2 * Copyright (c) 2008, Intel Corporation.
4 * This program is free software; you can redistribute it and/or modify it
5 * under the terms and conditions of the GNU General Public License,
6 * version 2, as published by the Free Software Foundation.
8 * This program is distributed in the hope it will be useful, but WITHOUT
9 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
10 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
13 * You should have received a copy of the GNU General Public License along with
14 * this program; if not, see <http://www.gnu.org/licenses/>.
16 * Author: Alexander Duyck <alexander.h.duyck@intel.com>
19 #include <linux/module.h>
20 #include <linux/init.h>
21 #include <linux/kernel.h>
22 #include <linux/skbuff.h>
23 #include <linux/rtnetlink.h>
24 #include <net/netlink.h>
25 #include <net/pkt_sched.h>
28 #include <net/dsfield.h>
30 #include <linux/tc_act/tc_skbedit.h>
31 #include <net/tc_act/tc_skbedit.h>
33 static unsigned int skbedit_net_id
;
34 static struct tc_action_ops act_skbedit_ops
;
36 static int tcf_skbedit_act(struct sk_buff
*skb
, const struct tc_action
*a
,
37 struct tcf_result
*res
)
39 struct tcf_skbedit
*d
= to_skbedit(a
);
40 struct tcf_skbedit_params
*params
;
43 tcf_lastuse_update(&d
->tcf_tm
);
44 bstats_cpu_update(this_cpu_ptr(d
->common
.cpu_bstats
), skb
);
46 params
= rcu_dereference_bh(d
->params
);
47 action
= READ_ONCE(d
->tcf_action
);
49 if (params
->flags
& SKBEDIT_F_PRIORITY
)
50 skb
->priority
= params
->priority
;
51 if (params
->flags
& SKBEDIT_F_INHERITDSFIELD
) {
52 int wlen
= skb_network_offset(skb
);
54 switch (tc_skb_protocol(skb
)) {
56 wlen
+= sizeof(struct iphdr
);
57 if (!pskb_may_pull(skb
, wlen
))
59 skb
->priority
= ipv4_get_dsfield(ip_hdr(skb
)) >> 2;
62 case htons(ETH_P_IPV6
):
63 wlen
+= sizeof(struct ipv6hdr
);
64 if (!pskb_may_pull(skb
, wlen
))
66 skb
->priority
= ipv6_get_dsfield(ipv6_hdr(skb
)) >> 2;
70 if (params
->flags
& SKBEDIT_F_QUEUE_MAPPING
&&
71 skb
->dev
->real_num_tx_queues
> params
->queue_mapping
)
72 skb_set_queue_mapping(skb
, params
->queue_mapping
);
73 if (params
->flags
& SKBEDIT_F_MARK
) {
74 skb
->mark
&= ~params
->mask
;
75 skb
->mark
|= params
->mark
& params
->mask
;
77 if (params
->flags
& SKBEDIT_F_PTYPE
)
78 skb
->pkt_type
= params
->ptype
;
82 qstats_drop_inc(this_cpu_ptr(d
->common
.cpu_qstats
));
86 static const struct nla_policy skbedit_policy
[TCA_SKBEDIT_MAX
+ 1] = {
87 [TCA_SKBEDIT_PARMS
] = { .len
= sizeof(struct tc_skbedit
) },
88 [TCA_SKBEDIT_PRIORITY
] = { .len
= sizeof(u32
) },
89 [TCA_SKBEDIT_QUEUE_MAPPING
] = { .len
= sizeof(u16
) },
90 [TCA_SKBEDIT_MARK
] = { .len
= sizeof(u32
) },
91 [TCA_SKBEDIT_PTYPE
] = { .len
= sizeof(u16
) },
92 [TCA_SKBEDIT_MASK
] = { .len
= sizeof(u32
) },
93 [TCA_SKBEDIT_FLAGS
] = { .len
= sizeof(u64
) },
96 static int tcf_skbedit_init(struct net
*net
, struct nlattr
*nla
,
97 struct nlattr
*est
, struct tc_action
**a
,
98 int ovr
, int bind
, bool rtnl_held
,
99 struct netlink_ext_ack
*extack
)
101 struct tc_action_net
*tn
= net_generic(net
, skbedit_net_id
);
102 struct tcf_skbedit_params
*params_new
;
103 struct nlattr
*tb
