[patch 4/4] ipmi: add new IPMI nmi watchdog handling

From: Corey Minyard
Date: Wed Feb 14 2007 - 15:13:19 EST


Convert over to the new NMI handling for getting IPMI watchdog
timeouts via an NMI. This add config options to know if there
is the ability to receive NMIs and if it has an NMI post processing
call. Then it modifies the IPMI watchdog to take advantage of
this so that it can know if an NMI comes in.

It also adds testing that the IPMI NMI watchdog works.

Signed-off-by: Corey Minyard <minyard@xxxxxxx>

Index: linux-2.6.19-rc6/arch/i386/Kconfig.debug
===================================================================
--- linux-2.6.19-rc6.orig/arch/i386/Kconfig.debug
+++ linux-2.6.19-rc6/arch/i386/Kconfig.debug
@@ -4,6 +4,10 @@ config TRACE_IRQFLAGS_SUPPORT
bool
default y

+config HAVE_STANDARD_NOTIFY_DIE
+ bool
+ default y
+
source "lib/Kconfig.debug"

config EARLY_PRINTK
Index: linux-2.6.19-rc6/arch/x86_64/Kconfig.debug
===================================================================
--- linux-2.6.19-rc6.orig/arch/x86_64/Kconfig.debug
+++ linux-2.6.19-rc6/arch/x86_64/Kconfig.debug
@@ -4,6 +4,10 @@ config TRACE_IRQFLAGS_SUPPORT
bool
default y

+config HAVE_STANDARD_NOTIFY_DIE
+ bool
+ default y
+
source "lib/Kconfig.debug"

config DEBUG_RODATA
Index: linux-2.6.19-rc6/arch/i386/kernel/traps.c
===================================================================
--- linux-2.6.19-rc6.orig/arch/i386/kernel/traps.c
+++ linux-2.6.19-rc6/arch/i386/kernel/traps.c
@@ -804,6 +804,11 @@ static __kprobes void default_do_nmi(str
*/
if (nmi_watchdog_tick(regs, reason))
return;
+#endif
+ if (notify_die(DIE_NMI_POST, "nmi_post", regs, reason, 2, 0)
+ == NOTIFY_STOP)
+ return;
+#ifdef CONFIG_X86_LOCAL_APIC
if (!do_nmi_callback(regs, smp_processor_id()))
#endif
unknown_nmi_error(reason, regs);
Index: linux-2.6.19-rc6/arch/x86_64/kernel/traps.c
===================================================================
--- linux-2.6.19-rc6.orig/arch/x86_64/kernel/traps.c
+++ linux-2.6.19-rc6/arch/x86_64/kernel/traps.c
@@ -835,6 +835,8 @@ asmlinkage __kprobes void default_do_nmi
*/
if (nmi_watchdog_tick(regs,reason))
return;
+ if (notify_die(DIE_NMI_POST, "nmi_post", regs, reason, 2, 0)
+ == NOTIFY_STOP)
if (!do_nmi_callback(regs,cpu))
unknown_nmi_error(reason, regs);

Index: linux-2.6.19-rc6/include/asm-i386/kdebug.h
===================================================================
--- linux-2.6.19-rc6.orig/include/asm-i386/kdebug.h
+++ linux-2.6.19-rc6/include/asm-i386/kdebug.h
@@ -38,9 +38,12 @@ enum die_val {
DIE_GPF,
DIE_CALL,
DIE_NMI_IPI,
+ DIE_NMI_POST,
DIE_PAGE_FAULT,
};

+#define HAVE_DIE_NMI_POST
+
static inline int notify_die(enum die_val val, const char *str,
struct pt_regs *regs, long err, int trap, int sig)
{
Index: linux-2.6.19-rc6/include/asm-x86_64/kdebug.h
===================================================================
--- linux-2.6.19-rc6.orig/include/asm-x86_64/kdebug.h
+++ linux-2.6.19-rc6/include/asm-x86_64/kdebug.h
@@ -33,9 +33,12 @@ enum die_val {
DIE_GPF,
DIE_CALL,
DIE_NMI_IPI,
+ DIE_NMI_POST,
DIE_PAGE_FAULT,
};

