Blame view
kernel/latencytop.c
7.23 KB
b886d83c5 treewide: Replace... |
1 |
// SPDX-License-Identifier: GPL-2.0-only |
9745512ce sched: latencytop... |
2 3 4 5 6 |
/* * latencytop.c: Latency display infrastructure * * (C) Copyright 2008 Intel Corporation * Author: Arjan van de Ven <arjan@linux.intel.com> |
9745512ce sched: latencytop... |
7 |
*/ |
ad0b0fd55 sched, latencytop... |
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 |
/* * CONFIG_LATENCYTOP enables a kernel latency tracking infrastructure that is * used by the "latencytop" userspace tool. The latency that is tracked is not * the 'traditional' interrupt latency (which is primarily caused by something * else consuming CPU), but instead, it is the latency an application encounters * because the kernel sleeps on its behalf for various reasons. * * This code tracks 2 levels of statistics: * 1) System level latency * 2) Per process latency * * The latency is stored in fixed sized data structures in an accumulated form; * if the "same" latency cause is hit twice, this will be tracked as one entry * in the data structure. Both the count, total accumulated latency and maximum * latency are tracked in this data structure. When the fixed size structure is * full, no new causes are tracked until the buffer is flushed by writing to * the /proc file; the userspace tool does this on a regular basis. * * A latency cause is identified by a stringified backtrace at the point that * the scheduler gets invoked. The userland tool will use this string to * identify the cause of the latency in human readable form. * * The information is exported via /proc/latency_stats and /proc/<pid>/latency. * These files look like this: * * Latency Top version : v0.1 * 70 59433 4897 i915_irq_wait drm_ioctl vfs_ioctl do_vfs_ioctl sys_ioctl * | | | | * | | | +----> the stringified backtrace * | | +---------> The maximum latency for this entry in microseconds * | +--------------> The accumulated latency for this entry (microseconds) * +-------------------> The number of times this entry is hit * * (note: the average latency is the accumulated latency divided by the number * of times) */ |
9745512ce sched: latencytop... |
45 46 47 48 49 |
#include <linux/kallsyms.h> #include <linux/seq_file.h> #include <linux/notifier.h> #include <linux/spinlock.h> #include <linux/proc_fs.h> |
cb2517653 sched/debug: Make... |
50 |
#include <linux/latencytop.h> |
9984de1a5 kernel: Map most ... |
51 |
#include <linux/export.h> |
9745512ce sched: latencytop... |
52 |
#include <linux/sched.h> |
b17b01533 sched/headers: Pr... |
53 |
#include <linux/sched/debug.h> |
3905f9ad4 sched/headers: Pr... |
54 |
#include <linux/sched/stat.h> |
9745512ce sched: latencytop... |
55 |
#include <linux/list.h> |
9745512ce sched: latencytop... |
56 |
#include <linux/stacktrace.h> |
757455d41 locking, latencyt... |
57 |
static DEFINE_RAW_SPINLOCK(latency_lock); |
9745512ce sched: latencytop... |
58 59 60 61 62 |
#define MAXLR 128 static struct latency_record latency_record[MAXLR]; int latencytop_enabled; |
e02c9b0d6 kernel/latencytop... |
63 |
void clear_tsk_latency_tracing(struct task_struct *p) |
9745512ce sched: latencytop... |
64 65 |
{ unsigned long flags; |
757455d41 locking, latencyt... |
66 |
raw_spin_lock_irqsave(&latency_lock, flags); |
9745512ce sched: latencytop... |
67 68 |
memset(&p->latency_record, 0, sizeof(p->latency_record)); p->latency_record_count = 0; |
757455d41 locking, latencyt... |
69 |
raw_spin_unlock_irqrestore(&latency_lock, flags); |
9745512ce sched: latencytop... |
70 71 72 73 74 |
} static void clear_global_latency_tracing(void) { unsigned long flags; |
757455d41 locking, latencyt... |
75 |
raw_spin_lock_irqsave(&latency_lock, flags); |
