Re: [tip: locking/urgent] locking/ww_mutex: Treat ww_mutex_lock() like a trylock

[Waiman Long]

On 3/17/21 1:45 PM, Peter Zijlstra wrote:
> On Wed, Mar 17, 2021 at 06:40:27PM +0100, Peter Zijlstra wrote:
> > On Wed, Mar 17, 2021 at 05:48:48PM +0100, Peter Zijlstra wrote:
> >
> > > I think you'll find that if you use ww_mutex_init() it'll all work. Let
> > > me go and zap this patch, and then I'll try and figure out why
> > > DEFINE_WW_MUTEX() is buggered.
> > Moo, I can't get the compiler to do as I want :/
> >
> > The below is so close but doesn't actually compile.. Maybe we should
> > just give up on DEFINE_WW_MUTEX and simply remove it.
> >
> > ---
> > diff --git a/include/linux/mutex.h b/include/linux/mutex.h
> > index 0cd631a19727..85f50538f26a 100644
> > --- a/include/linux/mutex.h
> > +++ b/include/linux/mutex.h
> > @@ -129,12 +129,15 @@ do {									\
> >   # define __DEP_MAP_MUTEX_INITIALIZER(lockname)
> >   #endif
> >   
> > -#define __MUTEX_INITIALIZER(lockname) \
> > +#define ___MUTEX_INITIALIZER(lockname, depmap) \
> >   		{ .owner = ATOMIC_LONG_INIT(0) \
> >   		, .wait_lock = __SPIN_LOCK_UNLOCKED(lockname.wait_lock) \
> >   		, .wait_list = LIST_HEAD_INIT(lockname.wait_list) \
> >   		__DEBUG_MUTEX_INITIALIZER(lockname) \
> > -		__DEP_MAP_MUTEX_INITIALIZER(lockname) }
> > +		depmap }
> > +
> > +#define __MUTEX_INITIALIZER(lockname) \
> > +		___MUTEX_INITIALIZER(lockname, __DEP_MAP_MUTEX_INITIALIZER(lockname))
> >   
> >   #define DEFINE_MUTEX(mutexname) \
> >   	struct mutex mutexname = __MUTEX_INITIALIZER(mutexname)
> > diff --git a/include/linux/ww_mutex.h b/include/linux/ww_mutex.h
> > index 6ecf2a0220db..c62a030652b4 100644
> > --- a/include/linux/ww_mutex.h
> > +++ b/include/linux/ww_mutex.h
> > @@ -50,9 +50,17 @@ struct ww_acquire_ctx {
> >   
> >   #ifdef CONFIG_DEBUG_LOCK_ALLOC
> >   # define __WW_CLASS_MUTEX_INITIALIZER(lockname, class) \
> > -		, .ww_class = class
> > +		, .ww_class = &(class)
> > +
> > +# define __DEP_MAP_WW_MUTEX_INITIALIZER(lockname, class) \
> > +		, .dep_map = { \
> > +			.key = &(class).mutex_key, \
> > +			.name = (class).mutex_name, \
> 			,name = #class "_mutex", \
>
> and it 'works', but shees!

The name string itself may be duplicated for multiple instances of 
DEFINE_WW_MUTEX(). Do you want to keep DEFINE_WW_MUTEX() or just use 
ww_mutex_init() for all?

I notice that the problem with DEFINE_WW_MUTEX is that the ww_mutex 
themselves has null key instead of the same key from the ww_class as 
with ww_mutex_init().

Cheers,
Longman