+#define HAVE_DIE_NMI_POST
+
static inline int notify_die(enum die_val val, const char *str,
struct pt_regs *regs, long err, int trap, int sig)
{
Index: linux-2.6.19/drivers/char/ipmi/ipmi_watchdog.c
===================================================================
--- linux-2.6.19.orig/drivers/char/ipmi/ipmi_watchdog.c 2006-12-30 12:40:48.000000000 -0600
+++ linux-2.6.19/drivers/char/ipmi/ipmi_watchdog.c 2006-12-30 12:57:12.000000000 -0600
@@ -49,9 +49,10 @@
#include <linux/poll.h>
#include <linux/string.h>
#include <linux/ctype.h>
+#include <linux/delay.h>
#include <asm/atomic.h>
-#ifdef CONFIG_X86_LOCAL_APIC
-#include <asm/apic.h>
+#ifdef CONFIG_HAVE_STANDARD_NOTIFY_DIE
+#include <asm/kdebug.h>
#endif

#define PFX "IPMI Watchdog: "
@@ -317,6 +318,11 @@
/* If a pretimeout occurs, this is used to allow only one panic to happen. */
static atomic_t preop_panic_excl = ATOMIC_INIT(-1);

+#ifdef HAVE_DIE_NMI_POST
+static int testing_nmi;
+static int nmi_handler_registered;
+#endif
+
static int ipmi_heartbeat(void);
static void panic_halt_ipmi_heartbeat(void);

@@ -919,6 +925,45 @@
printk(KERN_CRIT PFX "Unable to register misc device\n");
}

+#ifdef HAVE_DIE_NMI_POST
+ if (nmi_handler_registered) {
+ int old_pretimeout = pretimeout;
+ int old_timeout = timeout;
+ int old_preop_val = preop_val;
+
+ /* Set the pretimeout to go off in a second and give
+ ourselves plenty of time to stop the timer. */
+ ipmi_watchdog_state = WDOG_TIMEOUT_RESET;
+ preop_val = WDOG_PREOP_NONE; /* Make sure nothing happens */
+ pretimeout = 99;
+ timeout = 100;
+
+ testing_nmi = 1;
+
+ rv = ipmi_set_timeout(IPMI_SET_TIMEOUT_FORCE_HB);
+ if (rv) {
+ printk(KERN_WARNING PFX "Error starting timer to"
+ " test NMI: 0x%x. The NMI pretimeout will"
+ " likely not work\n", rv);
+ rv = 0;
+ goto out_restore;
+ }
+
+ msleep(1500);
+
+ if (testing_nmi != 2) {
+ printk(KERN_WARNING PFX "IPMI NMI didn't seem to"
+ " occur. The NMI pretimeout will"
+ " likely not work\n");
+ }
+ out_restore:
+ testing_nmi = 0;
+ preop_val = old_preop_val;
+ pretimeout = old_pretimeout;
+ timeout = old_timeout;
+ }
+#endif
+
out:
up_write(&register_sem);

@@ -928,6 +973,10 @@
ipmi_watchdog_state = action_val;
ipmi_set_timeout(IPMI_SET_TIMEOUT_FORCE_HB);
printk(KERN_INFO PFX "Starting now!\n");
+ } else {
+ /* Stop the timer now. */
+ ipmi_watchdog_state = WDOG_TIMEOUT_NONE;
+ ipmi_set_timeout(IPMI_SET_TIMEOUT_NO_HB);
}
}

@@ -964,17 +1013,28 @@
up_write(&register_sem);
}

-#ifdef HAVE_NMI_HANDLER
+#ifdef HAVE_DIE_NMI_POST
static int
-ipmi_nmi(void *dev_id, int cpu, int handled)
+ipmi_nmi(struct notifier_block *self, unsigned long val, void *data)
{
+ if (val != DIE_NMI_POST)
+ return NOTIFY_OK;
+
+ if (testing_nmi) {
+ testing_nmi = 2;
+ return NOTIFY_STOP;
+ }
+
/* If we are not expecting a timeout, ignore it. */
if (ipmi_watchdog_state == WDOG_TIMEOUT_NONE)
- return NOTIFY_DONE;
+ return NOTIFY_OK;
+
+ if (preaction_val != WDOG_PRETIMEOUT_NMI)
+ return NOTIFY_OK;

