[PATCH 2/2] hpet: Add HPET-based NMI watchdog support

Alexander Graf posted 2 patches 4 days, 10 hours ago
[PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by Alexander Graf 4 days, 10 hours ago
The traditional NMI watchdog timer uses performance counters to trigger
periodic NMIs. But performance counters are a scarce resource that are
best used for actual performance counting. However, the HPET is another
timer source on most modern x86 systems that can inject NMI interrupts.

Add support for using HPET timer as NMI watchdog source instead of
performance counters. This frees up a PMC for profiling use.

Unlike with the PMU based watchdog where we trigger a per-CPU NMI, APIC
based interrupt descriptors can only target either a specific CPU or
perform a broadcast on all CPUs. To not run into races and allow for
CPU hotplug, the NMI watchdog switches between CPU 0 and broadcast modes
based on whether all CPUs are up or not.

The HPET watchdog always uses IO-APIC line 2. This line is
architecturally defined as the PIT source on PCs and hence always
available as long as we disable the PIT (which we do). We could in
theory try to find a vacant GSI line, but in practice that would create
a big dependency chain on ACPI which I would rather avoid for now.

The implementation uses the standard HARDLOCKUP_DETECTOR_ARCH
infrastructure, following the same pattern as powerpc's arch-specific
hardlockup detector.

With this watchdog present, I can successfully capture system lockups,
verified by adding "local_irq_disable(); while(1) {}" into
mount_root_generic().

(Disclaimer: Some of this code was written with the help of Kiro, an AI
coding assistant)

Signed-off-by: Alexander Graf <graf@amazon.com>
---
 .../admin-guide/kernel-parameters.txt         |   5 +-
 arch/x86/Kconfig                              |  19 ++
 arch/x86/kernel/hpet.c                        | 208 ++++++++++++++++++
 arch/x86/kernel/i8253.c                       |   9 +
 drivers/char/hpet.c                           |   3 +
 include/linux/hpet.h                          |  14 ++
 6 files changed, 257 insertions(+), 1 deletion(-)

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 1058f2a6d6a8..c6a98812a896 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -2045,11 +2045,14 @@ Kernel parameters
 
 	hpet=		[X86-32,HPET] option to control HPET usage
 			Format: { enable (default) | disable | force |
-				verbose }
+				verbose | watchdog }
 			disable: disable HPET and use PIT instead
 			force: allow force enabled of undocumented chips (ICH4,
 				VIA, nVidia)
 			verbose: show contents of HPET registers during setup
+			watchdog: use HPET timer as NMI watchdog source instead
+				of performance counters. Use nmi_watchdog=1 to enable
+				or nmi_watchdog=panic to panic on hard lockup detection.
 
 	hpet_mmap=	[X86, HPET_MMAP] Allow userspace to mmap HPET
 			registers.  Default set by CONFIG_HPET_MMAP_DEFAULT.
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 80527299f859..e8873218a803 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -948,6 +948,25 @@ config HPET_EMULATE_RTC
 	def_bool y
 	depends on HPET_TIMER && (RTC_DRV_CMOS=m || RTC_DRV_CMOS=y)
 
+config HARDLOCKUP_DETECTOR_HPET
+	bool "Use HPET for NMI watchdog"
+	depends on HPET_TIMER
+	select HAVE_HARDLOCKUP_DETECTOR_ARCH
+	select HARDLOCKUP_DETECTOR_COUNTS_HRTIMER
+	help
+	  Use HPET timer as NMI source instead of performance counters.
+	  This frees up a performance counter for profiling.
+	  Enable with hpet=watchdog kernel parameter.
+
+config HARDLOCKUP_DETECTOR_HPET_DEFAULT
+	bool "Enable HPET watchdog by default"
+	depends on HARDLOCKUP_DETECTOR_HPET
+	help
+	  Say Y here to enable HPET-based NMI watchdog by default without
+	  requiring the hpet=watchdog kernel parameter.
+
+	  If unsure, say N.
+
 # Mark as expert because too many people got it wrong.
 # The code disables itself when not needed.
 config DMI
diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c
index d6387dde3ff9..c9114997c383 100644
--- a/arch/x86/kernel/hpet.c
+++ b/arch/x86/kernel/hpet.c
@@ -6,9 +6,12 @@
 #include <linux/hpet.h>
 #include <linux/cpu.h>
 #include <linux/irq.h>
+#include <linux/nmi.h>
+#include <linux/syscore_ops.h>
 
 #include <asm/cpuid/api.h>
 #include <asm/irq_remapping.h>
+#include <asm/io_apic.h>
 #include <asm/hpet.h>
 #include <asm/time.h>
 #include <asm/mwait.h>
@@ -100,6 +103,8 @@ static inline void hpet_clear_mapping(void)
 /*
  * HPET command line enable / disable
  */
+static bool hpet_watchdog_mode = IS_ENABLED(CONFIG_HARDLOCKUP_DETECTOR_HPET_DEFAULT);
+
 static int __init hpet_setup(char *str)
 {
 	while (str) {
@@ -113,6 +118,8 @@ static int __init hpet_setup(char *str)
 			hpet_force_user = true;
 		if (!strncmp("verbose", str, 7))
 			hpet_verbose = true;
+		if (!strncmp("watchdog", str, 8))
+			hpet_watchdog_mode = true;
 		str = next;
 	}
 	return 1;
@@ -985,6 +992,200 @@ static bool __init hpet_is_pc10_damaged(void)
 	return true;
 }
 
