lkml.org 
[lkml]   [2008]   [Feb]   [19]   [last100]   RSS Feed
Views: [more markup]   [less markup]   [headers]   [forward]  
 
Messages in this thread
/
DateTue, 19 Feb 2008 18:45:42 +0100
FromMarcin Slusarz <>
SubjectRe: [PATCH] ufs: [bl]e*_add_cpu conversion
On Tue, Feb 19, 2008 at 12:22:19AM +0100, Roel Kluin wrote:
> Andrew Morton wrote:
> > On Wed, 13 Feb 2008 10:41:44 +0100 Roel Kluin <12o3l@tiscali.nl> wrote:
> > 
> >> you may also want these:
> >> ---
> >> [bl]e_add_cpu conversion in return
> 
> > upsets powerpc (at least):
> > 
> > fs/ufs/swab.h: In function `fs64_add':
> > fs/ufs/swab.h:47: warning: passing arg 1 of `le64_add_cpu' from incompatible pointer type
> > fs/ufs/swab.h:49: warning: passing arg 1 of `be64_add_cpu' from incompatible pointer type
> > fs/ufs/swab.h: In function `fs64_sub':
> > fs/ufs/swab.h:58: warning: passing arg 1 of `le64_add_cpu' from incompatible pointer type
> > fs/ufs/swab.h:60: warning: passing arg 1 of `be64_add_cpu' from incompatible pointer type
> 
> sorry for this. Is it correct to cast like the patch below does?
I don't think so. Their prototypes are wrong. We can:
a) remove fs64_add and fs64_sub as nobody use them
b) fix them - change second parameter do __fs64 (and convert to [bl]e64_add_cpu)

Evgeniy?

> If not (anyone) feel free to correct and send a patch yourself.
> The patch below was *not* tested
> ---
> [bl]e_add_cpu conversion in return (with cast)
> 
> Signed-off-by: Roel Kluin <12o3l@tiscali.nl>
> ---
> diff --git a/fs/ufs/swab.h b/fs/ufs/swab.h
> index 1683d2b..594d6e8 100644
> --- a/fs/ufs/swab.h
> +++ b/fs/ufs/swab.h
> @@ -44,18 +44,22 @@ static __inline u32
>  fs64_add(struct super_block *sbp, u32 *n, int d)
>  {
>  	if (UFS_SB(sbp)->s_bytesex == BYTESEX_LE)
> -		return *n = cpu_to_le64(le64_to_cpu(*n)+d);
> +		le64_add_cpu((__le64 *)n, d);
>  	else
> -		return *n = cpu_to_be64(be64_to_cpu(*n)+d);
> +		be64_add_cpu((__le64 *)n, d);
> +
> +	return *n;
>  }
> 
>  static __inline u32
>  fs64_sub(struct super_block *sbp, u32 *n, int d)
>  {
>  	if (UFS_SB(sbp)->s_bytesex == BYTESEX_LE)
> -		return *n = cpu_to_le64(le64_to_cpu(*n)-d);
> +		le64_add_cpu((__le64 *)n, -d);
>  	else
> -		return *n = cpu_to_be64(be64_to_cpu(*n)-d);
> +		be64_add_cpu((__le64 *)n, -d);
> +
> +	return *n;
>  }
> 
>  static __inline u32


\
 
 \ /
  Last update: 2008-02-19 18:49    [from the cache]
©2003-2008