Loading...
1/*
2 * kernel/sched/cpupri.c
3 *
4 * CPU priority management
5 *
6 * Copyright (C) 2007-2008 Novell
7 *
8 * Author: Gregory Haskins <ghaskins@novell.com>
9 *
10 * This code tracks the priority of each CPU so that global migration
11 * decisions are easy to calculate. Each CPU can be in a state as follows:
12 *
13 * (INVALID), IDLE, NORMAL, RT1, ... RT99
14 *
15 * going from the lowest priority to the highest. CPUs in the INVALID state
16 * are not eligible for routing. The system maintains this state with
17 * a 2 dimensional bitmap (the first for priority class, the second for cpus
18 * in that class). Therefore a typical application without affinity
19 * restrictions can find a suitable CPU with O(1) complexity (e.g. two bit
20 * searches). For tasks with affinity restrictions, the algorithm has a
21 * worst case complexity of O(min(102, nr_domcpus)), though the scenario that
22 * yields the worst case search is fairly contrived.
23 *
24 * This program is free software; you can redistribute it and/or
25 * modify it under the terms of the GNU General Public License
26 * as published by the Free Software Foundation; version 2
27 * of the License.
28 */
29
30#include <linux/gfp.h>
31#include "cpupri.h"
32
33/* Convert between a 140 based task->prio, and our 102 based cpupri */
34static int convert_prio(int prio)
35{
36 int cpupri;
37
38 if (prio == CPUPRI_INVALID)
39 cpupri = CPUPRI_INVALID;
40 else if (prio == MAX_PRIO)
41 cpupri = CPUPRI_IDLE;
42 else if (prio >= MAX_RT_PRIO)
43 cpupri = CPUPRI_NORMAL;
44 else
45 cpupri = MAX_RT_PRIO - prio + 1;
46
47 return cpupri;
48}
49
50/**
51 * cpupri_find - find the best (lowest-pri) CPU in the system
52 * @cp: The cpupri context
53 * @p: The task
54 * @lowest_mask: A mask to fill in with selected CPUs (or NULL)
55 *
56 * Note: This function returns the recommended CPUs as calculated during the
57 * current invocation. By the time the call returns, the CPUs may have in
58 * fact changed priorities any number of times. While not ideal, it is not
59 * an issue of correctness since the normal rebalancer logic will correct
60 * any discrepancies created by racing against the uncertainty of the current
61 * priority configuration.
62 *
63 * Returns: (int)bool - CPUs were found
64 */
65int cpupri_find(struct cpupri *cp, struct task_struct *p,
66 struct cpumask *lowest_mask)
67{
68 int idx = 0;
69 int task_pri = convert_prio(p->prio);
70
71 if (task_pri >= MAX_RT_PRIO)
72 return 0;
73
74 for (idx = 0; idx < task_pri; idx++) {
75 struct cpupri_vec *vec = &cp->pri_to_cpu[idx];
76 int skip = 0;
77
78 if (!atomic_read(&(vec)->count))
79 skip = 1;
80 /*
81 * When looking at the vector, we need to read the counter,
82 * do a memory barrier, then read the mask.
83 *
84 * Note: This is still all racey, but we can deal with it.
85 * Ideally, we only want to look at masks that are set.
86 *
87 * If a mask is not set, then the only thing wrong is that we
88 * did a little more work than necessary.
89 *
90 * If we read a zero count but the mask is set, because of the
91 * memory barriers, that can only happen when the highest prio
92 * task for a run queue has left the run queue, in which case,
93 * it will be followed by a pull. If the task we are processing
94 * fails to find a proper place to go, that pull request will
95 * pull this task if the run queue is running at a lower
96 * priority.
97 */
98 smp_rmb();
99
100 /* Need to do the rmb for every iteration */
101 if (skip)
102 continue;
103
104 if (cpumask_any_and(&p->cpus_allowed, vec->mask) >= nr_cpu_ids)
105 continue;
106
107 if (lowest_mask) {
108 cpumask_and(lowest_mask, &p->cpus_allowed, vec->mask);
109
110 /*
111 * We have to ensure that we have at least one bit
112 * still set in the array, since the map could have
113 * been concurrently emptied between the first and
114 * second reads of vec->mask. If we hit this
115 * condition, simply act as though we never hit this
116 * priority level and continue on.