/* If no one else handled the NMI, we assume it was the IPMI
watchdog. */
- if ((!handled) && (preop_val == WDOG_PREOP_PANIC)) {
+ if (preop_val == WDOG_PREOP_PANIC) {
/* On some machines, the heartbeat will give
an error and not work unless we re-enable
the timer. So do so. */
@@ -983,18 +1043,12 @@
panic(PFX "pre-timeout");
}

- return NOTIFY_DONE;
+ return NOTIFY_STOP;
}

-static struct nmi_handler ipmi_nmi_handler =
-{
- .link = LIST_HEAD_INIT(ipmi_nmi_handler.link),
- .dev_name = "ipmi_watchdog",
- .dev_id = NULL,
- .handler = ipmi_nmi,
- .priority = 0, /* Call us last. */
+static struct notifier_block ipmi_nmi_handler = {
+ .notifier_call = ipmi_nmi
};
-int nmi_handler_registered;
#endif

static int wdog_reboot_handler(struct notifier_block *this,
@@ -1111,7 +1165,7 @@
preaction_val = WDOG_PRETIMEOUT_NONE;
else if (strcmp(inval, "pre_smi") == 0)
preaction_val = WDOG_PRETIMEOUT_SMI;
-#ifdef HAVE_NMI_HANDLER
+#ifdef HAVE_DIE_NMI_POST
else if (strcmp(inval, "pre_nmi") == 0)
preaction_val = WDOG_PRETIMEOUT_NMI;
#endif
@@ -1145,7 +1199,7 @@

static void check_parms(void)
{
-#ifdef HAVE_NMI_HANDLER
+#ifdef HAVE_DIE_NMI_POST
int do_nmi = 0;
int rv;

@@ -1158,20 +1212,9 @@
preop_op("preop_none", NULL);
do_nmi = 0;
}
-#ifdef CONFIG_X86_LOCAL_APIC
- if (nmi_watchdog == NMI_IO_APIC) {
- printk(KERN_WARNING PFX "nmi_watchdog is set to IO APIC"
- " mode (value is %d), that is incompatible"
- " with using NMI in the IPMI watchdog."
- " Disabling IPMI nmi pretimeout.\n",
- nmi_watchdog);
- preaction_val = WDOG_PRETIMEOUT_NONE;
- do_nmi = 0;
- }
-#endif
}
if (do_nmi && !nmi_handler_registered) {
- rv = request_nmi(&ipmi_nmi_handler);
+ rv = register_die_notifier(&ipmi_nmi_handler);
if (rv) {
printk(KERN_WARNING PFX
"Can't register nmi handler\n");
@@ -1179,7 +1222,7 @@
} else
nmi_handler_registered = 1;
} else if (!do_nmi && nmi_handler_registered) {
- release_nmi(&ipmi_nmi_handler);
+ unregister_die_notifier(&ipmi_nmi_handler);
nmi_handler_registered = 0;
}
#endif
@@ -1215,9 +1258,9 @@

rv = ipmi_smi_watcher_register(&smi_watcher);
if (rv) {
-#ifdef HAVE_NMI_HANDLER
- if (preaction_val == WDOG_PRETIMEOUT_NMI)
- release_nmi(&ipmi_nmi_handler);
+#ifdef HAVE_DIE_NMI_POST
+ if (nmi_handler_registered)
+ unregister_die_notifier(&ipmi_nmi_handler);
#endif
atomic_notifier_chain_unregister(&panic_notifier_list,
&wdog_panic_notifier);
@@ -1236,9 +1279,9 @@
ipmi_smi_watcher_unregister(&smi_watcher);
ipmi_unregister_watchdog(watchdog_ifnum);

-#ifdef HAVE_NMI_HANDLER
+#ifdef HAVE_DIE_NMI_POST
if (nmi_handler_registered)
- release_nmi(&ipmi_nmi_handler);
+ unregister_die_notifier(&ipmi_nmi_handler);
#endif

atomic_notifier_chain_unregister(&panic_notifier_list,
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/