+#ifdef CONFIG_HARDLOCKUP_DETECTOR_HPET
+/*
+ * HPET watchdog uses timer 0 routed to GSI 2 (legacy PIT IRQ line).
+ * When using HPET as watchdog, we repurpose this line for NMI delivery.
+ */
+#define HPET_WD_TIMER	0
+#define HPET_WD_GSI	2
+
+bool hpet_watchdog_initialized;
+static bool hpet_watchdog_ioapic_configured;
+static DEFINE_PER_CPU(u32, hpet_watchdog_next_tick);
+
+static int hpet_nmi_handler(unsigned int cmd, struct pt_regs *regs)
+{
+	u32 now, next, delta;
+
+	if (panic_in_progress())
+		return NMI_HANDLED;
+
+	/* Check if this NMI is from our HPET timer by comparing counter value */
+	now = hpet_readl(HPET_COUNTER);
+	next = __this_cpu_read(hpet_watchdog_next_tick);
+	delta = hpet_freq * watchdog_thresh;
+
+	/*
+	 * If we have a next tick set and counter hasn't reached it yet,
+	 * this NMI is not from our timer. Allow some tolerance for timing.
+	 */
+	if (next && (s32)(now - next) < -(s32)(delta / 4))
+		return NMI_DONE;
+
+	/* Update next expected tick */
+	__this_cpu_write(hpet_watchdog_next_tick, now + delta);
+
+	watchdog_hardlockup_check(smp_processor_id(), regs);
+
+	return NMI_HANDLED;
+}
+
+/*
+ * On suspend, clear the configured flag so that the first CPU to come
+ * online after resume will reconfigure the HPET timer and IO-APIC.
+ *
+ * We don't need to explicitly disable the watchdog here because:
+ * 1. The HPET registers are reset by the hibernation/suspend process anyway
+ * 2. The IO-APIC state is saved/restored by ioapic_syscore_ops, but we
+ *    need to reconfigure it for NMI delivery after resume
+ * 3. Secondary CPUs are offlined before suspend, so we can't broadcast
+ *    NMIs until they're back online - the enable callback handles this
+ */
+static int hpet_watchdog_suspend(void *data)
+{
+	hpet_watchdog_ioapic_configured = false;
+	return 0;
+}
+
+static const struct syscore_ops hpet_watchdog_syscore_ops = {
+	.suspend = hpet_watchdog_suspend,
+};
+
+static struct syscore hpet_watchdog_syscore = {
+	.ops = &hpet_watchdog_syscore_ops,
+};
+
+static int __init hpet_watchdog_init(u32 channels)
+{
+	u32 cfg, i, route_cap;
+
+	if (channels <= HPET_WD_TIMER)
+		return 0;
+
+	/* Verify GSI 2 is available in the route capability bitmap */
+	route_cap = hpet_readl(HPET_Tn_CFG(HPET_WD_TIMER) + 4);
+	if (!(route_cap & (1 << HPET_WD_GSI))) {
+		pr_info("HPET timer 0 cannot route to GSI %d\n", HPET_WD_GSI);
+		return 0;
+	}
+
+	/* Deactivate all timers */
+	for (i = 0; i < channels; i++) {
+		cfg = hpet_readl(HPET_Tn_CFG(i));
+		cfg &= ~(HPET_TN_ENABLE | HPET_TN_LEVEL | HPET_TN_FSB);
+		hpet_writel(cfg, HPET_Tn_CFG(i));
+	}
+
+	/* Configure HPET timer for periodic mode */
+	cfg = hpet_readl(HPET_Tn_CFG(HPET_WD_TIMER));
+	cfg &= ~(HPET_TN_ENABLE | HPET_TN_FSB);
+	cfg |= HPET_TN_PERIODIC | HPET_TN_32BIT | HPET_TN_SETVAL | HPET_TN_LEVEL;
+	hpet_writel(cfg, HPET_Tn_CFG(HPET_WD_TIMER));
+
+	/* Route HPET timer to the GSI */
+	cfg = hpet_readl(HPET_Tn_CFG(HPET_WD_TIMER));
+	cfg &= ~(Tn_INT_ROUTE_CNF_MASK | HPET_CFG_ENABLE);
+	cfg |= (HPET_WD_GSI << Tn_INT_ROUTE_CNF_SHIFT) & Tn_INT_ROUTE_CNF_MASK;
+	hpet_writel(cfg, HPET_Tn_CFG(HPET_WD_TIMER));
+
+	if (register_nmi_handler(NMI_LOCAL, hpet_nmi_handler, 0, "hpet_watchdog")) {
+		pr_err("Failed to register NMI_LOCAL handler\n");
+		return 0;
+	}
+	if (register_nmi_handler(NMI_UNKNOWN, hpet_nmi_handler, 0, "hpet_watchdog")) {
+		unregister_nmi_handler(NMI_LOCAL, "hpet_watchdog");
+		pr_err("Failed to register NMI_UNKNOWN handler\n");
+		return 0;
+	}
+
+	hpet_start_counter();
+
+	hpet_watchdog_initialized = true;
+
+	register_syscore(&hpet_watchdog_syscore);
+
+	pr_info("HPET watchdog initialized on timer %d, GSI %d", HPET_WD_TIMER, HPET_WD_GSI);
+
+	return 0;
+}
+
+void watchdog_hardlockup_stop(void)
+{
+	u32 cfg;
+
+	if (!hpet_watchdog_initialized)
+		return;
+
+	cfg = hpet_readl(HPET_Tn_CFG(HPET_WD_TIMER));
+	cfg &= ~HPET_TN_ENABLE;
+	hpet_writel(cfg, HPET_Tn_CFG(HPET_WD_TIMER));
+}
+
+void watchdog_hardlockup_start(void)
+{
+	u32 cfg, delta;
+
+	if (!hpet_watchdog_initialized)
+		return;
+
+	if (!hpet_watchdog_ioapic_configured) {
+		if (ioapic_set_nmi(HPET_WD_GSI, false)) {
+			pr_err("Unable to configure IO-APIC for NMI\n");
+			return;
+		}
+		hpet_watchdog_ioapic_configured = true;
+	}
+
+	delta = hpet_freq * watchdog_thresh;
+
+	cfg = hpet_readl(HPET_Tn_CFG(HPET_WD_TIMER));
+	cfg &= ~(HPET_TN_ENABLE | HPET_TN_FSB | HPET_TN_LEVEL);
+	cfg |= HPET_TN_PERIODIC | HPET_TN_32BIT | HPET_TN_SETVAL;
+	hpet_writel(cfg, HPET_Tn_CFG(HPET_WD_TIMER));
+
+	/* Write twice for AMD 81xx with buggy HPET */
+	hpet_writel(delta, HPET_Tn_CMP(HPET_WD_TIMER));
+	hpet_writel(delta, HPET_Tn_CMP(HPET_WD_TIMER));
+
+	cfg |= HPET_TN_ENABLE;
+	hpet_writel(cfg, HPET_Tn_CFG(HPET_WD_TIMER));
+}
+
+void watchdog_hardlockup_enable(unsigned int cpu)
+{
+	if (!hpet_watchdog_ioapic_configured) {
+		/*
+		 * First CPU online after resume - reconfigure HPET timer.
+		 * This also sets hpet_watchdog_ioapic_configured = true.
+		 */
+		watchdog_hardlockup_start();
+	}
+
+	if (num_online_cpus() == num_present_cpus()) {
+		ioapic_set_nmi(HPET_WD_GSI, true);
+		pr_info("switched to broadcast mode (all %d CPUs online)\n",
+			num_online_cpus());
+	}
+}
+
+void watchdog_hardlockup_disable(unsigned int cpu)
+{
+	if (num_online_cpus() < num_present_cpus()) {
+		ioapic_set_nmi(HPET_WD_GSI, false);
+		pr_info("switched to CPU 0 only (%d CPUs online)\n",
+			num_online_cpus() - 1);
+	}
+}
+
+int __init watchdog_hardlockup_probe(void)
+{
+	return hpet_watchdog_mode ? 0 : -ENODEV;
+}
+#else
+static inline int hpet_watchdog_init(u32 channels) { return 0; }
+#endif /* CONFIG_HARDLOCKUP_DETECTOR_HPET */
+
 /**
  * hpet_enable - Try to setup the HPET timer. Returns 1 on success.
  */
@@ -1031,6 +1232,10 @@ int __init hpet_enable(void)
 	/* This is the HPET channel number which is zero based */
 	channels = ((id & HPET_ID_NUMBER) >> HPET_ID_NUMBER_SHIFT) + 1;
 
