[PATCH 5/5] arm64/watchdog_hld: enable hard lockup on arm64 platform

@@ -193,6 +193,9 @@ config ARM64
 	select HAVE_PERF_EVENTS
 	select HAVE_PERF_REGS
 	select HAVE_PERF_USER_STACK_DUMP
+	select HAVE_PERF_EVENTS_NMI
+	select HAVE_HARDLOCKUP_DETECTOR_PERF \
+		if PERF_EVENTS && HW_PERF_EVENTS && HAVE_PERF_EVENTS_NMI
 	select HAVE_REGS_AND_STACK_ACCESS_API
 	select HAVE_FUNCTION_ARG_ACCESS_API
 	select HAVE_FUTEX_CMPXCHG if FUTEX

@@ -6,6 +6,7 @@
 #ifndef __ASM_PERF_EVENT_H
 #define __ASM_PERF_EVENT_H
 
+#include <linux/wait.h>
 #include <asm/stack_pointer.h>
 #include <asm/ptrace.h>
 

@@ -259,4 +260,8 @@ extern unsigned long perf_misc_flags(struct pt_regs *regs);
 	(regs)->pstate = PSR_MODE_EL1h;	\
 }
 
+extern bool arm_pmu_initialized;
+extern wait_queue_head_t arm_pmu_wait;
+extern bool check_pmu_nmi_ability(void);
+
 #endif

@@ -46,6 +46,7 @@ obj-$(CONFIG_MODULES)			+= module.o
 obj-$(CONFIG_ARM64_MODULE_PLTS)		+= module-plts.o
 obj-$(CONFIG_PERF_EVENTS)		+= perf_regs.o perf_callchain.o
 obj-$(CONFIG_HW_PERF_EVENTS)		+= perf_event.o
+obj-$(CONFIG_HARDLOCKUP_DETECTOR_PERF)	+= watchdog_hld.o
 obj-$(CONFIG_HAVE_HW_BREAKPOINT)	+= hw_breakpoint.o
 obj-$(CONFIG_CPU_PM)			+= sleep.o suspend.o
 obj-$(CONFIG_CPU_IDLE)			+= cpuidle.o

@@ -24,6 +24,9 @@
 #include <linux/sched_clock.h>
 #include <linux/smp.h>
 
+bool arm_pmu_initialized;
+DECLARE_WAIT_QUEUE_HEAD(arm_pmu_wait);
+
 /* ARMv8 Cortex-A53 specific event types. */
 #define ARMV8_A53_PERFCTR_PREF_LINEFILL				0xC2
 

@@ -1284,10 +1287,17 @@ static struct platform_driver armv8_pmu_driver = {
 
 static int __init armv8_pmu_driver_init(void)
 {
+	int ret;
+
 	if (acpi_disabled)
-		return platform_driver_register(&armv8_pmu_driver);
+		ret = platform_driver_register(&armv8_pmu_driver);
 	else
-		return arm_pmu_acpi_probe(armv8_pmuv3_init);
+		ret = arm_pmu_acpi_probe(armv8_pmuv3_init);
+
+	arm_pmu_initialized = true;
+	wake_up_all(&arm_pmu_wait);
+
+	return ret;
 }
 device_initcall(armv8_pmu_driver_init)
 

@@ -0,0 +1,83 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/kthread.h>
+#include <linux/sched.h>
+#include <linux/nmi.h>
+#include <linux/cpufreq.h>
+#include <asm/perf_event.h>
+
+/*
+ * Safe maximum CPU frequency in case a particular platform doesn't implement
+ * cpufreq driver. Although, architecture doesn't put any restrictions on
+ * maximum frequency but 5 GHz seems to be safe maximum given the available
+ * Arm CPUs in the market which are clocked much less than 5 GHz. On the other
+ * hand, we can't make it much higher as it would lead to a large hard-lockup
+ * detection timeout on parts which are running slower (eg. 1GHz on
+ * Developerbox) and doesn't possess a cpufreq driver.
+ */
+#define SAFE_MAX_CPU_FREQ	5000000000UL // 5 GHz
+u64 hw_nmi_get_sample_period(int watchdog_thresh)
+{
+	unsigned int cpu = smp_processor_id();
+	unsigned long max_cpu_freq;
+
+	max_cpu_freq = cpufreq_get_hw_max_freq(cpu) * 1000UL;
+	if (!max_cpu_freq)
+		max_cpu_freq = SAFE_MAX_CPU_FREQ;
+
+	return (u64)max_cpu_freq * watchdog_thresh;
+}
+
+static watchdog_nmi_status_reporter status_reporter;
+
+static int hld_enabled_thread_fun(void *unused)
+{
+	struct watchdog_nmi_status status;
+	watchdog_nmi_status_reporter local_reporter;
+	int ret;
+
+	wait_event(arm_pmu_wait, arm_pmu_initialized);
+	status.cpu = raw_smp_processor_id();
+
+	if (!check_pmu_nmi_ability()) {
+		status.status = -ENODEV;
+		goto report;
+	}
+
+	ret = hardlockup_detector_perf_enable();
+	status.status = ret;
+
+report:
+	local_reporter = (watchdog_nmi_status_reporter)atomic64_fetch_and(0,
+				(atomic64_t *)&status_reporter);
+	if (local_reporter)
+		(*local_reporter)(&status);
+
+	return 0;
+}
+
+/* As for watchdog_nmi_disable(), using the default implement */
+void watchdog_nmi_enable(unsigned int cpu)
+{
+	struct task_struct *t;
+
+	/* PMU is not ready */
+	if (!arm_pmu_initialized) {
+		t = kthread_create_on_cpu(hld_enabled_thread_fun, NULL, cpu,
+			       "arm64_hld.%u");
+		if (IS_ERR(t))
+			return;
+		wake_up_process(t);
+		return;
+	}
+
+	/* For hotplug in cpu */
+	hardlockup_detector_perf_enable();
+}
+
+int __init watchdog_nmi_probe(watchdog_nmi_status_reporter notifier)
+{
+	/* On arm64, arm pmu is not ready at this stage */
+	status_reporter = notifier;
+	return -EBUSY;
+}
+

@@ -105,6 +105,11 @@ static DEFINE_PER_CPU(const struct pmu_irq_ops *, cpu_irq_ops);
 
 static bool has_nmi;
 
+bool check_pmu_nmi_ability(void)
+{
+	return has_nmi;
+}
+
 static inline u64 arm_pmu_event_max_period(struct perf_event *event)
 {
 	if (event->hw.flags & ARMPMU_EVT_64BIT)