Orange Pi5 kernel

^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   1) // SPDX-License-Identifier: GPL-2.0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   2) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   3)  * Detect hard lockups on a system
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   4)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   5)  * started by Don Zickus, Copyright (C) 2010 Red Hat, Inc.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   6)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   7)  * Note: Most of this code is borrowed heavily from the original softlockup
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   8)  * detector, so thanks to Ingo for the initial implementation.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   9)  * Some chunks also taken from the old x86-specific nmi watchdog code, thanks
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  10)  * to those contributors as well.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  11)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  12) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  13) #define pr_fmt(fmt) "NMI watchdog: " fmt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  14) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  15) #include <linux/nmi.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  16) #include <linux/atomic.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  17) #include <linux/module.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  18) #include <linux/sched/debug.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  19) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  20) #include <asm/irq_regs.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  21) #include <linux/perf_event.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  22) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  23) static DEFINE_PER_CPU(bool, hard_watchdog_warn);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  24) static DEFINE_PER_CPU(bool, watchdog_nmi_touch);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  25) static DEFINE_PER_CPU(struct perf_event *, watchdog_ev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  26) static DEFINE_PER_CPU(struct perf_event *, dead_event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  27) static struct cpumask dead_events_mask;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  28) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  29) static unsigned long hardlockup_allcpu_dumped;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  30) static atomic_t watchdog_cpus = ATOMIC_INIT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  31) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  32) notrace void arch_touch_nmi_watchdog(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  33) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  34) 	/*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  35) 	 * Using __raw here because some code paths have
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  36) 	 * preemption enabled.  If preemption is enabled
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  37) 	 * then interrupts should be enabled too, in which
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  38) 	 * case we shouldn't have to worry about the watchdog
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  39) 	 * going off.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  40) 	 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  41) 	raw_cpu_write(watchdog_nmi_touch, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  42) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  43) EXPORT_SYMBOL(arch_touch_nmi_watchdog);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  44) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  45) #ifdef CONFIG_HARDLOCKUP_CHECK_TIMESTAMP
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  46) static DEFINE_PER_CPU(ktime_t, last_timestamp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  47) static DEFINE_PER_CPU(unsigned int, nmi_rearmed);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  48) static ktime_t watchdog_hrtimer_sample_threshold __read_mostly;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  49) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  50) void watchdog_update_hrtimer_threshold(u64 period)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  51) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  52) 	/*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  53) 	 * The hrtimer runs with a period of (watchdog_threshold * 2) / 5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  54) 	 *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  55) 	 * So it runs effectively with 2.5 times the rate of the NMI
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  56) 	 * watchdog. That means the hrtimer should fire 2-3 times before
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  57) 	 * the NMI watchdog expires. The NMI watchdog on x86 is based on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  58) 	 * unhalted CPU cycles, so if Turbo-Mode is enabled the CPU cycles
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  59) 	 * might run way faster than expected and the NMI fires in a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  60) 	 * smaller period than the one deduced from the nominal CPU
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  61) 	 * frequency. Depending on the Turbo-Mode factor this might be fast
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  62) 	 * enough to get the NMI period smaller than the hrtimer watchdog
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  63) 	 * period and trigger false positives.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  64) 	 *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  65) 	 * The sample threshold is used to check in the NMI handler whether
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  66) 	 * the minimum time between two NMI samples has elapsed. That
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  67) 	 * prevents false positives.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  68) 	 *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  69) 	 * Set this to 4/5 of the actual watchdog threshold period so the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  70) 	 * hrtimer is guaranteed to fire at least once within the real
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  71) 	 * watchdog threshold.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  72) 	 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  73) 	watchdog_hrtimer_sample_threshold = period * 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  74) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  75) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  76) static bool watchdog_check_timestamp(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  77) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  78) 	ktime_t delta, now = ktime_get_mono_fast_ns();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  79) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  80) 	delta = now - __this_cpu_read(last_timestamp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  81) 	if (delta < watchdog_hrtimer_sample_threshold) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  82) 		/*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  83) 		 * If ktime is jiffies based, a stalled timer would prevent
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  84) 		 * jiffies from being incremented and the filter would look
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  85) 		 * at a stale timestamp and never trigger.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  86) 		 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  87) 		if (__this_cpu_inc_return(nmi_rearmed) < 10)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  88) 			return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  89) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  90) 	__this_cpu_write(nmi_rearmed, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  91) 	__this_cpu_write(last_timestamp, now);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  92) 	return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  93) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  94) #else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  95) static inline bool watchdog_check_timestamp(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  96) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  97) 	return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  98) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  99) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) static struct perf_event_attr wd_hw_attr = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) 	.type		= PERF_TYPE_HARDWARE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) 	.config		= PERF_COUNT_HW_CPU_CYCLES,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) 	.size		= sizeof(struct perf_event_attr),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105) 	.pinned		= 1,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) 	.disabled	= 1,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109) /* Callback function for perf event subsystem */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) static void watchdog_overflow_callback(struct perf_event *event,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) 				       struct perf_sample_data *data,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112) 				       struct pt_regs *regs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) 	/* Ensure the watchdog never gets throttled */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) 	event->hw.interrupts = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) 	if (__this_cpu_read(watchdog_nmi_touch) == true) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) 		__this_cpu_write(watchdog_nmi_touch, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119) 		return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122) 	if (!watchdog_check_timestamp())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) 		return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125) 	/* check for a hardlockup
