tail_pointer(skb);
333	struct tcf_ipt *ipt = to_ipt(a);
334	struct xt_entry_target *t;
335	struct tcf_t tm;
336	struct tc_cnt c;
337
338	/* for simple targets kernel size == user size
339	 * user name = target name
340	 * for foolproof you need to not assume this
341	 */
342
343	spin_lock_bh(&ipt->tcf_lock);
344	t = kmemdup(ipt->tcfi_t, ipt->tcfi_t->u.user.target_size, GFP_ATOMIC);
345	if (unlikely(!t))
346		goto nla_put_failure;
347
348	c.bindcnt = atomic_read(&ipt->tcf_bindcnt) - bind;
349	c.refcnt = refcount_read(&ipt->tcf_refcnt) - ref;
350	strcpy(t->u.user.name, ipt->tcfi_t->u.kernel.target->name);
351
352	if (nla_put(skb, TCA_IPT_TARG, ipt->tcfi_t->u.user.target_size, t) ||
353	    nla_put_u32(skb, TCA_IPT_INDEX, ipt->tcf_index) ||
354	    nla_put_u32(skb, TCA_IPT_HOOK, ipt->tcfi_hook) ||
355	    nla_put(skb, TCA_IPT_CNT, sizeof(struct tc_cnt), &c) ||
356	    nla_put_string(skb, TCA_IPT_TABLE, ipt->tcfi_tname))
357		goto nla_put_failure;
358
359	tcf_tm_dump(&tm, &ipt->tcf_tm);
360	if (nla_put_64bit(skb, TCA_IPT_TM, sizeof(tm), &tm, TCA_IPT_PAD))
361		goto nla_put_failure;
362
363	spin_unlock_bh(&ipt->tcf_lock);
364	kfree(t);
365	return skb->len;
366
367nla_put_failure:
368	spin_unlock_bh(&ipt->tcf_lock);
369	nlmsg_trim(skb, b);
370	kfree(t);
371	return -1;
372}
373
374static struct tc_action_ops act_ipt_ops = {
375	.kind		=	"ipt",
376	.id		=	TCA_ID_IPT,
377	.owner		=	THIS_MODULE,
378	.act		=	tcf_ipt_act,
379	.dump		=	tcf_ipt_dump,
380	.cleanup	=	tcf_ipt_release,
381	.init		=	tcf_ipt_init,
382	.size		=	sizeof(struct tcf_ipt),
383};
384
385static __net_init int ipt_init_net(struct net *net)
386{
387	struct tc_action_net *tn = net_generic(net, act_ipt_ops.net_id);
388
389	return tc_action_net_init(net, tn, &act_ipt_ops);
390}
391
392static void __net_exit ipt_exit_net(struct list_head *net_list)
393{
394	tc_action_net_exit(net_list, act_ipt_ops.net_id);
395}
396
397static struct pernet_operations ipt_net_ops = {
398	.init = ipt_init_net,
399	.exit_batch = ipt_exit_net,
400	.id   = &act_ipt_ops.net_id,
401	.size = sizeof(struct tc_action_net),
402};
403
404static struct tc_action_ops act_xt_ops = {
405	.kind		=	"xt",
406	.id		=	TCA_ID_XT,
407	.owner		=	THIS_MODULE,
408	.act		=	tcf_ipt_act,
409	.dump		=	tcf_ipt_dump,
410	.cleanup	=	tcf_ipt_release,
411	.init		=	tcf_xt_init,
412	.size		=	sizeof(struct tcf_ipt),
413};
414
415static __net_init int xt_init_net(struct net *net)
416{
417	struct tc_action_net *tn = net_generic(net, act_xt_ops.net_id);
418
419	return tc_action_net_init(net, tn, &act_xt_ops);
420}
421
422static void __net_exit xt_exit_net(struct list_head *net_list)
423{
424	tc_action_net_exit(net_list, act_xt_ops.net_id);
425}
426
427static struct pernet_operations xt_net_ops = {
428	.init = xt_init_net,
429	.exit_batch = xt_exit_net,
430	.id   = &act_xt_ops.net_id,
431	.size = sizeof(struct tc_action_net),
432};
433
434MODULE_AUTHOR("Jamal Hadi Salim(2002-13)");
435MODULE_DESCRIPTION("Iptables target actions");
436MODULE_LICENSE("GPL");
437MODULE_ALIAS("act_xt");
438
439static int __init ipt_init_module(void)
440{
441	int ret1, ret2;
442
443	ret1 = tcf_register_action(&act_xt_ops, &xt_net_ops);
444	if (ret1 < 0)
445		pr_err("Failed to load xt action\n");
446
447	ret2 = tcf_register_action(&act_ipt_ops, &ipt_net_ops);
448	if (ret2 < 0)
449		pr_err("Failed to load ipt action\n");
450
451	if (ret1 < 0 && ret2 < 0) {
452		return ret1;
453	} else
454		return 0;
455}
456
457static void __exit ipt_cleanup_module(void)
458{
459	tcf_unregister_action(&act_ipt_ops, &ipt_net_ops);
460	tcf_unregister_action(&act_xt_ops, &xt_net_ops);
461}
462
463module_init(ipt_init_module);
464module_exit(ipt_cleanup_module);
465