Re: [PATCH] x86: fix spin_is_contended()

[Nick Piggin <npiggin@suse.de>]

On Mon, Aug 04, 2008 at 02:38:54PM +0100, Jan Beulich wrote:
> The masked difference is what needs to be compared against 1, rather
> than the difference of masked values (which can be negative).

Dang, thanks.

> 
> Signed-off-by: Jan Beulich <jbeulich@novell.com>
> Cc: Nick Piggin <npiggin@suse.de>
> 
> ---
>  include/asm-x86/spinlock.h |    4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> --- linux-2.6.27-rc1/include/asm-x86/spinlock.h	2008-07-31 17:08:34.000000000 +0200
> +++ 2.6.27-rc1-x86-spin-is-contended/include/asm-x86/spinlock.h	2008-08-04 11:09:33.000000000 +0200
> @@ -65,7 +65,7 @@ static inline int __ticket_spin_is_conte
>  {
>  	int tmp = ACCESS_ONCE(lock->slock);
>  
> -	return (((tmp >> 8) & 0xff) - (tmp & 0xff)) > 1;
> +	return (((tmp >> 8) - tmp) & 0xff) > 1;
>  }
>  
>  static __always_inline void __ticket_spin_lock(raw_spinlock_t *lock)
> @@ -127,7 +127,7 @@ static inline int __ticket_spin_is_conte
>  {
>  	int tmp = ACCESS_ONCE(lock->slock);
>  
> -	return (((tmp >> 16) & 0xffff) - (tmp & 0xffff)) > 1;
> +	return (((tmp >> 16) - tmp) & 0xffff) > 1;
>  }
>  
>  static __always_inline void __ticket_spin_lock(raw_spinlock_t *lock)
> 
>