Commit 696f9486 authored by Joel Becker's avatar Joel Becker Committed by Linus Torvalds
Browse files

[PATCH] hangcheck-timer: Update to 0.9.0. 



I recently realized that the in-kernel copy of hangcheck-timer was quite
stale.  Here's the latest.  It adds support for s390, ppc64, and ia64 too.

Signed-off-by: default avatarJoel Becker <joel.becker@oracle.com>
Signed-off-by: default avatarAndrew Morton <akpm@osdl.org>
Signed-off-by: default avatarLinus Torvalds <torvalds@osdl.org>
parent a40920b4

drivers/char/Kconfig

+1 −1
Original line number Diff line number Diff line
@@ -982,7 +982,7 @@ config MAX_RAW_DEVS 


config HANGCHECK_TIMER

	tristate "Hangcheck timer"

	depends on X86_64 || X86

	depends on X86_64 || X86 || IA64 || PPC64 || ARCH_S390

	help

	  The hangcheck-timer module detects when the system has gone

	  out to lunch past a certain margin.  It can reboot the system

drivers/char/hangcheck-timer.c

+95 −9
Original line number Diff line number Diff line
@@ -3,7 +3,7 @@ 
 *

 * Driver for a little io fencing timer.

 *

 * Copyright (C) 2002 Oracle Corporation.  All rights reserved.

 * Copyright (C) 2002, 2003 Oracle.  All rights reserved.

 *

 * Author: Joel Becker <joel.becker@oracle.com>

 *
@@ -44,11 +44,14 @@ 
#include <linux/fs.h>

#include <linux/mm.h>

#include <linux/reboot.h>

#include <linux/smp_lock.h>

#include <linux/init.h>

#include <linux/delay.h>

#include <asm/uaccess.h>

#include <linux/sysrq.h>





#define VERSION_STR "0.5.0"

#define VERSION_STR "0.9.0"



#define DEFAULT_IOFENCE_MARGIN 60	/* Default fudge factor, in seconds */

#define DEFAULT_IOFENCE_TICK 180	/* Default timer timeout, in seconds */
@@ -56,18 +59,89 @@ 
static int hangcheck_tick = DEFAULT_IOFENCE_TICK;

static int hangcheck_margin = DEFAULT_IOFENCE_MARGIN;

static int hangcheck_reboot;  /* Defaults to not reboot */

static int hangcheck_dump_tasks;  /* Defaults to not dumping SysRQ T */



/* Driver options */

/* options - modular */

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.");

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.");



MODULE_AUTHOR("Joel Becker");

MODULE_AUTHOR("Oracle");

MODULE_DESCRIPTION("Hangcheck-timer detects when the system has gone out to lunch past a certain margin.");

MODULE_LICENSE("GPL");

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 */



#if defined(CONFIG_X86) || defined(CONFIG_X86_64)

# define HAVE_MONOTONIC

# define TIMER_FREQ 1000000000ULL

#elif defined(CONFIG_ARCH_S390)

/* FA240000 is 1 Second in the IBM time universe (Page 4-38 Principles of Op for zSeries */

# define TIMER_FREQ 0xFA240000ULL

#elif defined(CONFIG_IA64)

# define TIMER_FREQ ((unsigned long long)local_cpu_data->itc_freq)

#elif defined(CONFIG_PPC64)

# define TIMER_FREQ (HZ*loops_per_jiffy)

#endif



#ifdef HAVE_MONOTONIC

extern unsigned long long monotonic_clock(void);

#else

static inline unsigned long long monotonic_clock(void)

{

# ifdef __s390__

	/* returns the TOD.  see 4-38 Principles of Op of zSeries */

	return get_clock();

# else

	return get_cycles();

# endif  /* __s390__ */

}

#endif  /* HAVE_MONOTONIC */





/* Last time scheduled */
@@ -78,7 +152,6 @@ static void hangcheck_fire(unsigned long); 
static struct timer_list hangcheck_ticktock =

		TIMER_INITIALIZER(hangcheck_fire, 0, 0);



extern unsigned long long monotonic_clock(void);



static void hangcheck_fire(unsigned long data)

{
@@ -92,6 +165,12 @@ static void hangcheck_fire(unsigned long data) 
		tsc_diff = (cur_tsc + (~0ULL - hangcheck_tsc)); /* or something */



	if (tsc_diff > hangcheck_tsc_margin) {

		if (hangcheck_dump_tasks) {

			printk(KERN_CRIT "Hangcheck: Task state:\n");

#ifdef CONFIG_MAGIC_SYSRQ

			handle_sysrq('t', NULL, NULL);

#endif  /* CONFIG_MAGIC_SYSRQ */

		}

		if (hangcheck_reboot) {

			printk(KERN_CRIT "Hangcheck: hangcheck is restarting the machine.\n");

			machine_restart(NULL);
@@ -108,10 +187,16 @@ static int __init hangcheck_init(void) 
{

	printk("Hangcheck: starting hangcheck timer %s (tick is %d seconds, margin is %d seconds).\n",

	       VERSION_STR, hangcheck_tick, hangcheck_margin);



	hangcheck_tsc_margin = hangcheck_margin + hangcheck_tick;

	hangcheck_tsc_margin *= 1000000000;



#if defined (HAVE_MONOTONIC)

	printk("Hangcheck: Using monotonic_clock().\n");

#elif defined(__s390__)

	printk("Hangcheck: Using TOD.\n");

#else

	printk("Hangcheck: Using get_cycles().\n");

#endif  /* HAVE_MONOTONIC */

	hangcheck_tsc_margin =

		(unsigned long long)(hangcheck_margin + hangcheck_tick);

	hangcheck_tsc_margin *= (unsigned long long)TIMER_FREQ;



	hangcheck_tsc = monotonic_clock();

	mod_timer(&hangcheck_ticktock, jiffies + (hangcheck_tick*HZ));
@@ -123,6 +208,7 @@ static int __init hangcheck_init(void) 
static void __exit hangcheck_exit(void)

{

	del_timer_sync(&hangcheck_ticktock);

        printk("Hangcheck: Stopped hangcheck timer.\n");

}



module_init(hangcheck_init);