block: add bio_start_io_acct_time() to control start_time
[linux-2.6-block.git] / include / net / fq_impl.h
CommitLineData
fb9e53cc 1/* SPDX-License-Identifier: GPL-2.0-only */
557fc4a0
MK
2/*
3 * Copyright (c) 2016 Qualcomm Atheros, Inc
4 *
557fc4a0
MK
5 * Based on net/sched/sch_fq_codel.c
6 */
7#ifndef __NET_SCHED_FQ_IMPL_H
8#define __NET_SCHED_FQ_IMPL_H
9
10#include <net/fq.h>
11
12/* functions that are embedded into includer */
13
07be2fed
FF
14
15static void
16__fq_adjust_removal(struct fq *fq, struct fq_flow *flow, unsigned int packets,
17 unsigned int bytes, unsigned int truesize)
18{
19 struct fq_tin *tin = flow->tin;
d7b64929 20 int idx;
07be2fed
FF
21
22 tin->backlog_bytes -= bytes;
23 tin->backlog_packets -= packets;
24 flow->backlog -= bytes;
25 fq->backlog -= packets;
26 fq->memory_usage -= truesize;
d7b64929
FF
27
28 if (flow->backlog)
29 return;
30
31 if (flow == &tin->default_flow) {
32 list_del_init(&tin->tin_list);
33 return;
34 }
35
36 idx = flow - fq->flows;
37 __clear_bit(idx, fq->flows_bitmap);
07be2fed
FF
38}
39
8c418b5b
JB
40static void fq_adjust_removal(struct fq *fq,
41 struct fq_flow *flow,
42 struct sk_buff *skb)
557fc4a0 43{
07be2fed 44 __fq_adjust_removal(fq, flow, 1, skb->len, skb->truesize);
8c418b5b
JB
45}
46
8c418b5b
JB
47static struct sk_buff *fq_flow_dequeue(struct fq *fq,
48 struct fq_flow *flow)
49{
50 struct sk_buff *skb;
51
52 lockdep_assert_held(&fq->lock);
53
54 skb = __skb_dequeue(&flow->queue);
55 if (!skb)
56 return NULL;
57
58 fq_adjust_removal(fq, flow, skb);
557fc4a0
MK
59
60 return skb;
61}
62
07be2fed
FF
63static int fq_flow_drop(struct fq *fq, struct fq_flow *flow,
64 fq_skb_free_t free_func)
65{
66 unsigned int packets = 0, bytes = 0, truesize = 0;
67 struct fq_tin *tin = flow->tin;
68 struct sk_buff *skb;
69 int pending;
70
71 lockdep_assert_held(&fq->lock);
72
73 pending = min_t(int, 32, skb_queue_len(&flow->queue) / 2);
74 do {
75 skb = __skb_dequeue(&flow->queue);
76 if (!skb)
77 break;
78
79 packets++;
80 bytes += skb->len;
81 truesize += skb->truesize;
82 free_func(fq, tin, flow, skb);
83 } while (packets < pending);
84
85 __fq_adjust_removal(fq, flow, packets, bytes, truesize);
07be2fed
FF
86
87 return packets;
88}
89
557fc4a0
MK
90static struct sk_buff *fq_tin_dequeue(struct fq *fq,
91 struct fq_tin *tin,
92 fq_tin_dequeue_t dequeue_func)
93{
94 struct fq_flow *flow;
95 struct list_head *head;
96 struct sk_buff *skb;
97
98 lockdep_assert_held(&fq->lock);
99
100begin:
101 head = &tin->new_flows;
102 if (list_empty(head)) {
103 head = &tin->old_flows;
104 if (list_empty(head))
105 return NULL;
106 }
107
108 flow = list_first_entry(head, struct fq_flow, flowchain);
109
110 if (flow->deficit <= 0) {
111 flow->deficit += fq->quantum;
112 list_move_tail(&flow->flowchain,
113 &tin->old_flows);
114 goto begin;
115 }
116
117 skb = dequeue_func(fq, tin, flow);
118 if (!skb) {
119 /* force a pass through old_flows to prevent starvation */
120 if ((head == &tin->new_flows) &&
121 !list_empty(&tin->old_flows)) {
122 list_move_tail(&flow->flowchain, &tin->old_flows);
123 } else {
124 list_del_init(&flow->flowchain);
125 flow->tin = NULL;
126 }
127 goto begin;
128 }
129
130 flow->deficit -= skb->len;
131 tin->tx_bytes += skb->len;
132 tin->tx_packets++;
133
134 return skb;
135}
136
f2af2df8
FF
137static u32 fq_flow_idx(struct fq *fq, struct sk_buff *skb)