117 */
118 if (cpumask_any(lowest_mask) >= nr_cpu_ids)
119 continue;
120 }
121
122 return 1;
123 }
124
125 return 0;
126}
127
128/**
129 * cpupri_set - update the cpu priority setting
130 * @cp: The cpupri context
131 * @cpu: The target cpu
132 * @newpri: The priority (INVALID-RT99) to assign to this CPU
133 *
134 * Note: Assumes cpu_rq(cpu)->lock is locked
135 *
136 * Returns: (void)
137 */
138void cpupri_set(struct cpupri *cp, int cpu, int newpri)
139{
140 int *currpri = &cp->cpu_to_pri[cpu];
141 int oldpri = *currpri;
142 int do_mb = 0;
143
144 newpri = convert_prio(newpri);
145
146 BUG_ON(newpri >= CPUPRI_NR_PRIORITIES);
147
148 if (newpri == oldpri)
149 return;
150
151 /*
152 * If the cpu was currently mapped to a different value, we
153 * need to map it to the new value then remove the old value.
154 * Note, we must add the new value first, otherwise we risk the
155 * cpu being missed by the priority loop in cpupri_find.
156 */
157 if (likely(newpri != CPUPRI_INVALID)) {
158 struct cpupri_vec *vec = &cp->pri_to_cpu[newpri];
159
160 cpumask_set_cpu(cpu, vec->mask);
161 /*
162 * When adding a new vector, we update the mask first,
163 * do a write memory barrier, and then update the count, to
164 * make sure the vector is visible when count is set.
165 */
166 smp_mb__before_atomic_inc();
167 atomic_inc(&(vec)->count);
168 do_mb = 1;
169 }
170 if (likely(oldpri != CPUPRI_INVALID)) {
171 struct cpupri_vec *vec = &cp->pri_to_cpu[oldpri];
172
173 /*
174 * Because the order of modification of the vec->count
175 * is important, we must make sure that the update
176 * of the new prio is seen before we decrement the
177 * old prio. This makes sure that the loop sees
178 * one or the other when we raise the priority of
179 * the run queue. We don't care about when we lower the
180 * priority, as that will trigger an rt pull anyway.
181 *
182 * We only need to do a memory barrier if we updated
183 * the new priority vec.
184 */
185 if (do_mb)
186 smp_mb__after_atomic_inc();
187
188 /*
189 * When removing from the vector, we decrement the counter first
190 * do a memory barrier and then clear the mask.
191 */
192 atomic_dec(&(vec)->count);
193 smp_mb__after_atomic_inc();
194 cpumask_clear_cpu(cpu, vec->mask);
195 }
196
197 *currpri = newpri;
198}
199
200/**
201 * cpupri_init - initialize the cpupri structure
202 * @cp: The cpupri context
203 *
204 * Returns: -ENOMEM if memory fails.
205 */
206int cpupri_init(struct cpupri *cp)
207{
208 int i;
209
210 memset(cp, 0, sizeof(*cp));
211
212 for (i = 0; i < CPUPRI_NR_PRIORITIES; i++) {
213 struct cpupri_vec *vec = &cp->pri_to_cpu[i];
214
215 atomic_set(&vec->count, 0);
216 if (!zalloc_cpumask_var(&vec->mask, GFP_KERNEL))
217 goto cleanup;
218 }
219
220 for_each_possible_cpu(i)
221 cp->cpu_to_pri[i] = CPUPRI_INVALID;
222 return 0;
223
224cleanup:
225 for (i--; i >= 0; i--)
226 free_cpumask_var(cp->pri_to_cpu[i].mask);
227 return -ENOMEM;
228}
229
230/**
231 * cpupri_cleanup - clean up the cpupri structure
232 * @cp: The cpupri context
233 */
234void cpupri_cleanup(struct cpupri *cp)
235{
236 int i;
237
238 for (i = 0; i < CPUPRI_NR_PRIORITIES; i++)
239 free_cpumask_var(cp->pri_to_cpu[i].mask);
240}
1/*
2 * kernel/sched/cpupri.c
3 *
4 * CPU priority management
5 *
6 * Copyright (C) 2007-2008 Novell
7 *
8 * Author: Gregory Haskins <ghaskins@novell.com>
9 *
10 * This code tracks the priority of each CPU so that global migration
11 * decisions are easy to calculate. Each CPU can be in a state as follows:
12 *
13 * (INVALID), IDLE, NORMAL, RT1, ... RT99
14 *
15 * going from the lowest priority to the highest. CPUs in the INVALID state
16 * are not eligible for routing. The system maintains this state with
17 * a 2 dimensional bitmap (the first for priority class, the second for CPUs
18 * in that class). Therefore a typical application without affinity
19 * restrictions can find a suitable CPU with O(1) complexity (e.g. two bit
20 * searches). For tasks with affinity restrictions, the algorithm has a
21 * worst case complexity of O(min(102, nr_domcpus)), though the scenario that
22 * yields the worst case search is fairly contrived.
