Line data Source code
1 : // SPDX-License-Identifier: GPL-2.0
2 : /*
3 : * Floating proportions with flexible aging period
4 : *
5 : * Copyright (C) 2011, SUSE, Jan Kara <jack@suse.cz>
6 : *
7 : * The goal of this code is: Given different types of event, measure proportion
8 : * of each type of event over time. The proportions are measured with
9 : * exponentially decaying history to give smooth transitions. A formula
10 : * expressing proportion of event of type 'j' is:
11 : *
12 : * p_{j} = (\Sum_{i>=0} x_{i,j}/2^{i+1})/(\Sum_{i>=0} x_i/2^{i+1})
13 : *
14 : * Where x_{i,j} is j's number of events in i-th last time period and x_i is
15 : * total number of events in i-th last time period.
16 : *
17 : * Note that p_{j}'s are normalised, i.e.
18 : *
19 : * \Sum_{j} p_{j} = 1,
20 : *
21 : * This formula can be straightforwardly computed by maintaining denominator
22 : * (let's call it 'd') and for each event type its numerator (let's call it
23 : * 'n_j'). When an event of type 'j' happens, we simply need to do:
24 : * n_j++; d++;
25 : *
26 : * When a new period is declared, we could do:
27 : * d /= 2
28 : * for each j
29 : * n_j /= 2
30 : *
31 : * To avoid iteration over all event types, we instead shift numerator of event
32 : * j lazily when someone asks for a proportion of event j or when event j
33 : * occurs. This can bit trivially implemented by remembering last period in
34 : * which something happened with proportion of type j.
35 : */
36 : #include <linux/flex_proportions.h>
37 :
38 1 : int fprop_global_init(struct fprop_global *p, gfp_t gfp)
39 : {
40 : int err;
41 :
42 1 : p->period = 0;
43 : /* Use 1 to avoid dealing with periods with 0 events... */
44 2 : err = percpu_counter_init(&p->events, 1, gfp);
45 : if (err)
46 : return err;
47 2 : seqcount_init(&p->sequence);
48 : return 0;
49 : }
50 :
51 0 : void fprop_global_destroy(struct fprop_global *p)
52 : {
53 0 : percpu_counter_destroy(&p->events);
54 0 : }
55 :
56 : /*
57 : * Declare @periods new periods. It is upto the caller to make sure period
58 : * transitions cannot happen in parallel.
59 : *
60 : * The function returns true if the proportions are still defined and false
61 : * if aging zeroed out all events. This can be used to detect whether declaring
62 : * further periods has any effect.
63 : */
64 0 : bool fprop_new_period(struct fprop_global *p, int periods)
65 : {
66 : s64 events;
67 : unsigned long flags;
68 :
69 0 : local_irq_save(flags);
70 0 : events = percpu_counter_sum(&p->events);
71 : /*
72 : * Don't do anything if there are no events.
73 : */
74 0 : if (events <= 1) {
75 0 : local_irq_restore(flags);
76 0 : return false;
77 : }
78 0 : write_seqcount_begin(&p->sequence);
79 0 : if (periods < 64)
80 0 : events -= events >> periods;
81 : /* Use addition to avoid losing events happening between sum and set */
82 0 : percpu_counter_add(&p->events, -events);
83 0 : p->period += periods;
84 0 : write_seqcount_end(&p->sequence);
85 0 : local_irq_restore(flags);
86 :
87 0 : return true;
88 : }
89 :
90 : /*
91 : * ---- SINGLE ----
92 : */
93 :
94 0 : int fprop_local_init_single(struct fprop_local_single *pl)
95 : {
96 0 : pl->events = 0;
97 0 : pl->period = 0;
98 : raw_spin_lock_init(&pl->lock);
99 0 : return 0;
100 : }
101 :
102 0 : void fprop_local_destroy_single(struct fprop_local_single *pl)
103 : {
104 0 : }
105 :
106 0 : static void fprop_reflect_period_single(struct fprop_global *p,
107 : struct fprop_local_single *pl)
108 : {
109 0 : unsigned int period = p->period;
110 : unsigned long flags;
111 :
112 : /* Fast path - period didn't change */
113 0 : if (pl->period == period)
114 : return;
115 0 : raw_spin_lock_irqsave(&pl->lock, flags);
116 : /* Someone updated pl->period while we were spinning? */
117 0 : if (pl->period >= period) {
118 0 : raw_spin_unlock_irqrestore(&pl->lock, flags);
119 : return;
120 : }
121 : /* Aging zeroed our fraction? */
122 0 : if (period - pl->period < BITS_PER_LONG)
123 0 : pl->events >>= period - pl->period;
124 : else
125 0 : pl->events = 0;
126 0 : pl->period = period;
127 0 : raw_spin_unlock_irqrestore(&pl->lock, flags);
128 : }
129 :
130 : /* Event of type pl happened */
131 0 : void __fprop_inc_single(struct fprop_global *p, struct fprop_local_single *pl)
132 : {
133 0 : fprop_reflect_period_single(p, pl);
134 0 : pl->events++;
135 0 : percpu_counter_add(&p->events, 1);
136 0 : }
137 :