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) 	 * This is done by making sure our timer interrupt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) 	 * is incrementing.  The timer interrupt should have
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) 	 * fired multiple times before we overflow'd.  If it hasn't
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) 	 * then this is a good indication the cpu is stuck
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) 	 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131) 	if (is_hardlockup()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132) 		int this_cpu = smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134) 		/* only print hardlockups once */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) 		if (__this_cpu_read(hard_watchdog_warn) == true)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136) 			return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) 		pr_emerg("Watchdog detected hard LOCKUP on cpu %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) 			 this_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140) 		print_modules();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141) 		print_irqtrace_events(current);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) 		if (regs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) 			show_regs(regs);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) 		else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145) 			dump_stack();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) 		/*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) 		 * Perform all-CPU dump only once to avoid multiple hardlockups
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149) 		 * generating interleaving traces
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150) 		 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) 		if (sysctl_hardlockup_all_cpu_backtrace &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) 				!test_and_set_bit(0, &hardlockup_allcpu_dumped))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153) 			trigger_allbutself_cpu_backtrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) 		if (hardlockup_panic)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156) 			nmi_panic(regs, "Hard LOCKUP");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) 		__this_cpu_write(hard_watchdog_warn, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) 		return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162) 	__this_cpu_write(hard_watchdog_warn, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) 	return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166) static int hardlockup_detector_event_create(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) 	unsigned int cpu = smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169) 	struct perf_event_attr *wd_attr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170) 	struct perf_event *evt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) 	wd_attr = &wd_hw_attr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173) 	wd_attr->sample_period = hw_nmi_get_sample_period(watchdog_thresh);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175) 	/* Try to register using hardware perf events */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176) 	evt = perf_event_create_kernel_counter(wd_attr, cpu, NULL,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) 					       watchdog_overflow_callback, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) 	if (IS_ERR(evt)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) 		pr_debug("Perf event create on CPU %d failed with %ld\n", cpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) 			 PTR_ERR(evt));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181) 		return PTR_ERR(evt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183) 	this_cpu_write(watchdog_ev, evt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) 	return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188)  * hardlockup_detector_perf_enable - Enable the local event
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190) void hardlockup_detector_perf_enable(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) 	if (hardlockup_detector_event_create())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) 		return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195) 	/* use original value for check */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) 	if (!atomic_fetch_inc(&watchdog_cpus))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) 		pr_info("Enabled. Permanently consumes one hw-PMU counter.\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) 	perf_event_enable(this_cpu_read(watchdog_ev));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203)  * hardlockup_detector_perf_disable - Disable the local event
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205) void hardlockup_detector_perf_disable(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207) 	struct perf_event *event = this_cpu_read(watchdog_ev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) 	if (event) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210) 		perf_event_disable(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211) 		this_cpu_write(watchdog_ev, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212) 		this_cpu_write(dead_event, event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213) 		cpumask_set_cpu(smp_processor_id(), &dead_events_mask);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214) 		atomic_dec(&watchdog_cpus);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219)  * hardlockup_detector_perf_cleanup - Cleanup disabled events and destroy them
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221)  * Called from lockup_detector_cleanup(). Serialized by the caller.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223) void hardlockup_detector_perf_cleanup(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225) 	int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227) 	for_each_cpu(cpu, &dead_events_mask) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228) 		struct perf_event *event = per_cpu(dead_event, cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) 		/*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231) 		 * Required because for_each_cpu() reports  unconditionally
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) 		 * CPU0 as set on UP kernels. Sigh.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) 		 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) 		if (event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) 			perf_event_release_kernel(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236) 		per_cpu(dead_event, cpu) = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) 	cpumask_clear(&dead_events_mask);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242)  * hardlockup_detector_perf_stop - Globally stop watchdog events
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244)  * Special interface for x86 to handle the perf HT bug.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) void __init hardlockup_detector_perf_stop(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248) 	int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250) 	lockdep_assert_cpus_held();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 251) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 252) 	for_each_online_cpu(cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 253) 		struct perf_event *event = per_cpu(watchdog_ev, cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255) 		if (event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) 			perf_event_disable(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261)  * hardlockup_detector_perf_restart - Globally restart watchdog events
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263)  * Special interface for x86 to handle the perf HT bug.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) void __init hardlockup_detector_perf_restart(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267) 	int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269) 	lockdep_assert_cpus_held();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) 	if (!(watchdog_enabled & NMI_WATCHDOG_ENABLED))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272) 		return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) 	for_each_online_cpu(cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275) 		struct perf_event *event = per_cpu(watchdog_ev, cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) 		if (event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278) 			perf_event_enable(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283)  * hardlockup_detector_perf_init - Probe whether NMI event is available at all
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285) int __init hardlockup_detector_perf_init(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287) 	int ret = hardlockup_detector_event_create();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) 	if (ret) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290) 		pr_info("Perf NMI watchdog permanently disabled\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291) 	} else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292) 		perf_event_release_kernel(this_cpu_read(watchdog_ev));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) 		this_cpu_write(watchdog_ev, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295) 	return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296) }