[PATCH v4 10/17] watchdog/hardlockup: Move perf hardlockup watchdog petting to watchdog.c

Douglas Anderson dianders at chromium.org
Thu May 4 15:13:42 PDT 2023


In preparation for the buddy hardlockup detector, which wants the same
petting logic as the current perf hardlockup detector, move the code
to watchdog.c. While doing this, rename the global variable to match
others nearby. The arch_touch_nmi_watchdog() function is not renamed
since that is exported with "EXPORT_SYMBOL" and is thus ABI.

Currently the code in watchdog.c is guarded by
CONFIG_HARDLOCKUP_DETECTOR_PERF, which makes this change seem
silly. However, a future patch will change this.

NOTE: there is a slight side effect to this change, though from code
analysis I believe it to be a beneficial one. Specifically the perf
hardlockup detector will now do check the "timestamp" before clearing
any watchdog pets. Previously the order was reversed. With the old
order if the watchdog perf event was firing super fast then it would
also be clearing existing watchdog pets super fast. The new behavior
of handling super-fast perf before clearing watchdog pets seems
better.

Signed-off-by: Douglas Anderson <dianders at chromium.org>
---

Changes in v4:
- ("Move perf hardlockup watchdog petting to watchdog.c") new for v4.

 include/linux/nmi.h    |  5 +++--
 kernel/watchdog.c      | 19 +++++++++++++++++++
 kernel/watchdog_perf.c | 19 -------------------
 3 files changed, 22 insertions(+), 21 deletions(-)

diff --git a/include/linux/nmi.h b/include/linux/nmi.h
index 2c9ea1ba285c..94ff84e1c8ec 100644
--- a/include/linux/nmi.h
+++ b/include/linux/nmi.h
@@ -88,7 +88,10 @@ static inline void hardlockup_detector_disable(void) {}
 #endif
 
 #if defined(CONFIG_HARDLOCKUP_DETECTOR_PERF)
+void arch_touch_nmi_watchdog(void);
 void watchdog_hardlockup_check(unsigned int cpu, struct pt_regs *regs);
+#elif !defined(CONFIG_HAVE_NMI_WATCHDOG)
+static inline void arch_touch_nmi_watchdog(void) { }
 #endif
 
 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
@@ -98,7 +101,6 @@ void watchdog_hardlockup_check(unsigned int cpu, struct pt_regs *regs);
 #endif
 
 #if defined(CONFIG_HARDLOCKUP_DETECTOR_PERF)
-extern void arch_touch_nmi_watchdog(void);
 extern void hardlockup_detector_perf_stop(void);
 extern void hardlockup_detector_perf_restart(void);
 extern void hardlockup_detector_perf_disable(void);
@@ -113,7 +115,6 @@ static inline void hardlockup_detector_perf_enable(void) { }
 static inline void hardlockup_detector_perf_cleanup(void) { }
 # if !defined(CONFIG_HAVE_NMI_WATCHDOG)
 static inline int hardlockup_detector_perf_init(void) { return -ENODEV; }
-static inline void arch_touch_nmi_watchdog(void) {}
 # else
 static inline int hardlockup_detector_perf_init(void) { return 0; }
 # endif
diff --git a/kernel/watchdog.c b/kernel/watchdog.c
index 367bea0167a5..9c17090611f2 100644
--- a/kernel/watchdog.c
+++ b/kernel/watchdog.c
@@ -90,8 +90,22 @@ __setup("nmi_watchdog=", hardlockup_panic_setup);
 static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts);
 static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved);
 static DEFINE_PER_CPU(bool, watchdog_hardlockup_processed);
+static DEFINE_PER_CPU(bool, watchdog_hardlockup_touch);
 static unsigned long watchdog_hardlockup_dumped_stacks;
 
+notrace void arch_touch_nmi_watchdog(void)
+{
+	/*
+	 * Using __raw here because some code paths have
+	 * preemption enabled.  If preemption is enabled
+	 * then interrupts should be enabled too, in which
+	 * case we shouldn't have to worry about the watchdog
+	 * going off.
+	 */
+	raw_cpu_write(watchdog_hardlockup_touch, true);
+}
+EXPORT_SYMBOL(arch_touch_nmi_watchdog);
+
 static bool watchdog_hardlockup_is_lockedup(unsigned int cpu)
 {
 	unsigned long hrint = per_cpu(hrtimer_interrupts, cpu);
@@ -111,6 +125,11 @@ static void watchdog_hardlockup_interrupt_count(void)
 
 void watchdog_hardlockup_check(unsigned int cpu, struct pt_regs *regs)
 {
+	if (__this_cpu_read(watchdog_hardlockup_touch)) {
+		__this_cpu_write(watchdog_hardlockup_touch, false);
+		return;
+	}
+
 	/*
 	 * Check for a hardlockup by making sure the CPU's timer
 	 * interrupt is incrementing. The timer interrupt should have
diff --git a/kernel/watchdog_perf.c b/kernel/watchdog_perf.c
index 9be90b2a2ea7..547917ebd5d3 100644
--- a/kernel/watchdog_perf.c
+++ b/kernel/watchdog_perf.c
@@ -20,26 +20,12 @@
 #include <asm/irq_regs.h>
 #include <linux/perf_event.h>
 
-static DEFINE_PER_CPU(bool, watchdog_nmi_touch);
 static DEFINE_PER_CPU(struct perf_event *, watchdog_ev);
 static DEFINE_PER_CPU(struct perf_event *, dead_event);
 static struct cpumask dead_events_mask;
 
 static atomic_t watchdog_cpus = ATOMIC_INIT(0);
 
-notrace void arch_touch_nmi_watchdog(void)
-{
-	/*
-	 * Using __raw here because some code paths have
-	 * preemption enabled.  If preemption is enabled
-	 * then interrupts should be enabled too, in which
-	 * case we shouldn't have to worry about the watchdog
-	 * going off.
-	 */
-	raw_cpu_write(watchdog_nmi_touch, true);
-}
-EXPORT_SYMBOL(arch_touch_nmi_watchdog);
-
 #ifdef CONFIG_HARDLOCKUP_CHECK_TIMESTAMP
 static DEFINE_PER_CPU(ktime_t, last_timestamp);
 static DEFINE_PER_CPU(unsigned int, nmi_rearmed);
@@ -112,11 +98,6 @@ static void watchdog_overflow_callback(struct perf_event *event,
 	/* Ensure the watchdog never gets throttled */
 	event->hw.interrupts = 0;
 
-	if (__this_cpu_read(watchdog_nmi_touch) == true) {
-		__this_cpu_write(watchdog_nmi_touch, false);
-		return;
-	}
-
 	if (!watchdog_check_timestamp())
 		return;
 
-- 
2.40.1.521.gf1e218fcd8-goog




More information about the linux-arm-kernel mailing list