9745512ce sched: latencytop... |
76 |
memset(&latency_record, 0, sizeof(latency_record)); |
757455d41 locking, latencyt... |
77 |
raw_spin_unlock_irqrestore(&latency_lock, flags); |
9745512ce sched: latencytop... |
78 79 80 |
} static void __sched |
eaa1809b9 kernel/latencytop... |
81 82 |
account_global_scheduler_latency(struct task_struct *tsk, struct latency_record *lat) |
9745512ce sched: latencytop... |
83 84 85 |
{ int firstnonnull = MAXLR + 1; int i; |
9745512ce sched: latencytop... |
86 87 88 89 90 |
/* skip kernel threads for now */ if (!tsk->mm) return; for (i = 0; i < MAXLR; i++) { |
19fb518c2 latencytop: optim... |
91 |
int q, same = 1; |
9745512ce sched: latencytop... |
92 93 94 95 96 97 |
/* Nothing stored: */ if (!latency_record[i].backtrace[0]) { if (firstnonnull > i) firstnonnull = i; continue; } |
ad0b0fd55 sched, latencytop... |
98 |
for (q = 0; q < LT_BACKTRACEDEPTH; q++) { |
19fb518c2 latencytop: optim... |
99 100 101 |
unsigned long record = lat->backtrace[q]; if (latency_record[i].backtrace[q] != record) { |
9745512ce sched: latencytop... |
102 |
same = 0; |
9745512ce sched: latencytop... |
103 |
break; |
19fb518c2 latencytop: optim... |
104 |
} |
accddc41b latency_top: Remo... |
105 106 |
/* 0 entry marks end of backtrace: */ if (!record) |
9745512ce sched: latencytop... |
107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 |
break; } if (same) { latency_record[i].count++; latency_record[i].time += lat->time; if (lat->time > latency_record[i].max) latency_record[i].max = lat->time; return; } } i = firstnonnull; if (i >= MAXLR - 1) return; /* Allocted a new one: */ memcpy(&latency_record[i], lat, sizeof(struct latency_record)); } |
ad0b0fd55 sched, latencytop... |
125 |
/** |
25985edce Fix common misspe... |
126 |
* __account_scheduler_latency - record an occurred latency |
ad0b0fd55 sched, latencytop... |
127 128 129 130 131 132 133 134 135 136 137 138 139 140 |
* @tsk - the task struct of the task hitting the latency * @usecs - the duration of the latency in microseconds * @inter - 1 if the sleep was interruptible, 0 if uninterruptible * * This function is the main entry point for recording latency entries * as called by the scheduler. * * This function has a few special cases to deal with normal 'non-latency' * sleeps: specifically, interruptible sleep longer than 5 msec is skipped * since this usually is caused by waiting for events via select() and co. * * Negative latencies (caused by time going backwards) are also explicitly * skipped. */ |
9745512ce sched: latencytop... |
141 |
void __sched |
ad0b0fd55 sched, latencytop... |
142 |
__account_scheduler_latency(struct task_struct *tsk, int usecs, int inter) |
9745512ce sched: latencytop... |
143 144 145 146 |
{ unsigned long flags; int i, q; struct latency_record lat; |
9745512ce sched: latencytop... |
147 148 149 |
/* Long interruptible waits are generally user requested... */ if (inter && usecs > 5000) return; |
ad0b0fd55 sched, latencytop... |
150 151 152 153 |
/* Negative sleeps are time going backwards */ /* Zero-time sleeps are non-interesting */ if (usecs <= 0) return; |
9745512ce sched: latencytop... |
154 155 156 157 |
memset(&lat, 0, sizeof(lat)); lat.count = 1; lat.time = usecs; lat.max = usecs; |
f93877214 latency_top: Simp... |
158 159 |
stack_trace_save_tsk(tsk, lat.backtrace, LT_BACKTRACEDEPTH, 0); |
9745512ce sched: latencytop... |
160 |
|
757455d41 locking, latencyt... |
161 |
raw_spin_lock_irqsave(&latency_lock, flags); |
9745512ce sched: latencytop... |
162 163 |
account_global_scheduler_latency(tsk, &lat); |
38715258a latencytop: fix p... |
164 |
for (i = 0; i < tsk->latency_record_count; i++) { |
9745512ce sched: latencytop... |
165 166 |
