[PATCH 11/33] cpuset: Provide lockdep check for cpuset lock held

[Chen Ridong]

On 2025/11/1 0:08, Frederic Weisbecker wrote:
> Le Tue, Oct 14, 2025 at 09:29:25PM +0800, Chen Ridong a écrit :
>>
>>
>> On 2025/10/14 4:31, Frederic Weisbecker wrote:
>>> cpuset modifies partitions, including isolated, while holding the cpuset
>>> mutex.
>>>
>>> This means that holding the cpuset mutex is safe to synchronize against
>>> housekeeping cpumask changes.
>>>
>>> Provide a lockdep check to validate that.
>>>
>>> Signed-off-by: Frederic Weisbecker <frederic at kernel.org>
>>> ---
>>>  include/linux/cpuset.h | 2 ++
>>>  kernel/cgroup/cpuset.c | 7 +++++++
>>>  2 files changed, 9 insertions(+)
>>>
>>> diff --git a/include/linux/cpuset.h b/include/linux/cpuset.h
>>> index 2ddb256187b5..051d36fec578 100644
>>> --- a/include/linux/cpuset.h
>>> +++ b/include/linux/cpuset.h
>>> @@ -18,6 +18,8 @@
>>>  #include <linux/mmu_context.h>
>>>  #include <linux/jump_label.h>
>>>  
>>> +extern bool lockdep_is_cpuset_held(void);
>>> +
>>>  #ifdef CONFIG_CPUSETS
>>>  
>>>  /*
>>> diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
>>> index 8595f1eadf23..aa1ac7bcf2ea 100644
>>> --- a/kernel/cgroup/cpuset.c
>>> +++ b/kernel/cgroup/cpuset.c
>>> @@ -279,6 +279,13 @@ void cpuset_full_unlock(void)
>>>  	cpus_read_unlock();
>>>  }
>>>  
>>> +#ifdef CONFIG_LOCKDEP
>>> +bool lockdep_is_cpuset_held(void)
>>> +{
>>> +	return lockdep_is_held(&cpuset_mutex);
>>> +}
>>> +#endif
>>> +
>>>  static DEFINE_SPINLOCK(callback_lock);
>>>  
>>>  void cpuset_callback_lock_irq(void)
>>
>> Is the lockdep_is_cpuset_held function actually being used?
>> If CONFIG_LOCKDEP is disabled, compilation would fail with an "undefined reference to
>> lockdep_is_cpuset_held" error.
> 
> Although counter-intuitive, this is how the lockdep_is_held() functions family
> do work.
> 
> This allows this kind of trick:
> 
> if (IS_ENABLED(CONFIG_LOCKDEP))
>    WARN_ON_ONCE(!lockdep_is_held(&some_lock))
> 
> This works during the compilation because the prototype of lockdep_is_held()
> is declared. And since the IS_ENABLED() is resolved during compilation as well,
> the conditional code is wiped out and therefore not linked. As a result the
> linker doesn't even look for the definition of lockdep_is_held() and we don't
> need to define an off case that would return a wrong assumption.
> 
> Thanks.
> 

Thank you for your explanation

-- 
Best regards,
Ridong