Re: [RFC PATCH 2/6] zsmalloc: make zspage lock preemptible

[Sergey Senozhatsky]

On (25/01/27 21:23), Uros Bizjak wrote:
> > +static void zspage_read_lock(struct zspage *zspage)
> > +{
> > +	atomic_t *lock = &zspage->lock;
> > +	int old;
> > +
> > +	while (1) {
> > +		old = atomic_read(lock);
> > +		if (old == ZS_PAGE_WRLOCKED) {
> > +			cpu_relax();
> > +			continue;
> > +		}
> > +
> > +		if (atomic_cmpxchg(lock, old, old + 1) == old)
> > +			return;
> 
> You can use atomic_try_cmpxchg() here:
> 
> if (atomic_try_cmpxchg(lock, &old, old + 1))
>         return;
> 
> > +
> > +		cpu_relax();
> > +	}
> > +}
> > +
> > +static void zspage_read_unlock(struct zspage *zspage)
> > +{
> > +	atomic_dec(&zspage->lock);
> > +}
> > +
> > +static void zspage_write_lock(struct zspage *zspage)
> > +{
> > +	atomic_t *lock = &zspage->lock;
> > +	int old;
> > +
> > +	while (1) {
> > +		old = atomic_cmpxchg(lock, ZS_PAGE_UNLOCKED, ZS_PAGE_WRLOCKED);
> > +		if (old == ZS_PAGE_UNLOCKED)
> > +			return;
> 
> Also, the above code can be rewritten as:
> 
> while (1) {
>         old = ZS_PAGE_UNLOCKED;
>         if (atomic_try_cmpxchg (lock, &old, ZS_PAGE_WRLOCKED))
>                 return;	
> > +
> > +		cpu_relax();
> > +	}
> > +}
> 
> The above change will result in a slightly better generated asm.

Thanks, I'll take a look for the next version.