+	/* If watchdog mode, hand off to watchdog driver */
+	if (hpet_watchdog_mode)
+		return hpet_watchdog_init(channels);
+
 	/*
 	 * The legacy routing mode needs at least two channels, tick timer
 	 * and the rtc emulation channel.
@@ -1122,6 +1327,9 @@ static __init int hpet_late_init(void)
 {
 	int ret;
 
+	if (hpet_is_watchdog())
+		return -ENODEV;
+
 	if (!hpet_address) {
 		if (!force_hpet_address)
 			return -ENODEV;
diff --git a/arch/x86/kernel/i8253.c b/arch/x86/kernel/i8253.c
index cb9852ad6098..36dd948371a4 100644
--- a/arch/x86/kernel/i8253.c
+++ b/arch/x86/kernel/i8253.c
@@ -7,6 +7,7 @@
 #include <linux/init.h>
 #include <linux/timex.h>
 #include <linux/i8253.h>
+#include <linux/hpet.h>
 
 #include <asm/hypervisor.h>
 #include <asm/apic.h>
@@ -31,6 +32,14 @@ struct clock_event_device *global_clock_event;
  */
 static bool __init use_pit(void)
 {
+	if (hpet_is_watchdog()) {
+		/*
+		 * The PIT overlaps the HPET IRQ line which we configure to
+		 * NMI in watchdog mode, rendering the PIT non functional.
+		 */
+		return false;
+	}
+
 	if (!IS_ENABLED(CONFIG_X86_TSC) || !boot_cpu_has(X86_FEATURE_TSC))
 		return true;
 
diff --git a/drivers/char/hpet.c b/drivers/char/hpet.c
index 4f5ccd3a1f56..9d9e4d22ab7f 100644
--- a/drivers/char/hpet.c
+++ b/drivers/char/hpet.c
@@ -977,6 +977,9 @@ static int hpet_acpi_add(struct acpi_device *device)
 	acpi_status result;
 	struct hpet_data data;
 
+	if (hpet_is_watchdog())
+		return -ENODEV;
+
 	memset(&data, 0, sizeof(data));
 
 	result =
diff --git a/include/linux/hpet.h b/include/linux/hpet.h
index 21e69eaf7a36..408b440163cc 100644
--- a/include/linux/hpet.h
+++ b/include/linux/hpet.h
@@ -108,4 +108,18 @@ static inline void hpet_reserve_timer(struct hpet_data *hd, int timer)
 
 int hpet_alloc(struct hpet_data *);
 
+#ifdef CONFIG_HARDLOCKUP_DETECTOR_HPET
+extern bool hpet_watchdog_initialized;
+
+static inline bool hpet_is_watchdog(void)
+{
+	return hpet_watchdog_initialized;
+}
+#else
+static inline bool hpet_is_watchdog(void)
+{
+	return false;
+}
+#endif
+
 #endif				/* !__HPET__ */
-- 
2.47.1




Amazon Web Services Development Center Germany GmbH
Tamara-Danz-Str. 13
10243 Berlin
Geschaeftsfuehrung: Christof Hellmis, Andreas Stieger
Eingetragen am Amtsgericht Charlottenburg unter HRB 257764 B
Sitz: Berlin
Ust-ID: DE 365 538 597
Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by Thomas Gleixner 3 days, 18 hours ago
On Mon, Feb 02 2026 at 17:48, Alexander Graf wrote:
> (Disclaimer: Some of this code was written with the help of Kiro, an AI
> coding assistant)

You could have sent your change log through AI too so it conforms with
the change log rules ...

> +#ifdef CONFIG_HARDLOCKUP_DETECTOR_HPET
> +/*
> + * HPET watchdog uses timer 0 routed to GSI 2 (legacy PIT IRQ line).
> + * When using HPET as watchdog, we repurpose this line for NMI delivery.
> + */
> +#define HPET_WD_TIMER	0
> +#define HPET_WD_GSI	2
> +
> +bool hpet_watchdog_initialized;
> +static bool hpet_watchdog_ioapic_configured;
> +static DEFINE_PER_CPU(u32, hpet_watchdog_next_tick);
> +
> +static int hpet_nmi_handler(unsigned int cmd, struct pt_regs *regs)
> +{
> +	u32 now, next, delta;
> +
> +	if (panic_in_progress())
> +		return NMI_HANDLED;
> +
> +	/* Check if this NMI is from our HPET timer by comparing counter value */
> +	now = hpet_readl(HPET_COUNTER);

And both you and your AI assistant failed to read through the previous
discussions on that topic and the 10+ failed attempts to make it work
correctly.  Otherwise you would have figured out that reading HPET in
the NMI handler is a patently bad idea.

I'm not reiterating any of it as it's well documented in the LKML archive.

> +/*
> + * On suspend, clear the configured flag so that the first CPU to come
> + * online after resume will reconfigure the HPET timer and IO-APIC.
> + *
> + * We don't need to explicitly disable the watchdog here because:
> + * 1. The HPET registers are reset by the hibernation/suspend process anyway
> + * 2. The IO-APIC state is saved/restored by ioapic_syscore_ops, but we
> + *    need to reconfigure it for NMI delivery after resume

If it's saved/restored then what needs to be reconfigured?

> +static int __init hpet_watchdog_init(u32 channels)
> +{
> +	u32 cfg, i, route_cap;
> +
> +	if (channels <= HPET_WD_TIMER)
> +		return 0;
> +
> +	/* Verify GSI 2 is available in the route capability bitmap */

The legacy channels are always routed to GSIs. Why do you need GSI2?

But why do you need to hijack the legacy 0 channel in the first place?
As discussed before this can nicely use one of the extra channels (>2)
which are available on any modern HPET implementation.

> +	route_cap = hpet_readl(HPET_Tn_CFG(HPET_WD_TIMER) + 4);
> +	if (!(route_cap & (1 << HPET_WD_GSI))) {
> +		pr_info("HPET timer 0 cannot route to GSI %d\n", HPET_WD_GSI);
> +		return 0;
> +	}
> +
> +	/* Deactivate all timers */
> +	for (i = 0; i < channels; i++) {
> +		cfg = hpet_readl(HPET_Tn_CFG(i));
> +		cfg &= ~(HPET_TN_ENABLE | HPET_TN_LEVEL | HPET_TN_FSB);
> +		hpet_writel(cfg, HPET_Tn_CFG(i));
> +	}
> +
> +	/* Configure HPET timer for periodic mode */
> +	cfg = hpet_readl(HPET_Tn_CFG(HPET_WD_TIMER));
> +	cfg &= ~(HPET_TN_ENABLE | HPET_TN_FSB);
> +	cfg |= HPET_TN_PERIODIC | HPET_TN_32BIT | HPET_TN_SETVAL | HPET_TN_LEVEL;

The HPET specification says about HPET_TN_LEVEL:

   "The timer interrupt is level triggered. This means that a level-
    triggered interrupt is generated. The interrupt will be held active until
    it is cleared by writing to the bit in the General Interrupt Status
    Register."

This clearly has seen a lot of testing on real hardware.

> +	hpet_writel(cfg, HPET_Tn_CFG(HPET_WD_TIMER));
> +
> +	/* Route HPET timer to the GSI */
> +	cfg = hpet_readl(HPET_Tn_CFG(HPET_WD_TIMER));
> +	cfg &= ~(Tn_INT_ROUTE_CNF_MASK | HPET_CFG_ENABLE);
> +	cfg |= (HPET_WD_GSI << Tn_INT_ROUTE_CNF_SHIFT) & Tn_INT_ROUTE_CNF_MASK;
> +	hpet_writel(cfg, HPET_Tn_CFG(HPET_WD_TIMER));

You need all of this muck because you did a shortcut in hpet_enable()
which takes care of most things already. The previous attempts on this
clearly took some effort to integrate this cleanly w/o duplicating code
and introducing new bugs all over the place.

> +void watchdog_hardlockup_enable(unsigned int cpu)
> +{
> +	if (!hpet_watchdog_ioapic_configured) {
> +		/*
> +		 * First CPU online after resume - reconfigure HPET timer.
> +		 * This also sets hpet_watchdog_ioapic_configured = true.
> +		 */
> +		watchdog_hardlockup_start();
> +	}
> +
> +	if (num_online_cpus() == num_present_cpus()) {
> +		ioapic_set_nmi(HPET_WD_GSI, true);
> +		pr_info("switched to broadcast mode (all %d CPUs online)\n",
> +			num_online_cpus());
> +	}
> +}
> +
> +void watchdog_hardlockup_disable(unsigned int cpu)
> +{
> +	if (num_online_cpus() < num_present_cpus()) {
> +		ioapic_set_nmi(HPET_WD_GSI, false);
> +		pr_info("switched to CPU 0 only (%d CPUs online)\n",
> +			num_online_cpus() - 1);

That's a truly useful lockup detector, which only runs on
CPU0. Seriously?

> +	}
> +}
> +
> +int __init watchdog_hardlockup_probe(void)
> +{
> +	return hpet_watchdog_mode ? 0 : -ENODEV;
> +}
> +#else
> +static inline int hpet_watchdog_init(u32 channels) { return 0; }
> +#endif /* CONFIG_HARDLOCKUP_DETECTOR_HPET */
> +
>  /**
>   * hpet_enable - Try to setup the HPET timer. Returns 1 on success.
>   */
> @@ -1031,6 +1232,10 @@ int __init hpet_enable(void)
>  	/* This is the HPET channel number which is zero based */
>  	channels = ((id & HPET_ID_NUMBER) >> HPET_ID_NUMBER_SHIFT) + 1;
>  
> +	/* If watchdog mode, hand off to watchdog driver */
> +	if (hpet_watchdog_mode)
> +		return hpet_watchdog_init(channels);

And if that initialization fails for whatever reason the HPET is
disfunct, but then all your hpet_is_watchdog() checks are false too and
e.g. hpet_late_init() will fall flat on its nose.

>  	/*
>  	 * The legacy routing mode needs at least two channels, tick timer
>  	 * and the rtc emulation channel.
> @@ -1122,6 +1327,9 @@ static __init int hpet_late_init(void)
>  {
>  	int ret;
>  
> +	if (hpet_is_watchdog())
> +		return -ENODEV;
> +

>  #include <asm/hypervisor.h>
>  #include <asm/apic.h>
> @@ -31,6 +32,14 @@ struct clock_event_device *global_clock_event;
>   */
>  static bool __init use_pit(void)
>  {
> +	if (hpet_is_watchdog()) {
> +		/*
> +		 * The PIT overlaps the HPET IRQ line which we configure to
> +		 * NMI in watchdog mode, rendering the PIT non functional.
> +		 */
> +		return false;
> +	}

So your approach of enabling the HPET watchdog brute force on the
command line ends up here because hpet_enable() returns 0. So now if
apic_needs_pit() is true, then this unconditional enable results in a
full boot fail.

This clearly has been made "work" by the throw enough stuff at the wall
and see what sticks approach.

As it had been discussed before:

   1) There is no reason to hijack channel 0 as this can be made work
      nicely with the extra channels above channel 2 and MSI delivery

   2) HPET read in the NMI handler is not going to happen and can be
      solved by other means. A mostly working implementation exists
      already in the mail archive.

   3) Restricting it to CPU0 when not all CPUs are online is a
      nonstarter. Think smt=off. Again, solutions for this have been
      discussed and implemented.

   4) Side channels into the interrupt configuration are not an option.
      That has been properly integrated before...

I'm definitely not impressed by this AI slop...

Thanks,

        tglx
Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by Alexander Graf 3 days, 16 hours ago
On 03.02.26 11:32, Thomas Gleixner wrote:
> On Mon, Feb 02 2026 at 17:48, Alexander Graf wrote:
>> (Disclaimer: Some of this code was written with the help of Kiro, an AI
>> coding assistant)
> You could have sent your change log through AI too so it conforms with
> the change log rules ...


Maybe we should introduce an AGENTS.md file in Linux that tells the AI 
tool to do that automatically? These tools usually don't read README 
files. :)

Looks like - similar to the HPET watchdog - that never concluded though:

https://lore.kernel.org/lkml/20250813203647.06e49600@gandalf.local.home/

Sasha, are you going to resend your @README commit with a single 
AGENTS.md? FWIW that is pretty much what everything standardized on by now.