[TCA_SKBEDIT_MAX
+ 1];
104 struct tc_skbedit
*parm
;
105 struct tcf_skbedit
*d
;
106 u32 flags
= 0, *priority
= NULL
, *mark
= NULL
, *mask
= NULL
;
107 u16
*queue_mapping
= NULL
, *ptype
= NULL
;
114 err
= nla_parse_nested(tb
, TCA_SKBEDIT_MAX
, nla
, skbedit_policy
, NULL
);
118 if (tb
[TCA_SKBEDIT_PARMS
] == NULL
)
121 if (tb
[TCA_SKBEDIT_PRIORITY
] != NULL
) {
122 flags
|= SKBEDIT_F_PRIORITY
;
123 priority
= nla_data(tb
[TCA_SKBEDIT_PRIORITY
]);
126 if (tb
[TCA_SKBEDIT_QUEUE_MAPPING
] != NULL
) {
127 flags
|= SKBEDIT_F_QUEUE_MAPPING
;
128 queue_mapping
= nla_data(tb
[TCA_SKBEDIT_QUEUE_MAPPING
]);
131 if (tb
[TCA_SKBEDIT_PTYPE
] != NULL
) {
132 ptype
= nla_data(tb
[TCA_SKBEDIT_PTYPE
]);
133 if (!skb_pkt_type_ok(*ptype
))
135 flags
|= SKBEDIT_F_PTYPE
;
138 if (tb
[TCA_SKBEDIT_MARK
] != NULL
) {
139 flags
|= SKBEDIT_F_MARK
;
140 mark
= nla_data(tb
[TCA_SKBEDIT_MARK
]);
143 if (tb
[TCA_SKBEDIT_MASK
] != NULL
) {
144 flags
|= SKBEDIT_F_MASK
;
145 mask
= nla_data(tb
[TCA_SKBEDIT_MASK
]);
148 if (tb
[TCA_SKBEDIT_FLAGS
] != NULL
) {
149 u64
*pure_flags
= nla_data(tb
[TCA_SKBEDIT_FLAGS
]);
151 if (*pure_flags
& SKBEDIT_F_INHERITDSFIELD
)
152 flags
|= SKBEDIT_F_INHERITDSFIELD
;
155 parm
= nla_data(tb
[TCA_SKBEDIT_PARMS
]);
157 err
= tcf_idr_check_alloc(tn
, &parm
->index
, a
, bind
);
166 tcf_idr_release(*a
, bind
);
168 tcf_idr_cleanup(tn
, parm
->index
);
173 ret
= tcf_idr_create(tn
, parm
->index
, est
, a
,
174 &act_skbedit_ops
, bind
, true);
176 tcf_idr_cleanup(tn
, parm
->index
);
185 tcf_idr_release(*a
, bind
);
190 params_new
= kzalloc(sizeof(*params_new
), GFP_KERNEL
);
191 if (unlikely(!params_new
)) {
192 if (ret
== ACT_P_CREATED
)
193 tcf_idr_release(*a
, bind
);
197 params_new
->flags
= flags
;
198 if (flags
& SKBEDIT_F_PRIORITY
)
199 params_new
->priority
= *priority
;
200 if (flags
& SKBEDIT_F_QUEUE_MAPPING
)
201 params_new
->queue_mapping
= *queue_mapping
;
202 if (flags
& SKBEDIT_F_MARK
)
203 params_new
->mark
= *mark
;
204 if (flags
& SKBEDIT_F_PTYPE
)
205 params_new
->ptype
= *ptype
;
206 /* default behaviour is to use all the bits */
207 params_new
->mask
= 0xffffffff;
208 if (flags
& SKBEDIT_F_MASK
)
209 params_new
->mask
= *mask
;
211 spin_lock_bh(&d
->tcf_lock
);
212 d
->tcf_action
= parm
->action
;
213 rcu_swap_protected(d
->params
, params_new
,
214 lockdep_is_held(&d
->tcf_lock
));
215 spin_unlock_bh(&d
->tcf_lock
);
217 kfree_rcu(params_new
, rcu
);
219 if (ret
== ACT_P_CREATED
)
220 tcf_idr_insert(tn
, *a
);
224 static int tcf_skbedit_dump(struct sk_buff
*skb
, struct tc_action
*a
,
227 unsigned char *b
= skb_tail_pointer(skb
);
228 struct tcf_skbedit
*d
= to_skbedit(a
);
229 struct tcf_skbedit_params
*params
;
230 struct tc_skbedit opt
= {
231 .index
= d
->tcf_index
,
232 .refcnt
= refcount_read(&d
->tcf_refcnt
) - ref
,
233 .bindcnt
= atomic_read(&d
->tcf_bindcnt
) - bind
,
238 spin_lock_bh(&d
->tcf_lock
);
239 params
= rcu_dereference_protected(d
->params
,
240 lockdep_is_held(&d
->tcf_lock
));
241 opt
.action
= d
->tcf_action
;
243 if (nla_put(skb
, TCA_SKBEDIT_PARMS
, sizeof(opt
), &opt
))
244 goto nla_put_failure
;
245 if ((params
->flags
& SKBEDIT_F_PRIORITY
) &&
