From: Waiman Long longman@redhat.com
[ Upstream commit fe37c699ae3eed6e02ee55fbf5cb9ceb7fcfd76c ]
Depending on the type of panics, it was found that the __register_nmi_handler() function can be called in NMI context from nmi_shootdown_cpus() leading to a lockdep splat:
WARNING: inconsistent lock state inconsistent {INITIAL USE} -> {IN-NMI} usage.
lock(&nmi_desc[0].lock); <Interrupt> lock(&nmi_desc[0].lock);
Call Trace: _raw_spin_lock_irqsave __register_nmi_handler nmi_shootdown_cpus kdump_nmi_shootdown_cpus native_machine_crash_shutdown __crash_kexec
In this particular case, the following panic message was printed before:
Kernel panic - not syncing: Fatal hardware error!
This message seemed to be given out from __ghes_panic() running in NMI context.
The __register_nmi_handler() function which takes the nmi_desc lock with irq disabled shouldn't be called from NMI context as this can lead to deadlock.
The nmi_shootdown_cpus() function can only be invoked once. After the first invocation, all other CPUs should be stuck in the newly added crash_nmi_callback() and cannot respond to a second NMI.
Fix it by adding a new emergency NMI handler to the nmi_desc structure and provide a new set_emergency_nmi_handler() helper to set crash_nmi_callback() in any context. The new emergency handler will preempt other handlers in the linked list. That will eliminate the need to take any lock and serve the panic in NMI use case.
Signed-off-by: Waiman Long longman@redhat.com Signed-off-by: Ingo Molnar mingo@kernel.org Acked-by: Rik van Riel riel@surriel.com Cc: Thomas Gleixner tglx@linutronix.de Link: https://lore.kernel.org/r/20250206191844.131700-1-longman@redhat.com Signed-off-by: Sasha Levin sashal@kernel.org --- arch/x86/include/asm/nmi.h | 2 ++ arch/x86/kernel/nmi.c | 42 ++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/reboot.c | 10 +++------ 3 files changed, 47 insertions(+), 7 deletions(-)
diff --git a/arch/x86/include/asm/nmi.h b/arch/x86/include/asm/nmi.h index 41a0ebb699ec6..f677382093f36 100644 --- a/arch/x86/include/asm/nmi.h +++ b/arch/x86/include/asm/nmi.h @@ -56,6 +56,8 @@ int __register_nmi_handler(unsigned int, struct nmiaction *);
void unregister_nmi_handler(unsigned int, const char *);
+void set_emergency_nmi_handler(unsigned int type, nmi_handler_t handler); + void stop_nmi(void); void restart_nmi(void); void local_touch_nmi(void); diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c index ed163c8c8604e..9a95d00f14233 100644 --- a/arch/x86/kernel/nmi.c +++ b/arch/x86/kernel/nmi.c @@ -40,8 +40,12 @@ #define CREATE_TRACE_POINTS #include <trace/events/nmi.h>
+/* + * An emergency handler can be set in any context including NMI + */ struct nmi_desc { raw_spinlock_t lock; + nmi_handler_t emerg_handler; struct list_head head; };
@@ -132,9 +136,22 @@ static void nmi_check_duration(struct nmiaction *action, u64 duration) static int nmi_handle(unsigned int type, struct pt_regs *regs) { struct nmi_desc *desc = nmi_to_desc(type); + nmi_handler_t ehandler; struct nmiaction *a; int handled=0;
+ /* + * Call the emergency handler, if set + * + * In the case of crash_nmi_callback() emergency handler, it will + * return in the case of the crashing CPU to enable it to complete + * other necessary crashing actions ASAP. Other handlers in the + * linked list won't need to be run. + */ + ehandler = desc->emerg_handler; + if (ehandler) + return ehandler(type, regs); + rcu_read_lock();
/* @@ -224,6 +241,31 @@ void unregister_nmi_handler(unsigned int type, const char *name) } EXPORT_SYMBOL_GPL(unregister_nmi_handler);
+/** + * set_emergency_nmi_handler - Set emergency handler + * @type: NMI type + * @handler: the emergency handler to be stored + * + * Set an emergency NMI handler which, if set, will preempt all the other + * handlers in the linked list. If a NULL handler is passed in, it will clear + * it. It is expected that concurrent calls to this function will not happen + * or the system is screwed beyond repair. + */ +void set_emergency_nmi_handler(unsigned int type, nmi_handler_t handler) +{ + struct nmi_desc *desc = nmi_to_desc(type); + + if (WARN_ON_ONCE(desc->emerg_handler == handler)) + return; + desc->emerg_handler = handler; + + /* + * Ensure the emergency handler is visible to other CPUs before + * function return + */ + smp_wmb(); +} + static void pci_serr_error(unsigned char reason, struct pt_regs *regs) { diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index dc1dd3f3e67fc..9aaac1f9f45b5 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -926,15 +926,11 @@ void nmi_shootdown_cpus(nmi_shootdown_cb callback) shootdown_callback = callback;
atomic_set(&waiting_for_crash_ipi, num_online_cpus() - 1); - /* Would it be better to replace the trap vector here? */ - if (register_nmi_handler(NMI_LOCAL, crash_nmi_callback, - NMI_FLAG_FIRST, "crash")) - return; /* Return what? */ + /* - * Ensure the new callback function is set before sending - * out the NMI + * Set emergency handler to preempt other handlers. */ - wmb(); + set_emergency_nmi_handler(NMI_LOCAL, crash_nmi_callback);
apic_send_IPI_allbutself(NMI_VECTOR);