138 : /* Return fraction of events of type pl */
139 0 : void fprop_fraction_single(struct fprop_global *p,
140 : struct fprop_local_single *pl,
141 : unsigned long *numerator, unsigned long *denominator)
142 : {
143 : unsigned int seq;
144 : s64 num, den;
145 :
146 : do {
147 0 : seq = read_seqcount_begin(&p->sequence);
148 0 : fprop_reflect_period_single(p, pl);
149 0 : num = pl->events;
150 0 : den = percpu_counter_read_positive(&p->events);
151 0 : } while (read_seqcount_retry(&p->sequence, seq));
152 :
153 : /*
154 : * Make fraction <= 1 and denominator > 0 even in presence of percpu
155 : * counter errors
156 : */
157 0 : if (den <= num) {
158 0 : if (num)
159 : den = num;
160 : else
161 0 : den = 1;
162 : }
163 0 : *denominator = den;
164 0 : *numerator = num;
165 0 : }
166 :
167 : /*
168 : * ---- PERCPU ----
169 : */
170 : #define PROP_BATCH (8*(1+ilog2(nr_cpu_ids)))
171 :
172 1 : int fprop_local_init_percpu(struct fprop_local_percpu *pl, gfp_t gfp)
173 : {
174 : int err;
175 :
176 2 : err = percpu_counter_init(&pl->events, 0, gfp);
177 : if (err)
178 : return err;
179 1 : pl->period = 0;
180 : raw_spin_lock_init(&pl->lock);
181 : return 0;
182 : }
183 :
184 0 : void fprop_local_destroy_percpu(struct fprop_local_percpu *pl)
185 : {
186 0 : percpu_counter_destroy(&pl->events);
187 0 : }
188 :
189 0 : static void fprop_reflect_period_percpu(struct fprop_global *p,
190 : struct fprop_local_percpu *pl)
191 : {
192 0 : unsigned int period = p->period;
193 : unsigned long flags;
194 :
195 : /* Fast path - period didn't change */
196 0 : if (pl->period == period)
197 : return;
198 0 : raw_spin_lock_irqsave(&pl->lock, flags);
199 : /* Someone updated pl->period while we were spinning? */
200 0 : if (pl->period >= period) {
201 0 : raw_spin_unlock_irqrestore(&pl->lock, flags);
202 : return;
203 : }
204 : /* Aging zeroed our fraction? */
205 0 : if (period - pl->period < BITS_PER_LONG) {
206 0 : s64 val = percpu_counter_read(&pl->events);
207 :
208 : if (val < (nr_cpu_ids * PROP_BATCH))
209 : val = percpu_counter_sum(&pl->events);
210 :
211 0 : percpu_counter_add_batch(&pl->events,
212 0 : -val + (val >> (period-pl->period)), PROP_BATCH);
213 : } else
214 0 : percpu_counter_set(&pl->events, 0);
215 0 : pl->period = period;
216 0 : raw_spin_unlock_irqrestore(&pl->lock, flags);
217 : }
218 :
219 : /* Event of type pl happened */
220 0 : void __fprop_add_percpu(struct fprop_global *p, struct fprop_local_percpu *pl,
221 : long nr)
222 : {
223 0 : fprop_reflect_period_percpu(p, pl);
224 0 : percpu_counter_add_batch(&pl->events, nr, PROP_BATCH);
225 0 : percpu_counter_add(&p->events, nr);
226 0 : }
227 :
228 0 : void fprop_fraction_percpu(struct fprop_global *p,
229 : struct fprop_local_percpu *pl,
230 : unsigned long *numerator, unsigned long *denominator)
231 : {
232 : unsigned int seq;
233 : s64 num, den;
234 :
235 : do {
236 0 : seq = read_seqcount_begin(&p->sequence);
237 0 : fprop_reflect_period_percpu(p, pl);
238 0 : num = percpu_counter_read_positive(&pl->events);
239 0 : den = percpu_counter_read_positive(&p->events);
240 0 : } while (read_seqcount_retry(&p->sequence, seq));
241 :
242 : /*
243 : * Make fraction <= 1 and denominator > 0 even in presence of percpu
244 : * counter errors
245 : */
246 0 : if (den <= num) {
247 0 : if (num)
248 : den = num;
249 : else
250 0 : den = 1;
251 : }
252 0 : *denominator = den;
253 0 : *numerator = num;
254 0 : }
255 :
256 : /*
257 : * Like __fprop_add_percpu() except that event is counted only if the given
258 : * type has fraction smaller than @max_frac/FPROP_FRAC_BASE
259 : */
260 0 : void __fprop_add_percpu_max(struct fprop_global *p,
261 : struct fprop_local_percpu *pl, int max_frac, long nr)
262 : {
263 0 : if (unlikely(max_frac < FPROP_FRAC_BASE)) {
264 : unsigned long numerator, denominator;
265 : s64 tmp;
266 :
267 0 : fprop_fraction_percpu(p, pl, &numerator, &denominator);
268 : /* Adding 'nr' to fraction exceeds max_frac/FPROP_FRAC_BASE? */
269 0 : tmp = (u64)denominator * max_frac -
270 0 : ((u64)numerator << FPROP_FRAC_SHIFT);
271 0 : if (tmp < 0) {
272 : /* Maximum fraction already exceeded? */
273 0 : return;
274 0 : } else if (tmp < nr * (FPROP_FRAC_BASE - max_frac)) {
275 : /* Add just enough for the fraction to saturate */
276 0 : nr = div_u64(tmp + FPROP_FRAC_BASE - max_frac - 1,
277 : FPROP_FRAC_BASE - max_frac);
278 : }
279 : }
280 :
281 : __fprop_add_percpu(p, pl, nr);
282 : }
|
