summaryrefslogtreecommitdiff
path: root/kernel/trace/trace_workqueue.c
blob: f8118d39ca9b7b8afce44b91a5d7a9e8adc337e4 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
/*
 * Workqueue statistical tracer.
 *
 * Copyright (C) 2008 Frederic Weisbecker <fweisbec@gmail.com>
 *
 */


#include <trace/workqueue.h>
#include <linux/list.h>
#include "trace_stat.h"
#include "trace.h"


/* A cpu workqueue thread */
struct cpu_workqueue_stats {
	struct list_head            list;
/* Useful to know if we print the cpu headers */
	bool		            first_entry;
	int		            cpu;
	pid_t 			    pid;
/* Can be inserted from interrupt or user context, need to be atomic */
	atomic_t 	            inserted;
/*
 *  Don't need to be atomic, works are serialized in a single workqueue thread
 *  on a single CPU.
 */
	unsigned int 	 	    executed;
};

/* List of workqueue threads on one cpu */
struct workqueue_global_stats {
	struct list_head	list;
	spinlock_t		lock;
};

/* Don't need a global lock because allocated before the workqueues, and
 * never freed.
 */
static struct workqueue_global_stats *all_workqueue_stat;

/* Insertion of a work */
static void
probe_workqueue_insertion(struct task_struct *wq_thread,
			  struct work_struct *work)
{
	int cpu = cpumask_first(&wq_thread->cpus_allowed);
	struct cpu_workqueue_stats *node, *next;
	unsigned long flags;

	spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
	list_for_each_entry_safe(node, next, &all_workqueue_stat[cpu].list,
							list) {
		if (node->pid == wq_thread->pid) {
			atomic_inc(&node->inserted);
			goto found;
		}
	}
	pr_debug("trace_workqueue: entry not found\n");
found:
	spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
}

/* Execution of a work */
static void
probe_workqueue_execution(struct task_struct *wq_thread,
			  struct work_struct *work)
{
	int cpu = cpumask_first(&wq_thread->cpus_allowed);
	struct cpu_workqueue_stats *node, *next;
	unsigned long flags;

	spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
	list_for_each_entry_safe(node, next, &all_workqueue_stat[cpu].list,
							list) {
		if (node->pid == wq_thread->pid) {
			node->executed++;
			goto found;
		}
	}
	pr_debug("trace_workqueue: entry not found\n");
found:
	spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
}

/* Creation of a cpu workqueue thread */
static void probe_workqueue_creation(struct task_struct *wq_thread, int cpu)
{
	struct cpu_workqueue_stats *cws;
	unsigned long flags;

	WARN_ON(cpu < 0 || cpu >= num_possible_cpus());

	/* Workqueues are sometimes created in atomic context */
	cws = kzalloc(sizeof(struct cpu_workqueue_stats), GFP_ATOMIC);
	if (!cws) {
		pr_warning("trace_workqueue: not enough memory\n");
		return;
	}
	tracing_record_cmdline(wq_thread);

	INIT_LIST_HEAD(&cws->list);
	cws->cpu = cpu;

	cws->pid = wq_thread->pid;

	spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
	if (list_empty(&all_workqueue_stat[cpu].list))
		cws->first_entry = true;
	list_add_tail(&cws->list, &all_workqueue_stat[cpu].list);
	spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
}

/* Destruction of a cpu workqueue thread */
static void probe_workqueue_destruction(struct task_struct *wq_thread)
{
	/* Workqueue only execute on one cpu */
	int cpu = cpumask_first(&wq_thread->cpus_allowed);
	struct cpu_workqueue_stats *node, *next;
	unsigned long flags;

	spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
	list_for_each_entry_safe(node, next, &all_workqueue_stat[cpu].list,
							list) {
		if (node->pid == wq_thread->pid) {
			list_del(&node->list);
			kfree(node);
			goto found;
		}
	}

	pr_debug("trace_workqueue: don't find workqueue to destroy\n");
found:
	spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);

