Commit f44310b9 authored by Wang YanQing's avatar Wang YanQing Committed by Ingo Molnar

smp: Fix SMP function call empty cpu mask race

I get the following warning every day with v3.7, once or
twice a day:

  [ 2235.186027] WARNING: at /mnt/sda7/kernel/linux/arch/x86/kernel/apic/ipi.c:109 default_send_IPI_mask_logical+0x2f/0xb8()

As explained by Linus as well:

 |
 | Once we've done the "list_add_rcu()" to add it to the
 | queue, we can have (another) IPI to the target CPU that can
 | now see it and clear the mask.
 |
 | So by the time we get to actually send the IPI, the mask might
 | have been cleared by another IPI.
 |

This patch also fixes a system hang problem, if the data->cpumask
gets cleared after passing this point:

        if (WARN_ONCE(!mask, "empty IPI mask"))
                return;

then the problem in commit 83d349f3 ("x86: don't send an IPI to
the empty set of CPU's") will happen again.
Signed-off-by: default avatarWang YanQing <udknight@gmail.com>
Acked-by: default avatarLinus Torvalds <torvalds@linux-foundation.org>
Acked-by: default avatarJan Beulich <jbeulich@suse.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: peterz@infradead.org
Cc: mina86@mina86.org
Cc: srivatsa.bhat@linux.vnet.ibm.com
Cc: <stable@kernel.org>
Link: http://lkml.kernel.org/r/20130126075357.GA3205@udknight
[ Tidied up the changelog and the comment in the code. ]
Signed-off-by: default avatarIngo Molnar <mingo@kernel.org>
parent c903f045
...@@ -33,6 +33,7 @@ struct call_function_data { ...@@ -33,6 +33,7 @@ struct call_function_data {
struct call_single_data csd; struct call_single_data csd;
atomic_t refs; atomic_t refs;
cpumask_var_t cpumask; cpumask_var_t cpumask;
cpumask_var_t cpumask_ipi;
}; };
static DEFINE_PER_CPU_SHARED_ALIGNED(struct call_function_data, cfd_data); static DEFINE_PER_CPU_SHARED_ALIGNED(struct call_function_data, cfd_data);
...@@ -56,6 +57,9 @@ hotplug_cfd(struct notifier_block *nfb, unsigned long action, void *hcpu) ...@@ -56,6 +57,9 @@ hotplug_cfd(struct notifier_block *nfb, unsigned long action, void *hcpu)
if (!zalloc_cpumask_var_node(&cfd->cpumask, GFP_KERNEL, if (!zalloc_cpumask_var_node(&cfd->cpumask, GFP_KERNEL,
cpu_to_node(cpu))) cpu_to_node(cpu)))
return notifier_from_errno(-ENOMEM); return notifier_from_errno(-ENOMEM);
if (!zalloc_cpumask_var_node(&cfd->cpumask_ipi, GFP_KERNEL,
cpu_to_node(cpu)))
return notifier_from_errno(-ENOMEM);
break; break;
#ifdef CONFIG_HOTPLUG_CPU #ifdef CONFIG_HOTPLUG_CPU
...@@ -65,6 +69,7 @@ hotplug_cfd(struct notifier_block *nfb, unsigned long action, void *hcpu) ...@@ -65,6 +69,7 @@ hotplug_cfd(struct notifier_block *nfb, unsigned long action, void *hcpu)
case CPU_DEAD: case CPU_DEAD:
case CPU_DEAD_FROZEN: case CPU_DEAD_FROZEN:
free_cpumask_var(cfd->cpumask); free_cpumask_var(cfd->cpumask);
free_cpumask_var(cfd->cpumask_ipi);
break; break;
#endif #endif
}; };
...@@ -526,6 +531,12 @@ void smp_call_function_many(const struct cpumask *mask, ...@@ -526,6 +531,12 @@ void smp_call_function_many(const struct cpumask *mask,
return; return;
} }
/*
* After we put an entry into the list, data->cpumask
* may be cleared again when another CPU sends another IPI for
* a SMP function call, so data->cpumask will be zero.
*/
cpumask_copy(data->cpumask_ipi, data->cpumask);
raw_spin_lock_irqsave(&call_function.lock, flags); raw_spin_lock_irqsave(&call_function.lock, flags);
/* /*
* Place entry at the _HEAD_ of the list, so that any cpu still * Place entry at the _HEAD_ of the list, so that any cpu still
...@@ -549,7 +560,7 @@ void smp_call_function_many(const struct cpumask *mask, ...@@ -549,7 +560,7 @@ void smp_call_function_many(const struct cpumask *mask,
smp_mb(); smp_mb();
/* Send a message to all CPUs in the map */ /* Send a message to all CPUs in the map */
arch_send_call_function_ipi_mask(data->cpumask); arch_send_call_function_ipi_mask(data->cpumask_ipi);
/* Optionally wait for the CPUs to complete */ /* Optionally wait for the CPUs to complete */
if (wait) if (wait)
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment