]> git.ipfire.org Git - thirdparty/kernel/stable.git/commitdiff
clockevent: Prevent dead lock on clockevents_lock
authorSuresh Siddha <suresh.b.siddha@intel.com>
Thu, 22 Apr 2010 09:47:51 +0000 (11:47 +0200)
committerGreg Kroah-Hartman <gregkh@suse.de>
Wed, 26 May 2010 21:27:07 +0000 (14:27 -0700)
This is a merge of two mainline commits, intended
for stable@kernel.org submission for 2.6.27 kernel.

commit f833bab87fca5c3ce13778421b1365845843b976
and

commit 918aae42aa9b611a3663b16ae849fdedc67c2292
Changelog of both:

    Currently clockevents_notify() is called with interrupts enabled at
    some places and interrupts disabled at some other places.

    This results in a deadlock in this scenario.

    cpu A holds clockevents_lock in clockevents_notify() with irqs enabled
    cpu B waits for clockevents_lock in clockevents_notify() with irqs disabled
    cpu C doing set_mtrr() which will try to rendezvous of all the cpus.

    This will result in C and A come to the rendezvous point and waiting
    for B. B is stuck forever waiting for the spinlock and thus not
    reaching the rendezvous point.

    Fix the clockevents code so that clockevents_lock is taken with
    interrupts disabled and thus avoid the above deadlock.

    Also call lapic_timer_propagate_broadcast() on the destination cpu so
    that we avoid calling smp_call_function() in the clockevents notifier
    chain.

    This issue left us wondering if we need to change the MTRR rendezvous
    logic to use stop machine logic (instead of smp_call_function) or add
    a check in spinlock debug code to see if there are other spinlocks
    which gets taken under both interrupts enabled/disabled conditions.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: "Brown Len" <len.brown@intel.com>
Cc: stable@kernel.org
    LKML-Reference: <1250544899.2709.210.camel@sbs-t61.sc.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
    I got following warning on ia64 box:
      In function 'acpi_processor_power_verify':
      642: warning: passing argument 2 of 'smp_call_function_single' from
      incompatible pointer type

    This smp_call_function_single() was introduced by a commit
    f833bab87fca5c3ce13778421b1365845843b976:

    The problem is that the lapic_timer_propagate_broadcast() has 2 versions:
    One is real code that modified in the above commit, and the other is NOP
    code that used when !ARCH_APICTIMER_STOPS_ON_C3:

      static void lapic_timer_propagate_broadcast(struct acpi_processor *pr) { }

    So I got warning because of !ARCH_APICTIMER_STOPS_ON_C3.

    We really want to do nothing here on !ARCH_APICTIMER_STOPS_ON_C3, so
    modify lapic_timer_propagate_broadcast() of real version to use
    smp_call_function_single() in it.

Signed-off-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Acked-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
arch/x86/kernel/process.c
drivers/acpi/processor_idle.c
kernel/time/clockevents.c
kernel/time/tick-broadcast.c

index 876e91890777ae9758e5efcb7dc72075a26e341e..f459e3ee047326f6ef617a5b5f3cc085fad6e59b 100644 (file)
@@ -283,16 +283,12 @@ static void c1e_idle(void)
                if (!cpu_isset(cpu, c1e_mask)) {
                        cpu_set(cpu, c1e_mask);
                        /*
-                        * Force broadcast so ACPI can not interfere. Needs
-                        * to run with interrupts enabled as it uses
-                        * smp_function_call.
+                        * Force broadcast so ACPI can not interfere.
                         */
-                       local_irq_enable();
                        clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_FORCE,
                                           &cpu);
                        printk(KERN_INFO "Switch to broadcast mode on CPU%d\n",
                               cpu);
-                       local_irq_disable();
                }
                clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_ENTER, &cpu);
 
index 5639e27cb4973ef41e82a46703741878bddff8f9..94483a2a61f74ec91c4f7b880ce85a3bb49b72e9 100644 (file)
@@ -317,8 +317,9 @@ static void acpi_timer_check_state(int state, struct acpi_processor *pr,
                pr->power.timer_broadcast_on_state = state;
 }
 
-static void acpi_propagate_timer_broadcast(struct acpi_processor *pr)
+static void __lapic_timer_propagate_broadcast(void *arg)
 {
+       struct acpi_processor *pr = (struct acpi_processor *) arg;
        unsigned long reason;
 
        reason = pr->power.timer_broadcast_on_state < INT_MAX ?
@@ -327,6 +328,12 @@ static void acpi_propagate_timer_broadcast(struct acpi_processor *pr)
        clockevents_notify(reason, &pr->id);
 }
 
