2 * Codel - The Controlled-Delay Active Queue Management algorithm
4 * Copyright (C) 2011-2012 Kathleen Nichols <nichols@pollere.com>
5 * Copyright (C) 2011-2012 Van Jacobson <van@pollere.net>
7 * Implemented on linux by :
8 * Copyright (C) 2012 Michael D. Taht <dave.taht@bufferbloat.net>
9 * Copyright (C) 2012,2015 Eric Dumazet <edumazet@google.com>
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions, and the following disclaimer,
16 * without modification.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 * 3. The names of the authors may not be used to endorse or promote products
21 * derived from this software without specific prior written permission.
23 * Alternatively, provided that this notice is retained in full, this
24 * software may be distributed under the terms of the GNU General
25 * Public License ("GPL") version 2, in which case the provisions of the
26 * GPL apply INSTEAD OF those given above.
28 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
29 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
30 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
31 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
32 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
33 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
34 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
35 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
36 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
37 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
38 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
43 #include <linux/module.h>
44 #include <linux/slab.h>
45 #include <linux/types.h>
46 #include <linux/kernel.h>
47 #include <linux/errno.h>
48 #include <linux/skbuff.h>
49 #include <linux/prefetch.h>
50 #include <net/pkt_sched.h>
51 #include <net/codel.h>
54 #define DEFAULT_CODEL_LIMIT 1000
56 struct codel_sched_data
{
57 struct codel_params params
;
58 struct codel_vars vars
;
59 struct codel_stats stats
;
63 /* This is the specific function called from codel_dequeue()
64 * to dequeue a packet from queue. Note: backlog is handled in
65 * codel, we dont need to reduce it here.
67 static struct sk_buff
*dequeue_func(struct codel_vars
*vars
, void *ctx
)
69 struct Qdisc
*sch
= ctx
;
70 struct sk_buff
*skb
= __skb_dequeue(&sch
->q
);
73 sch
->qstats
.backlog
-= qdisc_pkt_len(skb
);
75 prefetch(&skb
->end
); /* we'll need skb_shinfo() */
79 static void drop_func(struct sk_buff
*skb
, void *ctx
)
81 struct Qdisc
*sch
= ctx
;
86 static struct sk_buff
*codel_qdisc_dequeue(struct Qdisc
*sch
)
88 struct codel_sched_data
*q
= qdisc_priv(sch
);
91 skb
= codel_dequeue(sch
, &sch
->qstats
.backlog
, &q
->params
, &q
->vars
,
92 &q
->stats
, qdisc_pkt_len
, codel_get_enqueue_time
,
93 drop_func
, dequeue_func
);
95 /* We cant call qdisc_tree_reduce_backlog() if our qlen is 0,
96 * or HTB crashes. Defer it for next round.
98 if (q
->stats
.drop_count
&& sch
->q
.qlen
) {
99 qdisc_tree_reduce_backlog(sch
, q
->stats
.drop_count
, q
->stats
.drop_len
);
100 q
->stats
.drop_count
= 0;
101 q
->stats
.drop_len
= 0;
104 qdisc_bstats_update(sch
, skb
);
108 static int codel_qdisc_enqueue(struct sk_buff
*skb
, struct Qdisc
*sch
)
110 struct codel_sched_data
*q
;
112 if (likely(qdisc_qlen(sch
) < sch
->limit
)) {
113 codel_set_enqueue_time(skb
);
114 return qdisc_enqueue_tail(skb
, sch
);
118 return qdisc_drop(skb
, sch
);
121 static const struct nla_policy codel_policy
[TCA_CODEL_MAX
+ 1] = {
122 [TCA_CODEL_TARGET
] = { .type
= NLA_U32
},
123 [TCA_CODEL_LIMIT
] = { .type
= NLA_U32
},
124 [TCA_CODEL_INTERVAL
] = { .type
= NLA_U32
},
125 [TCA_CODEL_ECN
] = { .type
= NLA_U32
},
126 [TCA_CODEL_CE_THRESHOLD
]= { .type
= NLA_U32
},
129 static int codel_change(struct Qdisc
*sch
, struct nlattr
*opt
)
131 struct codel_sched_data
*q
= qdisc_priv(sch
);
132 struct nlattr
*tb
[TCA_CODEL_MAX
+ 1];
133 unsigned int qlen