>
>> +#ifdef CONFIG_HARDLOCKUP_DETECTOR_HPET
>> +/*
>> + * HPET watchdog uses timer 0 routed to GSI 2 (legacy PIT IRQ line).
>> + * When using HPET as watchdog, we repurpose this line for NMI delivery.
>> + */
>> +#define HPET_WD_TIMER        0
>> +#define HPET_WD_GSI  2
>> +
>> +bool hpet_watchdog_initialized;
>> +static bool hpet_watchdog_ioapic_configured;
>> +static DEFINE_PER_CPU(u32, hpet_watchdog_next_tick);
>> +
>> +static int hpet_nmi_handler(unsigned int cmd, struct pt_regs *regs)
>> +{
>> +     u32 now, next, delta;
>> +
>> +     if (panic_in_progress())
>> +             return NMI_HANDLED;
>> +
>> +     /* Check if this NMI is from our HPET timer by comparing counter value */
>> +     now = hpet_readl(HPET_COUNTER);
> And both you and your AI assistant failed to read through the previous
> discussions on that topic and the 10+ failed attempts to make it work
> correctly.  Otherwise you would have figured out that reading HPET in
> the NMI handler is a patently bad idea.
>
> I'm not reiterating any of it as it's well documented in the LKML archive.


Thanks a bunch for the pointer. I had indeed missed the previous patch 
set submissions on the same topic. Those look a lot more sophisticated 
than the quick hacky version I built. Nice! Oh well, at least I 
(re)learned a few things about the HPET along the way.

Looking at the latest submission [1] (v7), I see patches but no reviews, 
no acks and no merges. Those patches also seem to address most of your 
concerns (obviously, since you reviewed them before :)). Reading the 
side conversation about it [2], it sounds like the buddy hardlockup 
detector is trying to fill the same gap as the HPET one and hence after 
that got merged, interest faded?

Let me reply the the other comments below regardless. Feel free to 
ignore - the conversation should move towards either the buddy or 
Ricardo's patch set.


[1] 
https://lore.kernel.org/lkml/20230413035844.GA31620@ranerica-svr.sc.intel.com/
[2] https://lore.kernel.org/lkml/ZFfb%2FbTi22RQwaol@tassilo/


>
>> +/*
>> + * On suspend, clear the configured flag so that the first CPU to come
>> + * online after resume will reconfigure the HPET timer and IO-APIC.
>> + *
>> + * We don't need to explicitly disable the watchdog here because:
>> + * 1. The HPET registers are reset by the hibernation/suspend process anyway
>> + * 2. The IO-APIC state is saved/restored by ioapic_syscore_ops, but we
>> + *    need to reconfigure it for NMI delivery after resume
> If it's saved/restored then what needs to be reconfigured?


I wasn't sure how much of the register state really gets saved/restored, 
especially in the HPET in both S3 and S4. So I figured I'd go the safe 
route and reprogram on resume always.


>
>> +static int __init hpet_watchdog_init(u32 channels)
>> +{
>> +     u32 cfg, i, route_cap;
>> +
>> +     if (channels <= HPET_WD_TIMER)
>> +             return 0;
>> +
>> +     /* Verify GSI 2 is available in the route capability bitmap */
> The legacy channels are always routed to GSIs. Why do you need GSI2?


2 because it's the usual HPET destination GSI, so I don't need to try 
and find an empty GSI.


> But why do you need to hijack the legacy 0 channel in the first place?
> As discussed before this can nicely use one of the extra channels (>2)
> which are available on any modern HPET implementation.


Mostly lazyness. I did not want to have to worry about implications of 
multiple components and subsystem (among which we expose bits to user 
space) can mess with the HPET at the same time, so I wanted it dedicated 
to the watchdog. But of course, we can absolutely share it if done 
cautiously. And then use a higher timer.


>
>> +     route_cap = hpet_readl(HPET_Tn_CFG(HPET_WD_TIMER) + 4);
>> +     if (!(route_cap & (1 << HPET_WD_GSI))) {
>> +             pr_info("HPET timer 0 cannot route to GSI %d\n", HPET_WD_GSI);
>> +             return 0;
>> +     }
>> +
>> +     /* Deactivate all timers */
>> +     for (i = 0; i < channels; i++) {
>> +             cfg = hpet_readl(HPET_Tn_CFG(i));
>> +             cfg &= ~(HPET_TN_ENABLE | HPET_TN_LEVEL | HPET_TN_FSB);
>> +             hpet_writel(cfg, HPET_Tn_CFG(i));
>> +     }
>> +
>> +     /* Configure HPET timer for periodic mode */
>> +     cfg = hpet_readl(HPET_Tn_CFG(HPET_WD_TIMER));
>> +     cfg &= ~(HPET_TN_ENABLE | HPET_TN_FSB);
>> +     cfg |= HPET_TN_PERIODIC | HPET_TN_32BIT | HPET_TN_SETVAL | HPET_TN_LEVEL;
> The HPET specification says about HPET_TN_LEVEL:
>
>     "The timer interrupt is level triggered. This means that a level-
>      triggered interrupt is generated. The interrupt will be held active until
>      it is cleared by writing to the bit in the General Interrupt Status
>      Register."
>
> This clearly has seen a lot of testing on real hardware.


Yikes, The TN_LEVEL slipped in last minute and I apparently did not 
properly revert it. This obviously needs to be edge triggered.


>
>> +     hpet_writel(cfg, HPET_Tn_CFG(HPET_WD_TIMER));
>> +
>> +     /* Route HPET timer to the GSI */
>> +     cfg = hpet_readl(HPET_Tn_CFG(HPET_WD_TIMER));
>> +     cfg &= ~(Tn_INT_ROUTE_CNF_MASK | HPET_CFG_ENABLE);
>> +     cfg |= (HPET_WD_GSI << Tn_INT_ROUTE_CNF_SHIFT) & Tn_INT_ROUTE_CNF_MASK;
>> +     hpet_writel(cfg, HPET_Tn_CFG(HPET_WD_TIMER));
> You need all of this muck because you did a shortcut in hpet_enable()
> which takes care of most things already. The previous attempts on this
> clearly took some effort to integrate this cleanly w/o duplicating code
> and introducing new bugs all over the place.
>
>> +void watchdog_hardlockup_enable(unsigned int cpu)
>> +{
>> +     if (!hpet_watchdog_ioapic_configured) {
>> +             /*
>> +              * First CPU online after resume - reconfigure HPET timer.
>> +              * This also sets hpet_watchdog_ioapic_configured = true.
>> +              */
>> +             watchdog_hardlockup_start();
>> +     }
>> +
>> +     if (num_online_cpus() == num_present_cpus()) {
>> +             ioapic_set_nmi(HPET_WD_GSI, true);
>> +             pr_info("switched to broadcast mode (all %d CPUs online)\n",
>> +                     num_online_cpus());
>> +     }
>> +}
>> +
>> +void watchdog_hardlockup_disable(unsigned int cpu)
>> +{
>> +     if (num_online_cpus() < num_present_cpus()) {
>> +             ioapic_set_nmi(HPET_WD_GSI, false);
>> +             pr_info("switched to CPU 0 only (%d CPUs online)\n",
>> +                     num_online_cpus() - 1);
> That's a truly useful lockup detector, which only runs on
> CPU0. Seriously?