+static void lapic_timer_propagate_broadcast(struct acpi_processor *pr)
+{
+       smp_call_function_single(pr->id, __lapic_timer_propagate_broadcast,
+                                (void *)pr, 1);
+}
+
 /* Power(C) State timer broadcast control */
 static void acpi_state_timer_broadcast(struct acpi_processor *pr,
                                       struct acpi_processor_cx *cx,
@@ -347,7 +354,7 @@ static void acpi_state_timer_broadcast(struct acpi_processor *pr,
 
 static void acpi_timer_check_state(int state, struct acpi_processor *pr,
                                   struct acpi_processor_cx *cstate) { }
-static void acpi_propagate_timer_broadcast(struct acpi_processor *pr) { }
+static void lapic_timer_propagate_broadcast(struct acpi_processor *pr) { }
 static void acpi_state_timer_broadcast(struct acpi_processor *pr,
                                       struct acpi_processor_cx *cx,
                                       int broadcast)
@@ -1177,7 +1184,7 @@ static int acpi_processor_power_verify(struct acpi_processor *pr)
                        working++;
        }
 
-       acpi_propagate_timer_broadcast(pr);
+       lapic_timer_propagate_broadcast(pr);
 
        return (working);
 }
index f8d968063cea38e348f63590ce73d409385cc5e9..ae51d5e22c26291d57e1f529dd60087b6d0d4179 100644 (file)
@@ -124,11 +124,12 @@ int clockevents_program_event(struct clock_event_device *dev, ktime_t expires,
  */
 int clockevents_register_notifier(struct notifier_block *nb)
 {
+       unsigned long flags;
        int ret;
 
-       spin_lock(&clockevents_lock);
+       spin_lock_irqsave(&clockevents_lock, flags);
        ret = raw_notifier_chain_register(&clockevents_chain, nb);
-       spin_unlock(&clockevents_lock);
+       spin_unlock_irqrestore(&clockevents_lock, flags);
 
        return ret;
 }
@@ -165,6 +166,8 @@ static void clockevents_notify_released(void)
  */
 void clockevents_register_device(struct clock_event_device *dev)
 {
+       unsigned long flags;
+
        BUG_ON(dev->mode != CLOCK_EVT_MODE_UNUSED);
        /*
         * A nsec2cyc multiplicator of 0 is invalid and we'd crash
@@ -175,13 +178,13 @@ void clockevents_register_device(struct clock_event_device *dev)
                WARN_ON(1);
        }
 
-       spin_lock(&clockevents_lock);
+       spin_lock_irqsave(&clockevents_lock, flags);
 
        list_add(&dev->list, &clockevent_devices);
        clockevents_do_notify(CLOCK_EVT_NOTIFY_ADD, dev);
        clockevents_notify_released();
 
-       spin_unlock(&clockevents_lock);
+       spin_unlock_irqrestore(&clockevents_lock, flags);
 }
 
 /*
@@ -228,8 +231,9 @@ void clockevents_exchange_device(struct clock_event_device *old,
 void clockevents_notify(unsigned long reason, void *arg)
 {
        struct list_head *node, *tmp;
+       unsigned long flags;
 
-       spin_lock(&clockevents_lock);
+       spin_lock_irqsave(&clockevents_lock, flags);
        clockevents_do_notify(reason, arg);
 
        switch (reason) {
@@ -244,7 +248,7 @@ void clockevents_notify(unsigned long reason, void *arg)
        default:
                break;
        }
-       spin_unlock(&clockevents_lock);
+       spin_unlock_irqrestore(&clockevents_lock, flags);
 }
 EXPORT_SYMBOL_GPL(clockevents_notify);
 #endif
index cb01cd8f919b2752cfb7d97b4dc9ea4e87834a97..d1b32166be6f7a58c552def5e36c28e4a6630432 100644 (file)
@@ -205,11 +205,11 @@ static void tick_handle_periodic_broadcast(struct clock_event_device *dev)
  * Powerstate information: The system enters/leaves a state, where
  * affected devices might stop
  */
-static void tick_do_broadcast_on_off(void *why)
+static void tick_do_broadcast_on_off(unsigned long *reason)
 {
        struct clock_event_device *bc, *dev;
        struct tick_device *td;
-       unsigned long flags, *reason = why;
+       unsigned long flags;
        int cpu, bc_stopped;
 
        spin_lock_irqsave(&tick_broadcast_lock, flags);
@@ -276,8 +276,7 @@ void tick_broadcast_on_off(unsigned long reason, int *oncpu)
                printk(KERN_ERR "tick-broadcast: ignoring broadcast for "
                       "offline CPU #%d\n", *oncpu);
        else
-               smp_call_function_single(*oncpu, tick_do_broadcast_on_off,
-                                        &reason, 1);
+               tick_do_broadcast_on_off(&reason);
 }
 
 /*