Blame view
drivers/char/hangcheck-timer.c
5.09 KB
921a3d4d3
|
1 |
// SPDX-License-Identifier: GPL-2.0-only |
1da177e4c
|
2 3 4 5 6 |
/* * hangcheck-timer.c * * Driver for a little io fencing timer. * |
696f9486d
|
7 |
* Copyright (C) 2002, 2003 Oracle. All rights reserved. |
1da177e4c
|
8 9 |
* * Author: Joel Becker <joel.becker@oracle.com> |
1da177e4c
|
10 11 12 13 14 15 |
*/ /* * The hangcheck-timer driver uses the TSC to catch delays that * jiffies does not notice. A timer is set. When the timer fires, it * checks whether it was delayed and if that delay exceeds a given |
8dfba4d71
|
16 |
* margin of error. The hangcheck_tick module parameter takes the timer |
1da177e4c
|
17 18 19 20 21 |
* duration in seconds. The hangcheck_margin parameter defines the * margin of error, in seconds. The defaults are 60 seconds for the * timer and 180 seconds for the margin of error. IOW, a timer is set * for 60 seconds. When the timer fires, the callback checks the * actual duration that the timer waited. If the duration exceeds the |
8b932edfb
|
22 |
* allotted time and margin (here 60 + 180, or 240 seconds), the machine |
1da177e4c
|
23 24 25 26 27 28 29 30 31 32 33 34 |
* is restarted. A healthy machine will have the duration match the * expected timeout very closely. */ #include <linux/module.h> #include <linux/moduleparam.h> #include <linux/types.h> #include <linux/kernel.h> #include <linux/fs.h> #include <linux/mm.h> #include <linux/reboot.h> #include <linux/init.h> |
696f9486d
|
35 |
#include <linux/delay.h> |
7c0f6ba68
|
36 |
#include <linux/uaccess.h> |
696f9486d
|
37 |
#include <linux/sysrq.h> |
e8edc6e03
|
38 |
#include <linux/timer.h> |
2044fdb03
|
39 |
#include <linux/hrtimer.h> |
1da177e4c
|
40 |
|
940370fc8
|
41 |
#define VERSION_STR "0.9.1" |
1da177e4c
|
42 43 44 45 46 47 48 |
#define DEFAULT_IOFENCE_MARGIN 60 /* Default fudge factor, in seconds */ #define DEFAULT_IOFENCE_TICK 180 /* Default timer timeout, in seconds */ static int hangcheck_tick = DEFAULT_IOFENCE_TICK; static int hangcheck_margin = DEFAULT_IOFENCE_MARGIN; static int hangcheck_reboot; /* Defaults to not reboot */ |
696f9486d
|
49 |
static int hangcheck_dump_tasks; /* Defaults to not dumping SysRQ T */ |
1da177e4c
|
50 |
|
696f9486d
|
51 |
/* options - modular */ |
1da177e4c
|
52 53 54 55 56 57 |
module_param(hangcheck_tick, int, 0); MODULE_PARM_DESC(hangcheck_tick, "Timer delay."); module_param(hangcheck_margin, int, 0); MODULE_PARM_DESC(hangcheck_margin, "If the hangcheck timer has been delayed more than hangcheck_margin seconds, the driver will fire."); module_param(hangcheck_reboot, int, 0); MODULE_PARM_DESC(hangcheck_reboot, "If nonzero, the machine will reboot when the timer margin is exceeded."); |
696f9486d
|
58 59 |
module_param(hangcheck_dump_tasks, int, 0); MODULE_PARM_DESC(hangcheck_dump_tasks, "If nonzero, the machine will dump the system task state when the timer margin is exceeded."); |
1da177e4c
|
60 |
|
696f9486d
|
61 |
MODULE_AUTHOR("Oracle"); |
1da177e4c
|
62 63 |
MODULE_DESCRIPTION("Hangcheck-timer detects when the system has gone out to lunch past a certain margin."); MODULE_LICENSE("GPL"); |
696f9486d
|
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 |