}

static struct cpu_workqueue_stats *workqueue_stat_start_cpu(int cpu)
{
	unsigned long flags;
	struct cpu_workqueue_stats *ret = NULL;


	spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);

	if (!list_empty(&all_workqueue_stat[cpu].list))
		ret = list_entry(all_workqueue_stat[cpu].list.next,
				 struct cpu_workqueue_stats, list);

	spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);

	return ret;
}

static void *workqueue_stat_start(void)
{
	int cpu;
	void *ret = NULL;

	for_each_possible_cpu(cpu) {
		ret = workqueue_stat_start_cpu(cpu);
		if (ret)
			return ret;
	}
	return NULL;
}

static void *workqueue_stat_next(void *prev, int idx)
{
	struct cpu_workqueue_stats *prev_cws = prev;
	int cpu = prev_cws->cpu;
	unsigned long flags;
	void *ret = NULL;

	spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
	if (list_is_last(&prev_cws->list, &all_workqueue_stat[cpu].list)) {
		spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
		for (++cpu ; cpu < num_possible_cpus(); cpu++) {
			ret = workqueue_stat_start_cpu(cpu);
			if (ret)
				return ret;
		}
		return NULL;
	}
	spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);

	return list_entry(prev_cws->list.next, struct cpu_workqueue_stats,
			  list);
}

static int workqueue_stat_show(struct seq_file *s, void *p)
{
	struct cpu_workqueue_stats *cws = p;
	unsigned long flags;
	int cpu = cws->cpu;

	seq_printf(s, "%3d %6d     %6u       %s\n", cws->cpu,
		   atomic_read(&cws->inserted),
		   cws->executed,
		   trace_find_cmdline(cws->pid));

	spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
	if (&cws->list == all_workqueue_stat[cpu].list.next)
		seq_printf(s, "\n");
	spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);

	return 0;
}

static int workqueue_stat_headers(struct seq_file *s)
{
	seq_printf(s, "# CPU  INSERTED  EXECUTED   NAME\n");
	seq_printf(s, "# |      |         |          |\n\n");
	return 0;
}

struct tracer_stat workqueue_stats __read_mostly = {
	.name = "workqueues",
	.stat_start = workqueue_stat_start,
	.stat_next = workqueue_stat_next,
	.stat_show = workqueue_stat_show,
	.stat_headers = workqueue_stat_headers
};


int __init stat_workqueue_init(void)
{
	if (register_stat_tracer(&workqueue_stats)) {
		pr_warning("Unable to register workqueue stat tracer\n");
		return 1;
	}

	return 0;
}
fs_initcall(stat_workqueue_init);

/*
 * Workqueues are created very early, just after pre-smp initcalls.
 * So we must register our tracepoints at this stage.
 */
int __init trace_workqueue_early_init(void)
{
	int ret, cpu;

	ret = register_trace_workqueue_insertion(probe_workqueue_insertion);
	if (ret)
		goto out;

	ret = register_trace_workqueue_execution(probe_workqueue_execution);
	if (ret)
		goto no_insertion;

	ret = register_trace_workqueue_creation(probe_workqueue_creation);
	if (ret)
		goto no_execution;

	ret = register_trace_workqueue_destruction(probe_workqueue_destruction);
	if (ret)
		goto no_creation;

	all_workqueue_stat = kmalloc(sizeof(struct workqueue_global_stats)
				     * num_possible_cpus(), GFP_KERNEL);

	if (!all_workqueue_stat) {
		pr_warning("trace_workqueue: not enough memory\n");
		goto no_creation;
	}

	for_each_possible_cpu(cpu) {
		spin_lock_init(&all_workqueue_stat[cpu].lock);
		INIT_LIST_HEAD(&all_workqueue_stat[cpu].list);
	}

	return 0;

no_creation:
	unregister_trace_workqueue_creation(probe_workqueue_creation);
no_execution:
	unregister_trace_workqueue_execution(probe_workqueue_execution);
no_insertion:
	unregister_trace_workqueue_insertion(probe_workqueue_insertion);
out:
	pr_warning("trace_workqueue: unable to trace workqueues\n");

	return 1;
}
early_initcall(trace_workqueue_early_init);