23 *
24 * This program is free software; you can redistribute it and/or
25 * modify it under the terms of the GNU General Public License
26 * as published by the Free Software Foundation; version 2
27 * of the License.
28 */
29#include "sched.h"
30
31/* Convert between a 140 based task->prio, and our 102 based cpupri */
32static int convert_prio(int prio)
33{
34 int cpupri;
35
36 if (prio == CPUPRI_INVALID)
37 cpupri = CPUPRI_INVALID;
38 else if (prio == MAX_PRIO)
39 cpupri = CPUPRI_IDLE;
40 else if (prio >= MAX_RT_PRIO)
41 cpupri = CPUPRI_NORMAL;
42 else
43 cpupri = MAX_RT_PRIO - prio + 1;
44
45 return cpupri;
46}
47
48/**
49 * cpupri_find - find the best (lowest-pri) CPU in the system
50 * @cp: The cpupri context
51 * @p: The task
52 * @lowest_mask: A mask to fill in with selected CPUs (or NULL)
53 *
54 * Note: This function returns the recommended CPUs as calculated during the
55 * current invocation. By the time the call returns, the CPUs may have in
56 * fact changed priorities any number of times. While not ideal, it is not
57 * an issue of correctness since the normal rebalancer logic will correct
58 * any discrepancies created by racing against the uncertainty of the current
59 * priority configuration.
60 *
61 * Return: (int)bool - CPUs were found
62 */
63int cpupri_find(struct cpupri *cp, struct task_struct *p,
64 struct cpumask *lowest_mask)
65{
66 int idx = 0;
67 int task_pri = convert_prio(p->prio);
68
69 BUG_ON(task_pri >= CPUPRI_NR_PRIORITIES);
70
71 for (idx = 0; idx < task_pri; idx++) {
72 struct cpupri_vec *vec = &cp->pri_to_cpu[idx];
73 int skip = 0;
74
75 if (!atomic_read(&(vec)->count))
76 skip = 1;
77 /*
78 * When looking at the vector, we need to read the counter,
79 * do a memory barrier, then read the mask.
80 *
81 * Note: This is still all racey, but we can deal with it.
82 * Ideally, we only want to look at masks that are set.
83 *
84 * If a mask is not set, then the only thing wrong is that we
85 * did a little more work than necessary.
86 *
87 * If we read a zero count but the mask is set, because of the
88 * memory barriers, that can only happen when the highest prio
89 * task for a run queue has left the run queue, in which case,
90 * it will be followed by a pull. If the task we are processing
91 * fails to find a proper place to go, that pull request will
92 * pull this task if the run queue is running at a lower
93 * priority.
94 */
95 smp_rmb();
96
97 /* Need to do the rmb for every iteration */
98 if (skip)
99 continue;
100
101 if (cpumask_any_and(&p->cpus_allowed, vec->mask) >= nr_cpu_ids)
102 continue;
103
104 if (lowest_mask) {
105 cpumask_and(lowest_mask, &p->cpus_allowed, vec->mask);
106
107 /*
108 * We have to ensure that we have at least one bit
109 * still set in the array, since the map could have
110 * been concurrently emptied between the first and
111 * second reads of vec->mask. If we hit this
112 * condition, simply act as though we never hit this
113 * priority level and continue on.