MODULE_VERSION(VERSION_STR); /* options - nonmodular */ #ifndef MODULE static int __init hangcheck_parse_tick(char *str) { int par; if (get_option(&str,&par)) hangcheck_tick = par; return 1; } static int __init hangcheck_parse_margin(char *str) { int par; if (get_option(&str,&par)) hangcheck_margin = par; return 1; } static int __init hangcheck_parse_reboot(char *str) { int par; if (get_option(&str,&par)) hangcheck_reboot = par; return 1; } static int __init hangcheck_parse_dump_tasks(char *str) { int par; if (get_option(&str,&par)) hangcheck_dump_tasks = par; return 1; } __setup("hcheck_tick", hangcheck_parse_tick); __setup("hcheck_margin", hangcheck_parse_margin); __setup("hcheck_reboot", hangcheck_parse_reboot); __setup("hcheck_dump_tasks", hangcheck_parse_dump_tasks); #endif /* not MODULE */ |
2044fdb03
|
106 |
#define TIMER_FREQ 1000000000ULL |
1da177e4c
|
107 108 109 |
/* Last time scheduled */ static unsigned long long hangcheck_tsc, hangcheck_tsc_margin; |
24ed960ab
|
110 |
static void hangcheck_fire(struct timer_list *); |
1da177e4c
|
111 |
|
1d27e3e22
|
112 |
static DEFINE_TIMER(hangcheck_ticktock, hangcheck_fire); |
1da177e4c
|
113 |
|
24ed960ab
|
114 |
static void hangcheck_fire(struct timer_list *unused) |
1da177e4c
|
115 116 |
{ unsigned long long cur_tsc, tsc_diff; |
2044fdb03
|
117 |
cur_tsc = ktime_get_ns(); |
1da177e4c
|
118 119 120 121 122 123 124 |
if (cur_tsc > hangcheck_tsc) tsc_diff = cur_tsc - hangcheck_tsc; else tsc_diff = (cur_tsc + (~0ULL - hangcheck_tsc)); /* or something */ if (tsc_diff > hangcheck_tsc_margin) { |
696f9486d
|
125 126 127 128 |
if (hangcheck_dump_tasks) { printk(KERN_CRIT "Hangcheck: Task state: "); #ifdef CONFIG_MAGIC_SYSRQ |
f335397d1
|
129 |
handle_sysrq('t'); |
696f9486d
|
130 131 |
#endif /* CONFIG_MAGIC_SYSRQ */ } |
1da177e4c
|
132 133 134 |
if (hangcheck_reboot) { printk(KERN_CRIT "Hangcheck: hangcheck is restarting the machine. "); |
970d32443
|
135 |
emergency_restart(); |
1da177e4c
|
136 137 138 139 140 |
} else { printk(KERN_CRIT "Hangcheck: hangcheck value past margin! "); } } |
940370fc8
|
141 142 143 144 145 146 147 148 |
#if 0 /* * Enable to investigate delays in detail */ printk("Hangcheck: called %Ld ns since last time (%Ld ns overshoot) ", tsc_diff, tsc_diff - hangcheck_tick*TIMER_FREQ); #endif |
1da177e4c
|
149 |
mod_timer(&hangcheck_ticktock, jiffies + (hangcheck_tick*HZ)); |
2044fdb03
|
150 |
hangcheck_tsc = ktime_get_ns(); |
1da177e4c
|
151 152 153 154 155 156 157 158 |
} static int __init hangcheck_init(void) { printk("Hangcheck: starting hangcheck timer %s (tick is %d seconds, margin is %d seconds). ", VERSION_STR, hangcheck_tick, hangcheck_margin); |
696f9486d
|
159 |
hangcheck_tsc_margin = |
d0439a544
|
160 161 |
(unsigned long long)hangcheck_margin + hangcheck_tick; hangcheck_tsc_margin *= TIMER_FREQ; |
1da177e4c
|
162 |
|
2044fdb03
|
163 |
hangcheck_tsc = ktime_get_ns(); |
1da177e4c
|
164 165 166 167 168 169 170 171 172 |
mod_timer(&hangcheck_ticktock, jiffies + (hangcheck_tick*HZ)); return 0; } static void __exit hangcheck_exit(void) { del_timer_sync(&hangcheck_ticktock); |
696f9486d
|
173 174 |
printk("Hangcheck: Stopped hangcheck timer. "); |
1da177e4c
|
175 176 177 178 |
} module_init(hangcheck_init); module_exit(hangcheck_exit); |