, dropped
= 0;
139 err
= nla_parse_nested(tb
, TCA_CODEL_MAX
, opt
, codel_policy
);
145 if (tb
[TCA_CODEL_TARGET
]) {
146 u32 target
= nla_get_u32(tb
[TCA_CODEL_TARGET
]);
148 q
->params
.target
= ((u64
)target
* NSEC_PER_USEC
) >> CODEL_SHIFT
;
151 if (tb
[TCA_CODEL_CE_THRESHOLD
]) {
152 u64 val
= nla_get_u32(tb
[TCA_CODEL_CE_THRESHOLD
]);
154 q
->params
.ce_threshold
= (val
* NSEC_PER_USEC
) >> CODEL_SHIFT
;
157 if (tb
[TCA_CODEL_INTERVAL
]) {
158 u32 interval
= nla_get_u32(tb
[TCA_CODEL_INTERVAL
]);
160 q
->params
.interval
= ((u64
)interval
* NSEC_PER_USEC
) >> CODEL_SHIFT
;
163 if (tb
[TCA_CODEL_LIMIT
])
164 sch
->limit
= nla_get_u32(tb
[TCA_CODEL_LIMIT
]);
166 if (tb
[TCA_CODEL_ECN
])
167 q
->params
.ecn
= !!nla_get_u32(tb
[TCA_CODEL_ECN
]);
170 while (sch
->q
.qlen
> sch
->limit
) {
171 struct sk_buff
*skb
= __skb_dequeue(&sch
->q
);
173 dropped
+= qdisc_pkt_len(skb
);
174 qdisc_qstats_backlog_dec(sch
, skb
);
175 qdisc_drop(skb
, sch
);
177 qdisc_tree_reduce_backlog(sch
, qlen
- sch
->q
.qlen
, dropped
);
179 sch_tree_unlock(sch
);
183 static int codel_init(struct Qdisc
*sch
, struct nlattr
*opt
)
185 struct codel_sched_data
*q
= qdisc_priv(sch
);
187 sch
->limit
= DEFAULT_CODEL_LIMIT
;
189 codel_params_init(&q
->params
);
190 codel_vars_init(&q
->vars
);
191 codel_stats_init(&q
->stats
);
192 q
->params
.mtu
= psched_mtu(qdisc_dev(sch
));
195 int err
= codel_change(sch
, opt
);
202 sch
->flags
|= TCQ_F_CAN_BYPASS
;
204 sch
->flags
&= ~TCQ_F_CAN_BYPASS
;
209 static int codel_dump(struct Qdisc
*sch
, struct sk_buff
*skb
)
211 struct codel_sched_data
*q
= qdisc_priv(sch
);
214 opts
= nla_nest_start(skb
, TCA_OPTIONS
);
216 goto nla_put_failure
;
218 if (nla_put_u32(skb
, TCA_CODEL_TARGET
,
219 codel_time_to_us(q
->params
.target
)) ||
220 nla_put_u32(skb
, TCA_CODEL_LIMIT
,
222 nla_put_u32(skb
, TCA_CODEL_INTERVAL
,
223 codel_time_to_us(q
->params
.interval
)) ||
224 nla_put_u32(skb
, TCA_CODEL_ECN
,
226 goto nla_put_failure
;
227 if (q
->params
.ce_threshold
!= CODEL_DISABLED_THRESHOLD
&&
228 nla_put_u32(skb
, TCA_CODEL_CE_THRESHOLD
,
229 codel_time_to_us(q
->params
.ce_threshold
)))
230 goto nla_put_failure
;
231 return nla_nest_end(skb
, opts
);
234 nla_nest_cancel(skb
, opts
);
238 static int codel_dump_stats(struct Qdisc
*sch
, struct gnet_dump
*d
)
240 const struct codel_sched_data
*q
= qdisc_priv(sch
);
241 struct tc_codel_xstats st
= {
242 .maxpacket
= q
->stats
.maxpacket
,
243 .count
= q
->vars
.count
,
244 .lastcount
= q
->vars
.lastcount
,
245 .drop_overlimit
= q
->drop_overlimit
,
246 .ldelay
= codel_time_to_us(q
->vars
.ldelay
),
247 .dropping
= q
->vars
.dropping
,
248 .ecn_mark
= q
->stats
.ecn_mark
,
249 .ce_mark
= q
->stats
.ce_mark
,
252 if (q
->vars
.dropping
) {
253 codel_tdiff_t delta
= q
->vars
.drop_next
- codel_get_time();
256 st
.drop_next
= codel_time_to_us(delta
);
258 st
.drop_next
= -codel_time_to_us(-delta
);
261 return gnet_stats_copy_app(d
, &st
, sizeof(st
));
264 static void codel_reset(struct Qdisc
*sch
)
266 struct codel_sched_data
*q
= qdisc_priv(sch
);
268 qdisc_reset_queue(sch
);
269 codel_vars_init(&q
->vars
);
272 static struct Qdisc_ops codel_qdisc_ops __read_mostly
= {
274 .priv_size
= sizeof(struct codel_sched_data
),
276 .enqueue
= codel_qdisc_enqueue
,
277 .dequeue
= codel_qdisc_dequeue
,
278 .peek
= qdisc_peek_dequeued
,
280 .reset
= codel_reset
,
281 .change
= codel_change
,
283 .dump_stats
= codel_dump_stats
,
284 .owner
= THIS_MODULE
,
287 static int __init
codel_module_init(void)
289 return register_qdisc(&codel_qdisc_ops
);
292 static void __exit
codel_module_exit(void)
294 unregister_qdisc(&codel_qdisc_ops
);
297 module_init(codel_module_init
)
298 module_exit(codel_module_exit
)
300 MODULE_DESCRIPTION("Controlled Delay queue discipline");
301 MODULE_AUTHOR("Dave Taht");
302 MODULE_AUTHOR("Eric Dumazet");
303 MODULE_LICENSE("Dual BSD/GPL");