114 */
115 if (cpumask_any(lowest_mask) >= nr_cpu_ids)
116 continue;
117 }
118
119 return 1;
120 }
121
122 return 0;
123}
124
125/**
126 * cpupri_set - update the CPU priority setting
127 * @cp: The cpupri context
128 * @cpu: The target CPU
129 * @newpri: The priority (INVALID-RT99) to assign to this CPU
130 *
131 * Note: Assumes cpu_rq(cpu)->lock is locked
132 *
133 * Returns: (void)
134 */
135void cpupri_set(struct cpupri *cp, int cpu, int newpri)
136{
137 int *currpri = &cp->cpu_to_pri[cpu];
138 int oldpri = *currpri;
139 int do_mb = 0;
140
141 newpri = convert_prio(newpri);
142
143 BUG_ON(newpri >= CPUPRI_NR_PRIORITIES);
144
145 if (newpri == oldpri)
146 return;
147
148 /*
149 * If the CPU was currently mapped to a different value, we
150 * need to map it to the new value then remove the old value.
151 * Note, we must add the new value first, otherwise we risk the
152 * cpu being missed by the priority loop in cpupri_find.
153 */
154 if (likely(newpri != CPUPRI_INVALID)) {
155 struct cpupri_vec *vec = &cp->pri_to_cpu[newpri];
156
157 cpumask_set_cpu(cpu, vec->mask);
158 /*
159 * When adding a new vector, we update the mask first,
160 * do a write memory barrier, and then update the count, to
161 * make sure the vector is visible when count is set.
162 */
163 smp_mb__before_atomic();
164 atomic_inc(&(vec)->count);
165 do_mb = 1;
166 }
167 if (likely(oldpri != CPUPRI_INVALID)) {
168 struct cpupri_vec *vec = &cp->pri_to_cpu[oldpri];
169
170 /*
171 * Because the order of modification of the vec->count
172 * is important, we must make sure that the update
173 * of the new prio is seen before we decrement the
174 * old prio. This makes sure that the loop sees
175 * one or the other when we raise the priority of
176 * the run queue. We don't care about when we lower the
177 * priority, as that will trigger an rt pull anyway.
178 *
179 * We only need to do a memory barrier if we updated
180 * the new priority vec.
181 */
182 if (do_mb)
183 smp_mb__after_atomic();
184
185 /*
186 * When removing from the vector, we decrement the counter first
187 * do a memory barrier and then clear the mask.
188 */
189 atomic_dec(&(vec)->count);
190 smp_mb__after_atomic();
191 cpumask_clear_cpu(cpu, vec->mask);
192 }
193
194 *currpri = newpri;
195}
196
197/**
198 * cpupri_init - initialize the cpupri structure
199 * @cp: The cpupri context
200 *
201 * Return: -ENOMEM on memory allocation failure.
202 */
203int cpupri_init(struct cpupri *cp)
204{
205 int i;
206
207 for (i = 0; i < CPUPRI_NR_PRIORITIES; i++) {
208 struct cpupri_vec *vec = &cp->pri_to_cpu[i];
209
210 atomic_set(&vec->count, 0);
211 if (!zalloc_cpumask_var(&vec->mask, GFP_KERNEL))
212 goto cleanup;
213 }
214
215 cp->cpu_to_pri = kcalloc(nr_cpu_ids, sizeof(int), GFP_KERNEL);
216 if (!cp->cpu_to_pri)
217 goto cleanup;
218
219 for_each_possible_cpu(i)
220 cp->cpu_to_pri[i] = CPUPRI_INVALID;
221
222 return 0;
223
224cleanup:
225 for (i--; i >= 0; i--)
226 free_cpumask_var(cp->pri_to_cpu[i].mask);
227 return -ENOMEM;
228}
229
230/**
231 * cpupri_cleanup - clean up the cpupri structure
232 * @cp: The cpupri context
233 */
234void cpupri_cleanup(struct cpupri *cp)
235{
236 int i;
237
238 kfree(cp->cpu_to_pri);
239 for (i = 0; i < CPUPRI_NR_PRIORITIES; i++)
240 free_cpumask_var(cp->pri_to_cpu[i].mask);
241}