Re: [PATCH] i386 spinlocks should use the full 32 bits, not only 8 bits

[Andrew Morton <akpm@osdl.org>]

Ingo Molnar <mingo@elte.hu> wrote:
>
> +/*
>  + * We inline the unlock functions in the nondebug case:
>  + */
>  +#if defined(CONFIG_DEBUG_SPINLOCK) || defined(CONFIG_PREEMPT)
>  +# define spin_unlock(lock)		_spin_unlock(lock)
>  +# define read_unlock(lock)		_read_unlock(lock)
>  +# define write_unlock(lock)		_write_unlock(lock)
>  +#else
>  +# define spin_unlock(lock)		__raw_spin_unlock(&(lock)->raw_lock)
>  +# define read_unlock(lock)		__raw_read_unlock(&(lock)->raw_lock)
>  +# define write_unlock(lock)		__raw_write_unlock(&(lock)->raw_lock)
>  +#endif
>   

spin_lock is still uninlined.

static inline __attribute__((always_inline)) struct dentry *dget_parent(struct dentry *dentry)
{
 struct dentry *ret;

 _spin_lock(&dentry->d_lock);
 ret = dget(dentry->d_parent);
 __raw_spin_unlock(&(&dentry->d_lock)->raw_lock);
 return ret;
}

as is spin_lock_irqsave() and spin_lock_irq()

uninlining spin_lock will probably increase overall text size, but mainly in
the out-of-line section.

<looks>

we removed the out-of-line section :(


read_lock is out-of-line.   read_unlock is inlined

write_lock is out-of-line.  write_unlock is out-of-line.


Needs work ;)