On Wed, Sep 02, 2020 at 05:58:55PM +0200, Ulf Hansson wrote:
On Wed, 2 Sep 2020 at 14:14, peterz@infradead.org wrote:
On Wed, Sep 02, 2020 at 09:03:37AM +0200, Ulf Hansson wrote:
Lots of cpuidle drivers are using CPU_PM notifiers (grep for cpu_pm_enter and you will see) from their idlestates ->enter() callbacks. And for those we are already calling rcu_irq_enter_irqson|off() in cpu_pm_notify() when firing them.
Yeah, that particular trainwreck is on my todo list already ... then again, that list is forever overflowing.
I'm thinking cpu_pm_unregister_notifier() is not a common thing? The few I looked at seem to suggest 'never' is a good approximation.
The trend is that drivers are turning into regular modules that may also need to manage "->remove()", which may mean unregistering the notifier. Of course, I don't know for sure whether that becomes a problem, but it seems quite limiting.
You can pin modules, once they're loaded they can never be removed again.
Anyway, the below should 'work', I think.
--- diff --git a/kernel/cpu_pm.c b/kernel/cpu_pm.c index f7e1d0eccdbc..72804e0883d5 100644 --- a/kernel/cpu_pm.c +++ b/kernel/cpu_pm.c @@ -12,21 +12,18 @@ #include <linux/notifier.h> #include <linux/spinlock.h> #include <linux/syscore_ops.h> +#include <linux/cpu.h> +#include <linux/smp.h>
-static ATOMIC_NOTIFIER_HEAD(cpu_pm_notifier_chain); +static RAW_NOTIFIER_HEAD(cpu_pm_notifier_chain); +static DEFINE_SPINLOCK(cpu_pm_lock);
static int cpu_pm_notify(enum cpu_pm_event event) { int ret;
- /* - * atomic_notifier_call_chain has a RCU read critical section, which - * could be disfunctional in cpu idle. Copy RCU_NONIDLE code to let - * RCU know this. - */ - rcu_irq_enter_irqson(); - ret = atomic_notifier_call_chain(&cpu_pm_notifier_chain, event, NULL); - rcu_irq_exit_irqson(); + lockdep_assert_irqs_disabled(); + ret = raw_notifier_call_chain(&cpu_pm_notifier_chain, event, NULL);
return notifier_to_errno(ret); } @@ -35,9 +32,8 @@ static int cpu_pm_notify_robust(enum cpu_pm_event event_up, enum cpu_pm_event ev { int ret;
- rcu_irq_enter_irqson(); - ret = atomic_notifier_call_chain_robust(&cpu_pm_notifier_chain, event_up, event_down, NULL); - rcu_irq_exit_irqson(); + lockdep_assert_irqs_disabled(); + ret = raw_notifier_call_chain_robust(&cpu_pm_notifier_chain, event_up, event_down, NULL);
return notifier_to_errno(ret); } @@ -54,10 +50,28 @@ static int cpu_pm_notify_robust(enum cpu_pm_event event_up, enum cpu_pm_event ev */ int cpu_pm_register_notifier(struct notifier_block *nb) { - return atomic_notifier_chain_register(&cpu_pm_notifier_chain, nb); + unsigned long flags; + int ret; + + spin_lock_irqsave(&cpu_pm_lock, flags); + ret = raw_notifier_chain_register(&cpu_pm_notifier_chain, nb); + spin_unlock_irqrestore(&cpu_pm_lock, flags); + + return ret; } EXPORT_SYMBOL_GPL(cpu_pm_register_notifier);
+static bool __is_idle_cpu(int cpu, void *info) +{ + /* + * Racy as heck, however if we fail to see an idle task, it must be + * after we removed our element, so all is fine. + */ + return is_idle_task(curr_task(cpu)); +} + +static void __nop_func(void *arg) { } + /** * cpu_pm_unregister_notifier - unregister a driver with cpu_pm * @nb: notifier block to be unregistered @@ -69,7 +83,30 @@ EXPORT_SYMBOL_GPL(cpu_pm_register_notifier); */ int cpu_pm_unregister_notifier(struct notifier_block *nb) { - return atomic_notifier_chain_unregister(&cpu_pm_notifier_chain, nb); + unsigned long flags; + int ret, cpu; + + spin_lock_irqsave(&cpu_pm_lock, flags); + ret = raw_notifier_chain_unregister(&cpu_pm_notifier_chain, nb); + spin_unlock_irqrestore(&cpu_pm_lock, flags); + + /* + * Orders the removal above vs the __is_idle_cpu() test below. Matches + * schedule() switching to the idle task. + * + * Ensures that if we miss an idle task, it must be after the removal. + */ + smp_mb(); + + /* + * IPI all idle CPUs, this guarantees that no CPU is currently + * iterating the notifier list. + */ + cpus_read_lock(); + on_each_cpu_cond(__is_idle_cpu, __nop_func, NULL, 1); + cpus_read_unlock(); + + return ret; } EXPORT_SYMBOL_GPL(cpu_pm_unregister_notifier);