The patch titled
Subject: watchdog/hardlockup: move perf hardlockup checking/panic to common watchdog.c
has been added to the -mm mm-nonmm-unstable branch. Its filename is
watchdog-hardlockup-move-perf-hardlockup-checking-panic-to-common-watchdogc.patch
This patch will shortly appear at
https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches/watchdog-hardlockup-move-perf-hardlockup-checking-panic-to-common-watchdogc.patch
This patch will later appear in the mm-nonmm-unstable branch at
git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/process/submit-checklist.rst when testing your code ***
The -mm tree is included into linux-next via the mm-everything
branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
and is updated there every 2-3 working days
------------------------------------------------------
From: Douglas Anderson <dianders@xxxxxxxxxxxx>
Subject: watchdog/hardlockup: move perf hardlockup checking/panic to common watchdog.c
Date: Fri, 19 May 2023 10:18:32 -0700
The perf hardlockup detector works by looking at interrupt counts and
seeing if they change from run to run. The interrupt counts are managed
by the common watchdog code via its watchdog_timer_fn().
Currently the API between the perf detector and the common code is a
function: is_hardlockup(). When the hard lockup detector sees that
function return true then it handles printing out debug info and inducing
a panic if necessary.
Let's change the API a little bit in preparation for the buddy hardlockup
detector. The buddy hardlockup detector wants to print nearly the same
debug info and have nearly the same panic behavior. That means we want to
move all that code to the common file. For now, the code in the common
file will only be there if the perf hardlockup detector is enabled, but
eventually it will be selected by a common config.
Right now, this _just_ moves the code from the perf detector file to the
common file and changes the names. It doesn't make the changes that the
buddy hardlockup detector will need and doesn't do any style cleanups. A
future patch will do cleanup to make it more obvious what changed.
With the above, we no longer have any callers of is_hardlockup() outside
of the "watchdog.c" file, so we can remove it from the header, make it
static, and move it to the same "#ifdef" block as our new
watchdog_hardlockup_check(). While doing this, it can be noted that even
if no hardlockup detectors were configured the existing code used to still
have the code for counting/checking "hrtimer_interrupts" even if the perf
hardlockup detector wasn't configured. We didn't need to do that, so move
all the "hrtimer_interrupts" counting to only be there if the perf
hardlockup detector is configured as well.
This change is expected to be a no-op.
Link: https://lkml.kernel.org/r/20230519101840.v5.8.Id4133d3183e798122dc3b6205e7852601f289071@changeid
Signed-off-by: Douglas Anderson <dianders@xxxxxxxxxxxx>
Cc: Andi Kleen <ak@xxxxxxxxxxxxxxx>
Cc: Catalin Marinas <catalin.marinas@xxxxxxx>
Cc: Chen-Yu Tsai <wens@xxxxxxxx>
Cc: Christophe Leroy <christophe.leroy@xxxxxxxxxx>
Cc: Colin Cross <ccross@xxxxxxxxxxx>
Cc: Daniel Thompson <daniel.thompson@xxxxxxxxxx>
Cc: "David S. Miller" <davem@xxxxxxxxxxxxx>
Cc: Guenter Roeck <groeck@xxxxxxxxxxxx>
Cc: Ian Rogers <irogers@xxxxxxxxxx>
Cc: Lecopzer Chen <lecopzer.chen@xxxxxxxxxxxx>
Cc: Marc Zyngier <maz@xxxxxxxxxx>
Cc: Mark Rutland <mark.rutland@xxxxxxx>
Cc: Masayoshi Mizuma <msys.mizuma@xxxxxxxxx>
Cc: Matthias Kaehlcke <mka@xxxxxxxxxxxx>
Cc: Michael Ellerman <mpe@xxxxxxxxxxxxxx>