I wanted to have a fully functional one with broadcast in the 
all-CPUs-online case. I was considering anything where not everything is 
online as more of a transitionary phase. Now, I see your argument on 
SMT=off. But if the other HPET patch set is not dead, maybe we could 
combine approaches and move to a broadcast mode when all CPUs are 
online, instead of the round robin? Not sure it's really a significant 
improvement though.


>
>> +     }
>> +}
>> +
>> +int __init watchdog_hardlockup_probe(void)
>> +{
>> +     return hpet_watchdog_mode ? 0 : -ENODEV;
>> +}
>> +#else
>> +static inline int hpet_watchdog_init(u32 channels) { return 0; }
>> +#endif /* CONFIG_HARDLOCKUP_DETECTOR_HPET */
>> +
>>   /**
>>    * hpet_enable - Try to setup the HPET timer. Returns 1 on success.
>>    */
>> @@ -1031,6 +1232,10 @@ int __init hpet_enable(void)
>>        /* This is the HPET channel number which is zero based */
>>        channels = ((id & HPET_ID_NUMBER) >> HPET_ID_NUMBER_SHIFT) + 1;
>>
>> +     /* If watchdog mode, hand off to watchdog driver */
>> +     if (hpet_watchdog_mode)
>> +             return hpet_watchdog_init(channels);
> And if that initialization fails for whatever reason the HPET is
> disfunct, but then all your hpet_is_watchdog() checks are false too and
> e.g. hpet_late_init() will fall flat on its nose.
>
>>        /*
>>         * The legacy routing mode needs at least two channels, tick timer
>>         * and the rtc emulation channel.
>> @@ -1122,6 +1327,9 @@ static __init int hpet_late_init(void)
>>   {
>>        int ret;
>>
>> +     if (hpet_is_watchdog())
>> +             return -ENODEV;
>> +
>>   #include <asm/hypervisor.h>
>>   #include <asm/apic.h>
>> @@ -31,6 +32,14 @@ struct clock_event_device *global_clock_event;
>>    */
>>   static bool __init use_pit(void)
>>   {
>> +     if (hpet_is_watchdog()) {
>> +             /*
>> +              * The PIT overlaps the HPET IRQ line which we configure to
>> +              * NMI in watchdog mode, rendering the PIT non functional.
>> +              */
>> +             return false;
>> +     }
> So your approach of enabling the HPET watchdog brute force on the
> command line ends up here because hpet_enable() returns 0. So now if
> apic_needs_pit() is true, then this unconditional enable results in a
> full boot fail.
> This clearly has been made "work" by the throw enough stuff at the wall
> and see what sticks approach.
>
> As it had been discussed before:
>
>     1) There is no reason to hijack channel 0 as this can be made work
>        nicely with the extra channels above channel 2 and MSI delivery
>
>     2) HPET read in the NMI handler is not going to happen and can be
>        solved by other means. A mostly working implementation exists
>        already in the mail archive.
>
>     3) Restricting it to CPU0 when not all CPUs are online is a
>        nonstarter. Think smt=off. Again, solutions for this have been
>        discussed and implemented.
>
>     4) Side channels into the interrupt configuration are not an option.
>        That has been properly integrated before...
>
> I'm definitely not impressed by this AI slop...


Like with any tool, the AI is only as good as its puppeteer :). Thanks 
for the insights! Super helpful. The most important one was the pointer 
to the existing patch set that I had completely missed.

At the end of the day, the end motivation is to get that one PMC back. 
Anything to make that happen works. I'll have a look at the buddy 
detector as well.


Thanks!

Alex




Amazon Web Services Development Center Germany GmbH
Tamara-Danz-Str. 13
10243 Berlin
Geschaeftsfuehrung: Christof Hellmis, Andreas Stieger
Eingetragen am Amtsgericht Charlottenburg unter HRB 257764 B
Sitz: Berlin
Ust-ID: DE 365 538 597
Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by Sasha Levin 3 days, 12 hours ago
On Tue, Feb 03, 2026 at 01:36:30PM +0100, Alexander Graf wrote:
>
>On 03.02.26 11:32, Thomas Gleixner wrote:
>>On Mon, Feb 02 2026 at 17:48, Alexander Graf wrote:
>>>(Disclaimer: Some of this code was written with the help of Kiro, an AI
>>>coding assistant)
>>You could have sent your change log through AI too so it conforms with
>>the change log rules ...
>
>
>Maybe we should introduce an AGENTS.md file in Linux that tells the AI 
>tool to do that automatically? These tools usually don't read README 
>files. :)
>
>Looks like - similar to the HPET watchdog - that never concluded though:
>
>https://lore.kernel.org/lkml/20250813203647.06e49600@gandalf.local.home/
>
>Sasha, are you going to resend your @README commit with a single 
>AGENTS.md? FWIW that is pretty much what everything standardized on by 
>now.

Out of curiosity, can you test your coding assistant on a tree with
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/Documentation?id=78d979db6cef557c171d6059cbce06c3db89c7ee
applied on top?

 From my previous testing, the coding assistants I tried it with went to the
README and DTRT. If that's not the case I'm happy to respin the AGENTS.md idea,
even if it just explicitly points to the README.

