aboutsummaryrefslogtreecommitdiff
path: root/sys/i386/i386/mp_watchdog.c
blob: b6ac777316643a6df00a3a9c0a83e3997f2772f7 (plain) (blame)
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
/*-
 * Copyright (c) 2004 Robert N. M. Watson
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 * $FreeBSD$
 */

#include "opt_mp_watchdog.h"
#include "opt_sched.h"

#ifdef SCHED_ULE
#error MP_WATCHDOG cannot currently be used with SCHED_ULE
#endif

#include <sys/param.h>
#include <sys/kdb.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/pcpu.h>
#include <sys/proc.h>
#include <sys/sysctl.h>
#include <sys/systm.h>

#include <machine/smp.h>
#include <machine/apicreg.h>
#include <machine/apicvar.h>
#include <machine/mp_watchdog.h>

/*
 * mp_watchdog hijacks the idle thread on a specified CPU, prevents new work
 * from being scheduled there, and uses it as a "watchdog" to detect kernel
 * failure on other CPUs.  This is made reasonable by inclusion of logical
 * processors in Xeon hardware.  The watchdog is configured by setting the
 * debug.watchdog sysctl/tunable to the CPU of interest.  A callout will then
 * begin executing reseting a timer that is gradually lowered by the watching
 * thread.  If the timer reaches 0, the watchdog fires by ether dropping
 * directly to the debugger, or by sending an NMI IPI to the boot processor.
 * This is a somewhat less efficient substitute for dedicated watchdog
 * hardware, but can be quite an effective tool for debugging hangs.
 *
 * XXXRW: This should really use the watchdog(9)/watchdog(4) framework, but
 * doesn't yet.
 */
static int	watchdog_cpu = -1;
static int	watchdog_dontfire = 1;
static int	watchdog_timer = -1;
static int	watchdog_nmi = 1;

TUNABLE_INT("debug.watchdog", &watchdog_cpu);
SYSCTL_INT(_debug, OID_AUTO, watchdog_nmi, CTLFLAG_RW, &watchdog_nmi, 0,
    "IPI the boot processor with an NMI to enter the debugger");

static struct callout	watchdog_callout;

static void watchdog_change(int wdcpu);

/*
 * Number of seconds before the watchdog will fire if the callout fails to
 * reset the timer.
 */
#define	WATCHDOG_THRESHOLD	10

static void
watchdog_init(void *arg)
{

	callout_init(&watchdog_callout, CALLOUT_MPSAFE);
	if (watchdog_cpu != -1)
		watchdog_change(watchdog_cpu);
}

/*
 * This callout resets a timer until the watchdog kicks in.  It acquires some
 * critical locks to make sure things haven't gotten wedged with hose locks
 * held.
 */
static void
watchdog_function(void *arg)
{

	/*
	 * Since the timer ran, we must not be wedged.  Acquire some critical
	 * locks to make sure.  Then reset the timer.
	 */
	mtx_lock(&Giant);
	watchdog_timer = WATCHDOG_THRESHOLD;
	mtx_unlock(&Giant);
	callout_reset(&watchdog_callout, 1 * hz, watchdog_function, NULL);
}
SYSINIT(watchdog_init, SI_SUB_DRIVERS, SI_ORDER_ANY, watchdog_init, NULL);

static void
watchdog_change(int wdcpu)
{

	if (wdcpu == -1 || wdcpu == 0xffffffff) {
		/*
		 * Disable the watchdog.
		 */
		watchdog_cpu = -1;
		watchdog_dontfire = 1;
		callout_stop(&watchdog_callout);
		printf("watchdog stopped\n");
	} else {
		watchdog_timer = WATCHDOG_THRESHOLD;
		watchdog_dontfire = 0;
		watchdog_cpu = wdcpu;
		callout_reset(&watchdog_callout, 1 * hz, watchdog_function,
		    NULL);
	}
}

/*
 * This sysctl sets which CPU is the watchdog CPU.  Set to -1 or 0xffffffff
 * to disable the watchdog.
 */
static int
sysctl_watchdog(SYSCTL_HANDLER_ARGS)
{
	int error, temp;

	temp = watchdog_cpu;
	error = sysctl_handle_int(oidp, &temp, 0, req);
	if (error)
		return (error);

	if (req->newptr != NULL)
		watchdog_change(temp);
	return (0);
}
SYSCTL_PROC(_debug, OID_AUTO, watchdog, CTLTYPE_INT|CTLFLAG_RW, 0, 0,
    sysctl_watchdog, "I", "");

/*
 * Drop into the debugger by sending an IPI NMI to the boot processor.
 */
static void
watchdog_ipi_nmi(void)
{

	/*
	 * Deliver NMI to the boot processor.  Why not?
	 */
	lapic_ipi_raw(APIC_DEST_DESTFLD | APIC_TRIGMOD_EDGE |
	    APIC_LEVEL_ASSERT | APIC_DESTMODE_PHY | APIC_DELMODE_NMI,
	    boot_cpu_id);
	lapic_ipi_wait(-1);
}

/*
 * ap_watchdog() is called by the SMP idle loop code.  It works on the same
 * premise that the disabling of logical processors does: that if the cpu is
 * idle, then it can ignore the world from then on, as nothing will be
 * scheduled on it.  Leaving aside multi-runqueue schedulers (SCHED_ULE) and
 * explicit process migration (sched_bind()), this is not an unreasonable
 * assumption.
 */
void
ap_watchdog(u_int cpuid)
{
	char old_pcomm[MAXCOMLEN + 1];
	struct proc *p;

	if (watchdog_cpu != cpuid)
		return;

	printf("watchdog started on cpu %d\n", cpuid);
	p = curproc;
	bcopy(p->p_comm, old_pcomm, MAXCOMLEN + 1);
	snprintf(p->p_comm, MAXCOMLEN + 1, "mp_watchdog cpu %d", cpuid);
	while (1) {
		DELAY(1000000);				/* One second. */
		if (watchdog_cpu != cpuid)
			break;
		atomic_subtract_int(&watchdog_timer, 1);
		if (watchdog_timer < 4)
			printf("Watchdog timer: %d\n", watchdog_timer);
		if (watchdog_timer == 0 && watchdog_dontfire == 0) {
			printf("Watchdog firing!\n");
			watchdog_dontfire = 1;
			if (watchdog_nmi)
				watchdog_ipi_nmi();
			else
				kdb_enter("mp_watchdog");
		}
	}
	bcopy(old_pcomm, p->p_comm, MAXCOMLEN + 1);
	printf("watchdog stopped on cpu %d\n", cpuid);
}