Re: [PATCH v5 02/10] locking/mutex: introduce devm_mutex_init

[Christophe Leroy <christophe.leroy@csgroup.eu>]

Le 12/03/2024 à 00:47, George Stark a écrit :
> [Vous ne recevez pas souvent de courriers de gnstark@salutedevices.com. 
> Découvrez pourquoi ceci est important à 
> https://aka.ms/LearnAboutSenderIdentification ]
> 
> Hello Waiman, Marek
> 
> Thanks for the review.
> 
> I've never used lockdep for debug but it seems preferable to
> keep that feature working. It could be look like this:

For sure it is a must. I'm not used to it either hence my overlook.

> 
> 
> diff --git a/include/linux/mutex.h b/include/linux/mutex.h
> index f7611c092db7..574f6de6084d 100644
> --- a/include/linux/mutex.h
> +++ b/include/linux/mutex.h
> @@ -22,6 +22,8 @@
>   #include <linux/cleanup.h>
>   #include <linux/mutex_types.h>
> 
> +struct device;
> +
>   #ifdef CONFIG_DEBUG_LOCK_ALLOC
>   # define __DEP_MAP_MUTEX_INITIALIZER(lockname)                        \
>                 , .dep_map = {                                  \
> @@ -115,10 +117,31 @@ do 
> {                                                      \
> 
>   #ifdef CONFIG_DEBUG_MUTEXES
> 
> +int debug_devm_mutex_init(struct device *dev, struct mutex *lock);
> +
> +#define devm_mutex_init(dev, mutex)                    \
> +({                                                     \
> +       int ret;                                        \
> +       mutex_init(mutex);                              \
> +       ret = debug_devm_mutex_init(dev, mutex);        \
> +       ret;                                            \
> +})
> +

I think it would be preferable to minimise the number of macros.

If I were you I would keep your devm_mutex_init() as is but rename it 
__devm_mutex_init() and just remove the mutex_init() from it, then add 
only one macro that works independant of CONFIG_DEBUG_MUTEXES:

#define devm_mutex_init(dev, mutex)	\
({					\
	mutex_init(mutex);		\
	__devm_mutex_init(dev, mutex);	\
})

With that, no need of a second version of the macro and no need for the 
typecheck either.

Note the __ which is a clear indication that allthough that function is 
declared in public mutex.h, it is not meant to be used outside of it.