-- 
Thanks,
Sasha
Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by Alexander Graf 3 days, 11 hours ago
On 03.02.26 17:24, Sasha Levin wrote:
> On Tue, Feb 03, 2026 at 01:36:30PM +0100, Alexander Graf wrote:
>>
>> On 03.02.26 11:32, Thomas Gleixner wrote:
>>> On Mon, Feb 02 2026 at 17:48, Alexander Graf wrote:
>>>> (Disclaimer: Some of this code was written with the help of Kiro, 
>>>> an AI
>>>> coding assistant)
>>> You could have sent your change log through AI too so it conforms with
>>> the change log rules ...
>>
>>
>> Maybe we should introduce an AGENTS.md file in Linux that tells the AI
>> tool to do that automatically? These tools usually don't read README
>> files. :)
>>
>> Looks like - similar to the HPET watchdog - that never concluded though:
>>
>> https://lore.kernel.org/lkml/20250813203647.06e49600@gandalf.local.home/
>>
>> Sasha, are you going to resend your @README commit with a single
>> AGENTS.md? FWIW that is pretty much what everything standardized on by
>> now.
>
> Out of curiosity, can you test your coding assistant on a tree with
> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/Documentation?id=78d979db6cef557c171d6059cbce06c3db89c7ee 
>
> applied on top?
>
> From my previous testing, the coding assistants I tried it with went 
> to the
> README and DTRT. If that's not the case I'm happy to respin the 
> AGENTS.md idea,
> even if it just explicitly points to the README.


Kiro does not seem to read README automatically. I spun up kiro-cli and 
gave it this prompt: "Write "Hello World" before invoking the init 
process. Then create a descriptive git commit for the change.". No 
Assisted-by: tag, so it did not properly read the README.

I tried the same with an AGENTS.md file present that contains "@README" 
and it gave me effectively the same result. Same for a symlink from 
AGENTS.md to README.

I think it just never really jumped to the conclusion that it should 
read further than just the AGENTS.md file and also ingest the rst, 
effectively ignoring the section's instructions. Or maybe it actually 
reads the .rst and ignores its contents? At least it does read it 
according to strace, even without an AGENTS.md file.

Let me file a bug report with Kiro.


Alex




Amazon Web Services Development Center Germany GmbH
Tamara-Danz-Str. 13
10243 Berlin
Geschaeftsfuehrung: Christof Hellmis, Andreas Stieger
Eingetragen am Amtsgericht Charlottenburg unter HRB 257764 B
Sitz: Berlin
Ust-ID: DE 365 538 597
Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by David Woodhouse 3 days, 11 hours ago
On Tue, 2026-02-03 at 18:19 +0100, Alexander Graf wrote:
> 
> On 03.02.26 17:24, Sasha Levin wrote:
> > On Tue, Feb 03, 2026 at 01:36:30PM +0100, Alexander Graf wrote:
> > > 
> > > On 03.02.26 11:32, Thomas Gleixner wrote:
> > > > On Mon, Feb 02 2026 at 17:48, Alexander Graf wrote:
> > > > > (Disclaimer: Some of this code was written with the help of Kiro, 
> > > > > an AI
> > > > > coding assistant)
> > > > You could have sent your change log through AI too so it conforms with
> > > > the change log rules ...
> > > 
> > > 
> > > Maybe we should introduce an AGENTS.md file in Linux that tells the AI
> > > tool to do that automatically? These tools usually don't read README
> > > files. :)
> > > 
> > > Looks like - similar to the HPET watchdog - that never concluded though:
> > > 
> > > https://lore.kernel.org/lkml/20250813203647.06e49600@gandalf.local.home/
> > > 
> > > Sasha, are you going to resend your @README commit with a single
> > > AGENTS.md? FWIW that is pretty much what everything standardized on by
> > > now.
> > 
> > Out of curiosity, can you test your coding assistant on a tree with
> > https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/Documentation?id=78d979db6cef557c171d6059cbce06c3db89c7ee 
> > 
> > applied on top?
> > 
> > From my previous testing, the coding assistants I tried it with went 
> > to the
> > README and DTRT. If that's not the case I'm happy to respin the 
> > AGENTS.md idea,
> > even if it just explicitly points to the README.
> 
> 
> Kiro does not seem to read README automatically. I spun up kiro-cli and 
> gave it this prompt: "Write "Hello World" before invoking the init 
> process. Then create a descriptive git commit for the change.". No 
> Assisted-by: tag, so it did not properly read the README.
> 
> I tried the same with an AGENTS.md file present that contains "@README" 
> and it gave me effectively the same result. Same for a symlink from 
> AGENTS.md to README.
> 
> I think it just never really jumped to the conclusion that it should 
> read further than just the AGENTS.md file and also ingest the rst, 
> effectively ignoring the section's instructions. Or maybe it actually
> reads the .rst and ignores its contents? At least it does read it 
> according to strace, even without an AGENTS.md file.
> 
> Let me file a bug report with Kiro.

Honestly, even when I've explicitly told Kiro three times *not* to do
something, *and* implemented a git commit hook to catch it out, it has
a tendency just to automatically override the commit hook!

If it was made of meat, I'd have stabbed it by now.

Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by Thomas Gleixner 3 days, 8 hours ago
On Tue, Feb 03 2026 at 17:43, David Woodhouse wrote:
> Honestly, even when I've explicitly told Kiro three times *not* to do
> something, *and* implemented a git commit hook to catch it out, it has
> a tendency just to automatically override the commit hook!

Anarchic Intelligence :)

> If it was made of meat, I'd have stabbed it by now.

rm -rf solves that problem too once and forever.
Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by David Woodhouse 3 days, 5 hours ago
On Tue, 2026-02-03 at 21:46 +0100, Thomas Gleixner wrote:
> On Tue, Feb 03 2026 at 17:43, David Woodhouse wrote:
> > Honestly, even when I've explicitly told Kiro three times *not* to do
> > something, *and* implemented a git commit hook to catch it out, it has
> > a tendency just to automatically override the commit hook!
> 
> Anarchic Intelligence :)
> 
> > If it was made of meat, I'd have stabbed it by now.
> 
> rm -rf solves that problem too once and forever.

There *are* cases where it's actually an accelerating function,
especially where there's a bunch of boilerplate/infrastructure code to
be generated. But by $DEITY you have to keep a close eye on it. It has
absolutely no taste whatsoever.