246 nla_put_u32(skb
, TCA_SKBEDIT_PRIORITY
, params
->priority
))
247 goto nla_put_failure
;
248 if ((params
->flags
& SKBEDIT_F_QUEUE_MAPPING
) &&
249 nla_put_u16(skb
, TCA_SKBEDIT_QUEUE_MAPPING
, params
->queue_mapping
))
250 goto nla_put_failure
;
251 if ((params
->flags
& SKBEDIT_F_MARK
) &&
252 nla_put_u32(skb
, TCA_SKBEDIT_MARK
, params
->mark
))
253 goto nla_put_failure
;
254 if ((params
->flags
& SKBEDIT_F_PTYPE
) &&
255 nla_put_u16(skb
, TCA_SKBEDIT_PTYPE
, params
->ptype
))
256 goto nla_put_failure
;
257 if ((params
->flags
& SKBEDIT_F_MASK
) &&
258 nla_put_u32(skb
, TCA_SKBEDIT_MASK
, params
->mask
))
259 goto nla_put_failure
;
260 if (params
->flags
& SKBEDIT_F_INHERITDSFIELD
)
261 pure_flags
|= SKBEDIT_F_INHERITDSFIELD
;
262 if (pure_flags
!= 0 &&
263 nla_put(skb
, TCA_SKBEDIT_FLAGS
, sizeof(pure_flags
), &pure_flags
))
264 goto nla_put_failure
;
266 tcf_tm_dump(&t
, &d
->tcf_tm
);
267 if (nla_put_64bit(skb
, TCA_SKBEDIT_TM
, sizeof(t
), &t
, TCA_SKBEDIT_PAD
))
268 goto nla_put_failure
;
269 spin_unlock_bh(&d
->tcf_lock
);
274 spin_unlock_bh(&d
->tcf_lock
);
279 static void tcf_skbedit_cleanup(struct tc_action
*a
)
281 struct tcf_skbedit
*d
= to_skbedit(a
);
282 struct tcf_skbedit_params
*params
;
284 params
= rcu_dereference_protected(d
->params
, 1);
286 kfree_rcu(params
, rcu
);
289 static int tcf_skbedit_walker(struct net
*net
, struct sk_buff
*skb
,
290 struct netlink_callback
*cb
, int type
,
291 const struct tc_action_ops
*ops
,
292 struct netlink_ext_ack
*extack
)
294 struct tc_action_net
*tn
= net_generic(net
, skbedit_net_id
);
296 return tcf_generic_walker(tn
, skb
, cb
, type
, ops
, extack
);
299 static int tcf_skbedit_search(struct net
*net
, struct tc_action
**a
, u32 index
)
301 struct tc_action_net
*tn
= net_generic(net
, skbedit_net_id
);
303 return tcf_idr_search(tn
, a
, index
);
306 static struct tc_action_ops act_skbedit_ops
= {
308 .type
= TCA_ACT_SKBEDIT
,
309 .owner
= THIS_MODULE
,
310 .act
= tcf_skbedit_act
,
311 .dump
= tcf_skbedit_dump
,
312 .init
= tcf_skbedit_init
,
313 .cleanup
= tcf_skbedit_cleanup
,
314 .walk
= tcf_skbedit_walker
,
315 .lookup
= tcf_skbedit_search
,
316 .size
= sizeof(struct tcf_skbedit
),
319 static __net_init
int skbedit_init_net(struct net
*net
)
321 struct tc_action_net
*tn
= net_generic(net
, skbedit_net_id
);
323 return tc_action_net_init(tn
, &act_skbedit_ops
);
326 static void __net_exit
skbedit_exit_net(struct list_head
*net_list
)
328 tc_action_net_exit(net_list
, skbedit_net_id
);
331 static struct pernet_operations skbedit_net_ops
= {
332 .init
= skbedit_init_net
,
333 .exit_batch
= skbedit_exit_net
,
334 .id
= &skbedit_net_id
,
335 .size
= sizeof(struct tc_action_net
),
338 MODULE_AUTHOR("Alexander Duyck, <alexander.h.duyck@intel.com>");
339 MODULE_DESCRIPTION("SKB Editing");
340 MODULE_LICENSE("GPL");
342 static int __init
skbedit_init_module(void)
344 return tcf_register_action(&act_skbedit_ops
, &skbedit_net_ops
);
347 static void __exit
skbedit_cleanup_module(void)
349 tcf_unregister_action(&act_skbedit_ops
, &skbedit_net_ops
);
352 module_init(skbedit_init_module
);
353 module_exit(skbedit_cleanup_module
);