138{
48a54f6b 139 u32 hash = skb_get_hash(skb);
f2af2df8
FF
140
141 return reciprocal_scale(hash, fq->flows_cnt);
142}
143
557fc4a0 144static struct fq_flow *fq_flow_classify(struct fq *fq,
f2af2df8 145 struct fq_tin *tin, u32 idx,
bf9009bf 146 struct sk_buff *skb)
557fc4a0
MK
147{
148 struct fq_flow *flow;
557fc4a0
MK
149
150 lockdep_assert_held(&fq->lock);
151
557fc4a0 152 flow = &fq->flows[idx];
557fc4a0 153 if (flow->tin && flow->tin != tin) {
bf9009bf 154 flow = &tin->default_flow;
557fc4a0
MK
155 tin->collisions++;
156 fq->collisions++;
157 }
158
159 if (!flow->tin)
160 tin->flows++;
161
162 return flow;
163}
164
d7b64929 165static struct fq_flow *fq_find_fattest_flow(struct fq *fq)
b43e7199 166{
d7b64929
FF
167 struct fq_tin *tin;
168 struct fq_flow *flow = NULL;
169 u32 len = 0;
170 int i;
b43e7199 171
d7b64929
FF
172 for_each_set_bit(i, fq->flows_bitmap, fq->flows_cnt) {
173 struct fq_flow *cur = &fq->flows[i];
174 unsigned int cur_len;
b43e7199 175
d7b64929
FF
176 cur_len = cur->backlog;
177 if (cur_len <= len)
178 continue;
179
180 flow = cur;
181 len = cur_len;
182 }
183
184 list_for_each_entry(tin, &fq->tin_backlog, tin_list) {
185 unsigned int cur_len = tin->default_flow.backlog;
b43e7199 186
d7b64929
FF
187 if (cur_len <= len)
188 continue;
189
190 flow = &tin->default_flow;
191 len = cur_len;
192 }
193
194 return flow;
b43e7199
MK
195}
196
557fc4a0 197static void fq_tin_enqueue(struct fq *fq,
f2af2df8 198 struct fq_tin *tin, u32 idx,
557fc4a0 199 struct sk_buff *skb,
bf9009bf 200 fq_skb_free_t free_func)
557fc4a0
MK
201{
202 struct fq_flow *flow;
0bfe649f 203 bool oom;
557fc4a0
MK
204
205 lockdep_assert_held(&fq->lock);
206
bf9009bf 207 flow = fq_flow_classify(fq, tin, idx, skb);
557fc4a0 208
d7b64929
FF
209 if (!flow->backlog) {
210 if (flow != &tin->default_flow)
211 __set_bit(idx, fq->flows_bitmap);
212 else if (list_empty(&tin->tin_list))
213 list_add(&tin->tin_list, &fq->tin_backlog);
214 }
215
557fc4a0
MK
216 flow->tin = tin;
217 flow->backlog += skb->len;
218 tin->backlog_bytes += skb->len;
219 tin->backlog_packets++;
097b065b 220 fq->memory_usage += skb->truesize;
557fc4a0
MK
221 fq->backlog++;
222
557fc4a0
MK
223 if (list_empty(&flow->flowchain)) {
224 flow->deficit = fq->quantum;
225 list_add_tail(&flow->flowchain,
226 &tin->new_flows);
227 }
228
229 __skb_queue_tail(&flow->queue, skb);
0bfe649f
THJ
230 oom = (fq->memory_usage > fq->memory_limit);
231 while (fq->backlog > fq->limit || oom) {
d7b64929 232 flow = fq_find_fattest_flow(fq);
557fc4a0
MK
233 if (!flow)
234 return;
235
07be2fed 236 if (!fq_flow_drop(fq, flow, free_func))
557fc4a0
MK
237 return;
238
557fc4a0
MK
239 flow->tin->overlimit++;
240 fq->overlimit++;
0bfe649f 241 if (oom) {
097b065b 242 fq->overmemory++;
0bfe649f
THJ
243 oom = (fq->memory_usage > fq->memory_limit);
244 }
557fc4a0
MK
245 }
246}
247
8c418b5b
JB
248static void fq_flow_filter(struct fq *fq,
249 struct fq_flow *flow,
250 fq_skb_filter_t filter_func,
251 void *filter_data,
252 fq_skb_free_t free_func)
253{
254 struct fq_tin *tin = flow->tin;
255 struct sk_buff *skb, *tmp;
256
257 lockdep_assert_held(&fq->lock);
258
259 skb_queue_walk_safe(&flow->queue, skb, tmp) {
260 if (!filter_func(fq, tin, flow, skb, filter_data))
261 continue;
262
263 __skb_unlink(skb, &flow->queue);
