[PATCH v4 15/17] watchdog/perf: Adapt the watchdog_perf interface for async model

[Douglas Anderson]

From: Lecopzer Chen <lecopzer.chen at mediatek.com>

When lockup_detector_init()->watchdog_hardlockup_probe(), PMU may be
not ready yet. E.g. on arm64, PMU is not ready until
device_initcall(armv8_pmu_driver_init).  And it is deeply integrated
with the driver model and cpuhp. Hence it is hard to push this
initialization before smp_init().

But it is easy to take an opposite approach and try to initialize
the watchdog once again later.
The delayed probe is called using workqueues. It need to allocate
memory and must be proceed in a normal context.
The delayed probe is able to use if watchdog_hardlockup_probe() returns
non-zero which means the return code returned when PMU is not ready yet.

Provide an API - lockup_detector_retry_init() for anyone who needs
to delayed init lockup detector if they had ever failed at
lockup_detector_init().

The original assumption is: nobody should use delayed probe after
lockup_detector_check() which has __init attribute.
That is, anyone uses this API must call between lockup_detector_init()
and lockup_detector_check(), and the caller must have __init attribute

Reviewed-by: Petr Mladek <pmladek at suse.com>
Co-developed-by: Pingfan Liu <kernelfans at gmail.com>
Signed-off-by: Pingfan Liu <kernelfans at gmail.com>
Signed-off-by: Lecopzer Chen <lecopzer.chen at mediatek.com>
Suggested-by: Petr Mladek <pmladek at suse.com>
Signed-off-by: Douglas Anderson <dianders at chromium.org>
---
I yanked this patch from the mailing lists [1] into my series just to
make it easier to avoid conflicts between my series and the one adding
the arm64 perf hardlockup detector, in case someone wanted to test
them both together.

As part of making this match with my series, I resolved a few
conflicts and did a few renames. I also fixed the kernel test robot
yell [2] by providing a dummy implementation of the retry function if
CONFIG_LOCKUP_DETECTOR wasn't defined. That led me to move the
definition of the function and sanitize the name of the function to
match others around it.

This patch makes less sense without the patches to add support for the
arm64 perf hardlockup detector. Thus, I've put it at the end of the
series.

I removed the "kernel test robot" tag since that didn't really make
sense. Presumably the robot found a problem on a previous version of
the patch, but that's not normally a reason to add the Reported-by.

[1] https://lore.kernel.org/r/20220903093415.15850-5-lecopzer.chen@mediatek.com/
[2] https://lore.kernel.org/r/202209050639.jDaWd49E-lkp@intel.com/

Changes in v4:
- Pulled ("Adapt the watchdog_hld interface ...") into my series for v4.

 include/linux/nmi.h |  2 ++
 kernel/watchdog.c   | 67 ++++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 68 insertions(+), 1 deletion(-)

diff --git a/include/linux/nmi.h b/include/linux/nmi.h
index 9caea5ba494d..3cacc5fc16b9 100644
--- a/include/linux/nmi.h
+++ b/include/linux/nmi.h
@@ -13,6 +13,7 @@
 
 #ifdef CONFIG_LOCKUP_DETECTOR
 void lockup_detector_init(void);
+void lockup_detector_retry_init(void);
 void lockup_detector_soft_poweroff(void);
 void lockup_detector_cleanup(void);
 
@@ -34,6 +35,7 @@ extern int sysctl_hardlockup_all_cpu_backtrace;
 
 #else /* CONFIG_LOCKUP_DETECTOR */
 static inline void lockup_detector_init(void) { }
+static inline void lockup_detector_retry_init(void) { }
 static inline void lockup_detector_soft_poweroff(void) { }
 static inline void lockup_detector_cleanup(void) { }
 #endif /* !CONFIG_LOCKUP_DETECTOR */
diff --git a/kernel/watchdog.c b/kernel/watchdog.c
index 678d55172ef4..55471634d932 100644
--- a/kernel/watchdog.c
+++ b/kernel/watchdog.c
@@ -211,7 +211,13 @@ void __weak watchdog_hardlockup_enable(unsigned int cpu) { }
 
 void __weak watchdog_hardlockup_disable(unsigned int cpu) { }
 
-/* Return 0, if a hardlockup watchdog is available. Error code otherwise */
+/*
+ * Watchdog-detector specific API.
+ *
+ * Return 0 when hardlockup watchdog is available, negative value otherwise.
+ * Note that the negative value means that a delayed probe might
+ * succeed later.
+ */
 int __weak __init watchdog_hardlockup_probe(void)
 {
 	/*
@@ -957,6 +963,62 @@ static void __init watchdog_sysctl_init(void)
 #define watchdog_sysctl_init() do { } while (0)
 #endif /* CONFIG_SYSCTL */
 
+static void __init lockup_detector_delay_init(struct work_struct *work);
+static bool allow_lockup_detector_init_retry __initdata;
+
+static struct work_struct detector_work __initdata =
+		__WORK_INITIALIZER(detector_work, lockup_detector_delay_init);
+
+static void __init lockup_detector_delay_init(struct work_struct *work)
+{
+	int ret;
+
+	ret = watchdog_hardlockup_probe();
+	if (ret) {
+		pr_info("Delayed init of the lockup detector failed: %d\n", ret);
+		pr_info("Hard watchdog permanently disabled\n");
+		return;
+	}
+
+	allow_lockup_detector_init_retry = false;
+
+	nmi_watchdog_available = true;
+	lockup_detector_setup();
+}
+
+/*
+ * lockup_detector_retry_init - retry init lockup detector if possible.
+ *
+ * Retry hardlockup detector init. It is useful when it requires some
+ * functionality that has to be initialized later on a particular
+ * platform.
+ */
+void __init lockup_detector_retry_init(void)
+{
+	/* Must be called before late init calls */
+	if (!allow_lockup_detector_init_retry)
+		return;
+
+	schedule_work(&detector_work);
+}
+
+/*
+ * Ensure that optional delayed hardlockup init is proceed before
+ * the init code and memory is freed.
+ */
+static int __init lockup_detector_check(void)
+{
+	/* Prevent any later retry. */
+	allow_lockup_detector_init_retry = false;
+
+	/* Make sure no work is pending. */
+	flush_work(&detector_work);
+
+	return 0;
+
+}
+late_initcall_sync(lockup_detector_check);
+
 void __init lockup_detector_init(void)
 {
 	if (tick_nohz_full_enabled())
@@ -967,6 +1029,9 @@ void __init lockup_detector_init(void)
 
 	if (!watchdog_hardlockup_probe())
 		nmi_watchdog_available = true;
+	else
+		allow_lockup_detector_init_retry = true;
+
 	lockup_detector_setup();
 	watchdog_sysctl_init();
 }
-- 
2.40.1.521.gf1e218fcd8-goog