>   void mutex_destroy(struct mutex *lock);
> 
>   #else
> 
> +/*
> +* When CONFIG_DEBUG_MUTEXES is off mutex_destroy is just a nop so
> +* there's no really need to register it in devm subsystem.
> +*/
> +#define devm_mutex_init(dev, mutex)                    \
> +({                                                     \
> +       typecheck(struct device *, dev);                \
> +       mutex_init(mutex);                              \
> +       0;                                              \
> +})
> +
>   static inline void mutex_destroy(struct mutex *lock) {}
> 
>   #endif
> diff --git a/kernel/locking/mutex-debug.c b/kernel/locking/mutex-debug.c
> index bc8abb8549d2..967a5367c79a 100644
> --- a/kernel/locking/mutex-debug.c
> +++ b/kernel/locking/mutex-debug.c
> @@ -19,6 +19,7 @@
>   #include <linux/kallsyms.h>
>   #include <linux/interrupt.h>
>   #include <linux/debug_locks.h>
> +#include <linux/device.h>
> 
>   #include "mutex.h"
> 
> @@ -89,6 +90,16 @@ void debug_mutex_init(struct mutex *lock, const char
> *name,
>         lock->magic = lock;
>   }
> 
> +static void devm_mutex_release(void *res)
> +{
> +       mutex_destroy(res);
> +}
> +
> +int debug_devm_mutex_init(struct device *dev, struct mutex *lock)
> +{
> +       return devm_add_action_or_reset(dev, devm_mutex_release, lock);
> +}
> +
>   /***
>    * mutex_destroy - mark a mutex unusable
>    * @lock: the mutex to be destroyed
> -- 
> 2.25.1
> 
> 
> 
> And now I would drop the the refactoring patch with moving down
> mutex_destroy. devm block is big enough to be declared standalone.
> 
> 
> On 3/7/24 19:44, Marek Behún wrote:
>> On Thu, 7 Mar 2024 08:39:46 -0500
>> Waiman Long <longman@redhat.com> wrote:
>>
>>> On 3/7/24 04:56, Marek Behún wrote:
>>>> On Thu, Mar 07, 2024 at 05:40:26AM +0300, George Stark wrote:
>>>>> Using of devm API leads to a certain order of releasing resources.
>>>>> So all dependent resources which are not devm-wrapped should be 
>>>>> deleted
>>>>> with respect to devm-release order. Mutex is one of such objects that
>>>>> often is bound to other resources and has no own devm wrapping.
>>>>> Since mutex_destroy() actually does nothing in non-debug builds
>>>>> frequently calling mutex_destroy() is just ignored which is safe 
>>>>> for now
>>>>> but wrong formally and can lead to a problem if mutex_destroy() 
>>>>> will be
>>>>> extended so introduce devm_mutex_init()
>>>>>
>>>>> Signed-off-by: George Stark <gnstark@salutedevices.com>
>>>>> Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
>>>>> ---
>>>>>    Hello Christophe. Hope you don't mind I put you SoB tag because 
>>>>> you helped alot
>>>>>    to make this patch happen.
>>>>>
>>>>>    include/linux/mutex.h        | 13 +++++++++++++
>>>>>    kernel/locking/mutex-debug.c | 22 ++++++++++++++++++++++
>>>>>    2 files changed, 35 insertions(+)
>>>>>
>>>>> diff --git a/include/linux/mutex.h b/include/linux/mutex.h
>>>>> index f7611c092db7..9bcf72cb941a 100644
>>>>> --- a/include/linux/mutex.h
>>>>> +++ b/include/linux/mutex.h
>>>>> @@ -22,6 +22,8 @@
>>>>>    #include <linux/cleanup.h>
>>>>>    #include <linux/mutex_types.h>
>>>>>
>>>>> +struct device;
>>>>> +
>>>>>    #ifdef CONFIG_DEBUG_LOCK_ALLOC
>>>>>    # define __DEP_MAP_MUTEX_INITIALIZER(lockname)                  \
>>>>>                    , .dep_map = {                                  \
>>>>> @@ -115,10 +117,21 @@ do 
>>>>> {                                                 \
>>>>>
>>>>>    #ifdef CONFIG_DEBUG_MUTEXES
>>>>>
>>>>> +int devm_mutex_init(struct device *dev, struct mutex *lock);
>>>>>    void mutex_destroy(struct mutex *lock);
>>>>>
>>>>>    #else
>>>>>
>>>>> +static inline int devm_mutex_init(struct device *dev, struct mutex 
>>>>> *lock)
>>>>> +{
>>>>> +  /*
>>>>> +   * since mutex_destroy is nop actually there's no need to 
>>>>> register it
>>>>> +   * in devm subsystem.
>>>>> +   */
>>>>> +  mutex_init(lock);
>>>>> +  return 0;
>>>>> +}
>>>>> +
>>>>>    static inline void mutex_destroy(struct mutex *lock) {}
>>>>>
>>>>>    #endif
>>>>> diff --git a/kernel/locking/mutex-debug.c 
>>>>> b/kernel/locking/mutex-debug.c
>>>>> index bc8abb8549d2..c9efab1a8026 100644
>>>>> --- a/kernel/locking/mutex-debug.c
>>>>> +++ b/kernel/locking/mutex-debug.c
>>>>> @@ -19,6 +19,7 @@
>>>>>    #include <linux/kallsyms.h>
>>>>>    #include <linux/interrupt.h>
>>>>>    #include <linux/debug_locks.h>
>>>>> +#include <linux/device.h>
>>>>>
>>>>>    #include "mutex.h"
>>>>>
>>>>> @@ -104,3 +105,24 @@ void mutex_destroy(struct mutex *lock)
>>>>>    }
>>>>>
>>>>>    EXPORT_SYMBOL_GPL(mutex_destroy);
>>>>> +
>>>>> +static void devm_mutex_release(void *res)
>>>>> +{
>>>>> +  mutex_destroy(res);
>>>>> +}
>>>>> +
>>>>> +/**
>>>>> + * devm_mutex_init - Resource-managed mutex initialization
>>>>> + * @dev:  Device which lifetime mutex is bound to
>>>>> + * @lock: Pointer to a mutex
>>>>> + *
>>>>> + * Initialize mutex which is automatically destroyed when the 
>>>>> driver is detached.
>>>>> + *
>>>>> + * Returns: 0 on success or a negative error code on failure.
>>>>> + */
>>>>> +int devm_mutex_init(struct device *dev, struct mutex *lock)
>>>>> +{
>>>>> +  mutex_init(lock);
>>>>> +  return devm_add_action_or_reset(dev, devm_mutex_release, lock);
>>>>> +}
>>>>> +EXPORT_SYMBOL_GPL(devm_mutex_init);
>>>> Hi George,
>>>>
>>>> look at
>>>> https://lore.kernel.org/lkml/7013bf9e-2663-4613-ae61-61872e81355b@redhat.com/
>>>> where Matthew and Hans explain that devm_mutex_init needs to be a macro
>>>> because of the static lockdep key.
>>>>
>>>> so this should be something like:
>>>>
>>>> static inline int __devm_mutex_init(struct device *dev, struct mutex 
>>>> *mutex,
>>>>                                 const char *name,
>>>>                                 struct lock_class_key *key)
>>>> {
>>>>     __mutex_init(mutex, name, key);
>>>>     return devm_add_action_or_reset(dev, devm_mutex_release, mutex);
>>>> }
>>>>
>>>> #define devm_mutex_init(dev, mutex)                         \
>>>> do {                                                                \
>>>>     static struct lock_class_key __key;                     \
>>>>                                                             \
>>>>     __devm_mutex_init(dev, (mutex), #mutex, &__key);        \
>>>> } while (0);
>>>>
>>>>
>>>> Marek
>>>
>>> Making devm_mutex_init() a function will make all the devm_mutex share
>>> the same lockdep key. Making it a macro will make each caller of
>>> devm_mutex_init() have a distinct lockdep key. It all depends on whether
>>> all the devm_mutexes have the same lock usage pattern or not and whether
>>> it is possible for one devm_mutex to be nested inside another. So either
>>> way can be fine depending on the mutex usage pattern. My suggestion is
>>> to use a function, if possible, unless it will cause a false positive
>>> lockdep splat as there is a limit on the maximum # of lockdep keys that
>>> can be used.
>>
>> devm_mutex_init() should behave like other similar function
>> initializing stuff with resource management. I.e. it should behave like
>> mutex_init(), but with resource management.
>>
>> mutex_init() is a macro generating static lockdep key for each instance,
>> so devm_mutex_init() should also generate static lockdep key for each
>> instance.
>>
>> Marek
> 
> -- 
> Best regards
> George