struct latency_record *mylat; int same = 1; |
19fb518c2 latencytop: optim... |
167 |
|
9745512ce sched: latencytop... |
168 |
mylat = &tsk->latency_record[i]; |
ad0b0fd55 sched, latencytop... |
169 |
for (q = 0; q < LT_BACKTRACEDEPTH; q++) { |
19fb518c2 latencytop: optim... |
170 171 172 |
unsigned long record = lat.backtrace[q]; if (mylat->backtrace[q] != record) { |
9745512ce sched: latencytop... |
173 |
same = 0; |
9745512ce sched: latencytop... |
174 |
break; |
19fb518c2 latencytop: optim... |
175 |
} |
accddc41b latency_top: Remo... |
176 177 |
/* 0 entry is end of backtrace */ if (!record) |
9745512ce sched: latencytop... |
178 179 180 181 182 183 184 185 186 187 |
break; } if (same) { mylat->count++; mylat->time += lat.time; if (lat.time > mylat->max) mylat->max = lat.time; goto out_unlock; } } |
38715258a latencytop: fix p... |
188 189 190 191 192 |
/* * short term hack; if we're > 32 we stop; future we recycle: */ if (tsk->latency_record_count >= LT_SAVECOUNT) goto out_unlock; |
9745512ce sched: latencytop... |
193 |
/* Allocated a new one: */ |
38715258a latencytop: fix p... |
194 |
i = tsk->latency_record_count++; |
9745512ce sched: latencytop... |
195 196 197 |
memcpy(&tsk->latency_record[i], &lat, sizeof(struct latency_record)); out_unlock: |
757455d41 locking, latencyt... |
198 |
raw_spin_unlock_irqrestore(&latency_lock, flags); |
9745512ce sched: latencytop... |
199 200 201 202 203 204 205 206 207 208 |
} static int lstats_show(struct seq_file *m, void *v) { int i; seq_puts(m, "Latency Top version : v0.1 "); for (i = 0; i < MAXLR; i++) { |
34e49d4f6 fs/proc/base.c, k... |
209 210 211 |
struct latency_record *lr = &latency_record[i]; if (lr->backtrace[0]) { |
9745512ce sched: latencytop... |
212 |
int q; |
34e49d4f6 fs/proc/base.c, k... |
213 214 |
seq_printf(m, "%i %lu %lu", lr->count, lr->time, lr->max); |
9745512ce sched: latencytop... |
215 |
for (q = 0; q < LT_BACKTRACEDEPTH; q++) { |
34e49d4f6 fs/proc/base.c, k... |
216 |
unsigned long bt = lr->backtrace[q]; |
accddc41b latency_top: Remo... |
217 |
|
34e49d4f6 fs/proc/base.c, k... |
218 |
if (!bt) |
9745512ce sched: latencytop... |
219 |
break; |
accddc41b latency_top: Remo... |
220 |
|
34e49d4f6 fs/proc/base.c, k... |
221 |
seq_printf(m, " %ps", (void *)bt); |
9745512ce sched: latencytop... |
222 |
} |
eaa1809b9 kernel/latencytop... |
223 224 |
seq_puts(m, " "); |
9745512ce sched: latencytop... |
225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 |
} } return 0; } static ssize_t lstats_write(struct file *file, const char __user *buf, size_t count, loff_t *offs) { clear_global_latency_tracing(); return count; } static int lstats_open(struct inode *inode, struct file *filp) { return single_open(filp, lstats_show, NULL); } |
97a32539b proc: convert eve... |
243 244 245 246 247 248 |
static const struct proc_ops lstats_proc_ops = { .proc_open = lstats_open, .proc_read = seq_read, .proc_write = lstats_write, .proc_lseek = seq_lseek, .proc_release = single_release, |
9745512ce sched: latencytop... |
249 250 251 252 |
}; static int __init init_lstats_procfs(void) { |
97a32539b proc: convert eve... |
253 |
proc_create("latency_stats", 0644, NULL, &lstats_proc_ops); |
9745512ce sched: latencytop... |
254 255 |
return 0; } |
cb2517653 sched/debug: Make... |
256 |
|
32927393d sysctl: pass kern... |
257 258 |
int sysctl_latencytop(struct ctl_table *table, int write, void *buffer, size_t *lenp, loff_t *ppos) |
cb2517653 sched/debug: Make... |
259 260 261 262 263 264 265 266 267 |
{ int err; err = proc_dointvec(table, write, buffer, lenp, ppos); if (latencytop_enabled) force_schedstat_enabled(); return err; } |
ad0b0fd55 sched, latencytop... |
268 |
device_initcall(init_lstats_procfs); |