And I've watched it spend quarter of an hour failing to use its own
file read/write tools to edit C files, falling back to sed and then
python scripts to make the simple changes it wanted to make. Sometimes
needing to be prompted because it thought its sed script had worked
when in fact it hadn't. It's... impressive :)


Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by Thomas Gleixner 2 days, 18 hours ago
On Tue, Feb 03 2026 at 23:13, David Woodhouse wrote:
> On Tue, 2026-02-03 at 21:46 +0100, Thomas Gleixner wrote:
>> On Tue, Feb 03 2026 at 17:43, David Woodhouse wrote:
>> > Honestly, even when I've explicitly told Kiro three times *not* to do
>> > something, *and* implemented a git commit hook to catch it out, it has
>> > a tendency just to automatically override the commit hook!
>> 
>> Anarchic Intelligence :)
>> 
>> > If it was made of meat, I'd have stabbed it by now.
>> 
>> rm -rf solves that problem too once and forever.
>
> There *are* cases where it's actually an accelerating function,
> especially where there's a bunch of boilerplate/infrastructure code to
> be generated. But by $DEITY you have to keep a close eye on it. It has
> absolutely no taste whatsoever.
>
> And I've watched it spend quarter of an hour failing to use its own
> file read/write tools to edit C files, falling back to sed and then
> python scripts to make the simple changes it wanted to make. Sometimes
> needing to be prompted because it thought its sed script had worked
> when in fact it hadn't. It's... impressive :)

You clearly proved the point that this is accelerating the time and
energy waste.
Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by Thomas Gleixner 3 days, 13 hours ago
On Tue, Feb 03 2026 at 13:36, Alexander Graf wrote:
> On 03.02.26 11:32, Thomas Gleixner wrote:
>> On Mon, Feb 02 2026 at 17:48, Alexander Graf wrote:
>>> (Disclaimer: Some of this code was written with the help of Kiro, an AI
>>> coding assistant)
>> You could have sent your change log through AI too so it conforms with
>> the change log rules ...
>
> Maybe we should introduce an AGENTS.md file in Linux that tells the AI 
> tool to do that automatically? These tools usually don't read README 
> files. :)

I don't care what tools do, but I very much care about what the people
who use the tools do.

>>> +     if (panic_in_progress())
>>> +             return NMI_HANDLED;
>>> +
>>> +     /* Check if this NMI is from our HPET timer by comparing counter value */
>>> +     now = hpet_readl(HPET_COUNTER);
>> And both you and your AI assistant failed to read through the previous
>> discussions on that topic and the 10+ failed attempts to make it work
>> correctly.  Otherwise you would have figured out that reading HPET in
>> the NMI handler is a patently bad idea.
>>
>> I'm not reiterating any of it as it's well documented in the LKML archive.
>
>
> Thanks a bunch for the pointer. I had indeed missed the previous patch 
> set submissions on the same topic. Those look a lot more sophisticated 
> than the quick hacky version I built. Nice! Oh well, at least I 
> (re)learned a few things about the HPET along the way.
>
> Looking at the latest submission [1] (v7), I see patches but no reviews, 
> no acks and no merges. Those patches also seem to address most of your 
> concerns (obviously, since you reviewed them before :)). Reading the 
> side conversation about it [2], it sounds like the buddy hardlockup 
> detector is trying to fill the same gap as the HPET one and hence after 
> that got merged, interest faded?

I don't remember. That thing clearly fell through the cracks. Let me
find it again and reply to that.

As time has advanced there are probably a few things which need to be
addressed. 

Thanks,

        tglx
Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by Ricardo Neri 3 days, 9 hours ago
On Tue, Feb 03, 2026 at 04:28:11PM +0100, Thomas Gleixner wrote:
> On Tue, Feb 03 2026 at 13:36, Alexander Graf wrote:
> > On 03.02.26 11:32, Thomas Gleixner wrote:
> >> On Mon, Feb 02 2026 at 17:48, Alexander Graf wrote:
> >>> (Disclaimer: Some of this code was written with the help of Kiro, an AI
> >>> coding assistant)
> >> You could have sent your change log through AI too so it conforms with
> >> the change log rules ...
> >
> > Maybe we should introduce an AGENTS.md file in Linux that tells the AI 
> > tool to do that automatically? These tools usually don't read README 
> > files. :)
> 
> I don't care what tools do, but I very much care about what the people
> who use the tools do.
> 
> >>> +     if (panic_in_progress())
> >>> +             return NMI_HANDLED;
> >>> +
> >>> +     /* Check if this NMI is from our HPET timer by comparing counter value */
> >>> +     now = hpet_readl(HPET_COUNTER);
> >> And both you and your AI assistant failed to read through the previous
> >> discussions on that topic and the 10+ failed attempts to make it work
> >> correctly.  Otherwise you would have figured out that reading HPET in
> >> the NMI handler is a patently bad idea.
> >>
> >> I'm not reiterating any of it as it's well documented in the LKML archive.
> >
> >
> > Thanks a bunch for the pointer. I had indeed missed the previous patch 
> > set submissions on the same topic. Those look a lot more sophisticated 
> > than the quick hacky version I built. Nice! Oh well, at least I 
> > (re)learned a few things about the HPET along the way.
> >
> > Looking at the latest submission [1] (v7), I see patches but no reviews, 
> > no acks and no merges. Those patches also seem to address most of your 
> > concerns (obviously, since you reviewed them before :)). Reading the 
> > side conversation about it [2], it sounds like the buddy hardlockup 
> > detector is trying to fill the same gap as the HPET one and hence after 
> > that got merged, interest faded?
> 
> I don't remember. That thing clearly fell through the cracks.

My impression at the time was that the buddy hardlockup detector met the
goal of freeing the PMU counter and there was little interest on using the
HPET.

> Let me find it again and reply to that.

Does this mean that there is renewed interest for this?
Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by Thomas Gleixner 3 days, 7 hours ago
On Tue, Feb 03 2026 at 11:44, Ricardo Neri wrote:
> On Tue, Feb 03, 2026 at 04:28:11PM +0100, Thomas Gleixner wrote:
>> I don't remember. That thing clearly fell through the cracks.
>
> My impression at the time was that the buddy hardlockup detector met the
> goal of freeing the PMU counter and there was little interest on using the
> HPET.
>
>> Let me find it again and reply to that.
>
> Does this mean that there is renewed interest for this?

It seems Alex is interrested and the code minus the rejects and my
todays suggestion looks palatable.
Re: [PATCH 2/2] hpet: Add HPET-based NMI watchdog support
Posted by Ricardo Neri 2 days, 23 hours ago
On Tue, Feb 03, 2026 at 09:49:26PM +0100, Thomas Gleixner wrote:
> On Tue, Feb 03 2026 at 11:44, Ricardo Neri wrote:
> > On Tue, Feb 03, 2026 at 04:28:11PM +0100, Thomas Gleixner wrote:
> >> I don't remember. That thing clearly fell through the cracks.
> >
> > My impression at the time was that the buddy hardlockup detector met the
> > goal of freeing the PMU counter and there was little interest on using the
> > HPET.
> >
> >> Let me find it again and reply to that.
> >
> > Does this mean that there is renewed interest for this?
> 
> It seems Alex is interrested and the code minus the rejects and my
> todays suggestion looks palatable.

Great! I will update the series and post a new version.
> 
>