Cc: Nicholas Piggin <npiggin@xxxxxxxxx>
Cc: Petr Mladek <pmladek@xxxxxxxx>
Cc: Pingfan Liu <kernelfans@xxxxxxxxx>
Cc: Randy Dunlap <rdunlap@xxxxxxxxxxxxx>
Cc: "Ravi V. Shankar" <ravi.v.shankar@xxxxxxxxx>
Cc: Ricardo Neri <ricardo.neri@xxxxxxxxx>
Cc: Stephane Eranian <eranian@xxxxxxxxxx>
Cc: Stephen Boyd <swboyd@xxxxxxxxxxxx>
Cc: Sumit Garg <sumit.garg@xxxxxxxxxx>
Cc: Tzung-Bi Shih <tzungbi@xxxxxxxxxxxx>
Cc: Will Deacon <will@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
include/linux/nmi.h | 5 +-
kernel/watchdog.c | 93 ++++++++++++++++++++++++++++++---------
kernel/watchdog_perf.c | 42 -----------------
3 files changed, 78 insertions(+), 62 deletions(-)
--- a/include/linux/nmi.h~watchdog-hardlockup-move-perf-hardlockup-checking-panic-to-common-watchdogc
+++ a/include/linux/nmi.h
@@ -15,7 +15,6 @@
void lockup_detector_init(void);
void lockup_detector_soft_poweroff(void);
void lockup_detector_cleanup(void);
-bool is_hardlockup(void);
extern int watchdog_user_enabled;
extern int nmi_watchdog_user_enabled;
@@ -88,6 +87,10 @@ extern unsigned int hardlockup_panic;
static inline void hardlockup_detector_disable(void) {}
#endif
+#if defined(CONFIG_HARDLOCKUP_DETECTOR_PERF)
+void watchdog_hardlockup_check(struct pt_regs *regs);
+#endif
+
#if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
# define NMI_WATCHDOG_SYSCTL_PERM 0644
#else
--- a/kernel/watchdog.c~watchdog-hardlockup-move-perf-hardlockup-checking-panic-to-common-watchdogc
+++ a/kernel/watchdog.c
@@ -85,6 +85,78 @@ __setup("nmi_watchdog=", hardlockup_pani
#endif /* CONFIG_HARDLOCKUP_DETECTOR */
+#if defined(CONFIG_HARDLOCKUP_DETECTOR_PERF)
+
+static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts);
+static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved);
+static DEFINE_PER_CPU(bool, hard_watchdog_warn);
+static unsigned long hardlockup_allcpu_dumped;
+
+static bool is_hardlockup(void)
+{
+ unsigned long hrint = __this_cpu_read(hrtimer_interrupts);
+
+ if (__this_cpu_read(hrtimer_interrupts_saved) == hrint)
+ return true;
+
+ __this_cpu_write(hrtimer_interrupts_saved, hrint);
+ return false;
+}
+
+static void watchdog_hardlockup_kick(void)
+{
+ __this_cpu_inc(hrtimer_interrupts);
+}
+
+void watchdog_hardlockup_check(struct pt_regs *regs)
+{
+ /* check for a hardlockup
+ * This is done by making sure our timer interrupt
+ * is incrementing. The timer interrupt should have
+ * fired multiple times before we overflow'd. If it hasn't
+ * then this is a good indication the cpu is stuck
+ */
+ if (is_hardlockup()) {
+ int this_cpu = smp_processor_id();
+
+ /* only print hardlockups once */
+ if (__this_cpu_read(hard_watchdog_warn) == true)
+ return;
+
+ pr_emerg("Watchdog detected hard LOCKUP on cpu %d\n",
+ this_cpu);
+ print_modules();
+ print_irqtrace_events(current);
+ if (regs)
+ show_regs(regs);
+ else
+ dump_stack();
+
+ /*
+ * Perform all-CPU dump only once to avoid multiple hardlockups
+ * generating interleaving traces
+ */
+ if (sysctl_hardlockup_all_cpu_backtrace &&
+ !test_and_set_bit(0, &hardlockup_allcpu_dumped))
+ trigger_allbutself_cpu_backtrace();
+
+ if (hardlockup_panic)
+ nmi_panic(regs, "Hard LOCKUP");
+
+ __this_cpu_write(hard_watchdog_warn, true);
+ return;
+ }
+
+ __this_cpu_write(hard_watchdog_warn, false);
+ return;
+}
+
+#else /* CONFIG_HARDLOCKUP_DETECTOR_PERF */
+
+static inline void watchdog_hardlockup_kick(void) { }
+
+#endif /* !CONFIG_HARDLOCKUP_DETECTOR_PERF */
+
/*
* These functions can be overridden if an architecture implements its
* own hardlockup detector.