264 fq_adjust_removal(fq, flow, skb);
265 free_func(fq, tin, flow, skb);
266 }
8c418b5b
JB
267}
268
269static void fq_tin_filter(struct fq *fq,
270 struct fq_tin *tin,
271 fq_skb_filter_t filter_func,
272 void *filter_data,
273 fq_skb_free_t free_func)
274{
275 struct fq_flow *flow;
276
277 lockdep_assert_held(&fq->lock);
278
279 list_for_each_entry(flow, &tin->new_flows, flowchain)
280 fq_flow_filter(fq, flow, filter_func, filter_data, free_func);
281 list_for_each_entry(flow, &tin->old_flows, flowchain)
282 fq_flow_filter(fq, flow, filter_func, filter_data, free_func);
283}
284
557fc4a0
MK
285static void fq_flow_reset(struct fq *fq,
286 struct fq_flow *flow,
287 fq_skb_free_t free_func)
288{
d7b64929 289 struct fq_tin *tin = flow->tin;
557fc4a0
MK
290 struct sk_buff *skb;
291
292 while ((skb = fq_flow_dequeue(fq, flow)))
d7b64929 293 free_func(fq, tin, flow, skb);
557fc4a0 294
d7b64929 295 if (!list_empty(&flow->flowchain)) {
557fc4a0 296 list_del_init(&flow->flowchain);
d7b64929
FF
297 if (list_empty(&tin->new_flows) &&
298 list_empty(&tin->old_flows))
299 list_del_init(&tin->tin_list);
300 }
557fc4a0
MK
301
302 flow->tin = NULL;
303
304 WARN_ON_ONCE(flow->backlog);
305}
306
307static void fq_tin_reset(struct fq *fq,
308 struct fq_tin *tin,
309 fq_skb_free_t free_func)
310{
311 struct list_head *head;
312 struct fq_flow *flow;
313
314 for (;;) {
315 head = &tin->new_flows;
316 if (list_empty(head)) {
317 head = &tin->old_flows;
318 if (list_empty(head))
319 break;
320 }
321
322 flow = list_first_entry(head, struct fq_flow, flowchain);
323 fq_flow_reset(fq, flow, free_func);
324 }
325
d7b64929 326 WARN_ON_ONCE(!list_empty(&tin->tin_list));
557fc4a0
MK
327 WARN_ON_ONCE(tin->backlog_bytes);
328 WARN_ON_ONCE(tin->backlog_packets);
329}
330
331static void fq_flow_init(struct fq_flow *flow)
332{
333 INIT_LIST_HEAD(&flow->flowchain);
557fc4a0
MK
334 __skb_queue_head_init(&flow->queue);
335}
336
337static void fq_tin_init(struct fq_tin *tin)
338{
339 INIT_LIST_HEAD(&tin->new_flows);
340 INIT_LIST_HEAD(&tin->old_flows);
d7b64929 341 INIT_LIST_HEAD(&tin->tin_list);
bf9009bf 342 fq_flow_init(&tin->default_flow);
557fc4a0
MK
343}
344
345static int fq_init(struct fq *fq, int flows_cnt)
346{
347 int i;
348
349 memset(fq, 0, sizeof(fq[0]));
557fc4a0 350 spin_lock_init(&fq->lock);
d7b64929 351 INIT_LIST_HEAD(&fq->tin_backlog);
557fc4a0 352 fq->flows_cnt = max_t(u32, flows_cnt, 1);
557fc4a0
MK
353 fq->quantum = 300;
354 fq->limit = 8192;
097b065b 355 fq->memory_limit = 16 << 20; /* 16 MBytes */
557fc4a0 356
71e67c3b 357 fq->flows = kvcalloc(fq->flows_cnt, sizeof(fq->flows[0]), GFP_KERNEL);
557fc4a0
MK
358 if (!fq->flows)
359 return -ENOMEM;
360
d7b64929
FF
361 fq->flows_bitmap = kcalloc(BITS_TO_LONGS(fq->flows_cnt), sizeof(long),
362 GFP_KERNEL);
363 if (!fq->flows_bitmap) {
364 kvfree(fq->flows);
365 fq->flows = NULL;
366 return -ENOMEM;
367 }
368
557fc4a0
MK
369 for (i = 0; i < fq->flows_cnt; i++)
370 fq_flow_init(&fq->flows[i]);
371
372 return 0;
373}
374
375static void fq_reset(struct fq *fq,
376 fq_skb_free_t free_func)
377{
378 int i;
379
380 for (i = 0; i < fq->flows_cnt; i++)
381 fq_flow_reset(fq, &fq->flows[i], free_func);
382
71e67c3b 383 kvfree(fq->flows);
557fc4a0 384 fq->flows = NULL;
d7b64929
FF
385
386 kfree(fq->flows_bitmap);
387 fq->flows_bitmap = NULL;
557fc4a0
MK
388}
389
390#endif