@@ -176,8 +248,6 @@ static DEFINE_PER_CPU(unsigned long, wat
static DEFINE_PER_CPU(unsigned long, watchdog_report_ts);
static DEFINE_PER_CPU(struct hrtimer, watchdog_hrtimer);
static DEFINE_PER_CPU(bool, softlockup_touch_sync);
-static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts);
-static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved);
static unsigned long soft_lockup_nmi_warn;
static int __init nowatchdog_setup(char *str)
@@ -312,22 +382,6 @@ static int is_softlockup(unsigned long t
}
/* watchdog detector functions */
-bool is_hardlockup(void)
-{
- unsigned long hrint = __this_cpu_read(hrtimer_interrupts);
-
- if (__this_cpu_read(hrtimer_interrupts_saved) == hrint)
- return true;
-
- __this_cpu_write(hrtimer_interrupts_saved, hrint);
- return false;
-}
-
-static void watchdog_interrupt_count(void)
-{
- __this_cpu_inc(hrtimer_interrupts);
-}
-
static DEFINE_PER_CPU(struct completion, softlockup_completion);
static DEFINE_PER_CPU(struct cpu_stop_work, softlockup_stop_work);
@@ -358,8 +412,7 @@ static enum hrtimer_restart watchdog_tim
if (!watchdog_enabled)
return HRTIMER_NORESTART;
- /* kick the hardlockup detector */
- watchdog_interrupt_count();
+ watchdog_hardlockup_kick();
/* kick the softlockup detector */
if (completion_done(this_cpu_ptr(&softlockup_completion))) {
--- a/kernel/watchdog_perf.c~watchdog-hardlockup-move-perf-hardlockup-checking-panic-to-common-watchdogc
+++ a/kernel/watchdog_perf.c
@@ -20,13 +20,11 @@
#include <asm/irq_regs.h>
#include <linux/perf_event.h>
-static DEFINE_PER_CPU(bool, hard_watchdog_warn);
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 unsigned long hardlockup_allcpu_dumped;
static atomic_t watchdog_cpus = ATOMIC_INIT(0);
notrace void arch_touch_nmi_watchdog(void)
@@ -122,45 +120,7 @@ static void watchdog_overflow_callback(s
return;
}
- /* check for a hardlockup
- * This is done by making sure our timer interrupt
- * is incrementing. The timer interrupt should have
- * fired multiple times before we overflow'd. If it hasn't
- * then this is a good indication the cpu is stuck
- */
- if (is_hardlockup()) {
- int this_cpu = smp_processor_id();
-
- /* only print hardlockups once */
- if (__this_cpu_read(hard_watchdog_warn) == true)
- return;
-
- pr_emerg("Watchdog detected hard LOCKUP on cpu %d\n",
- this_cpu);
- print_modules();
- print_irqtrace_events(current);
- if (regs)
- show_regs(regs);
- else
- dump_stack();
-
- /*
- * Perform all-CPU dump only once to avoid multiple hardlockups
- * generating interleaving traces
- */
- if (sysctl_hardlockup_all_cpu_backtrace &&
- !test_and_set_bit(0, &hardlockup_allcpu_dumped))
- trigger_allbutself_cpu_backtrace();
-
- if (hardlockup_panic)
- nmi_panic(regs, "Hard LOCKUP");
-
- __this_cpu_write(hard_watchdog_warn, true);
- return;
- }
-
- __this_cpu_write(hard_watchdog_warn, false);
- return;
+ watchdog_hardlockup_check(regs);
}
static int hardlockup_detector_event_create(void)
_
Patches currently in -mm which might be from dianders@xxxxxxxxxxxx are
migrate_pages-avoid-blocking-for-io-in-migrate_sync_light.patch
watchdog-perf-define-dummy-watchdog_update_hrtimer_threshold-on-correct-config.patch
watchdog-perf-more-properly-prevent-false-positives-with-turbo-modes.patch
watchdog-hardlockup-add-comments-to-touch_nmi_watchdog.patch
watchdog-perf-rename-watchdog_hldc-to-watchdog_perfc.patch
watchdog-hardlockup-move-perf-hardlockup-checking-panic-to-common-watchdogc.patch
watchdog-hardlockup-style-changes-to-watchdog_hardlockup_check-is_hardlockup.patch
watchdog-hardlockup-add-a-cpu-param-to-watchdog_hardlockup_check.patch
watchdog-hardlockup-move-perf-hardlockup-watchdog-petting-to-watchdogc.patch
watchdog-hardlockup-rename-some-nmi-watchdog-constants-function.patch
watchdog-hardlockup-have-the-perf-hardlockup-use-__weak-functions-more-cleanly.patch
watchdog-hardlockup-detect-hard-lockups-using-secondary-buddy-cpus.patch
watchdog-perf-add-a-weak-function-for-an-arch-to-detect-if-perf-can-use-nmis.patch
arm64-enable-perf-events-based-hard-lockup-detector.patch
