[Cluster-devel] [GFS2 Patch] GFS2: Add readahead to sequential directory traversal

[Steven Whitehouse <swhiteho@redhat.com>]

Hi,

On Mon, 2011-10-24 at 15:56 -0400, Bob Peterson wrote:
> ----- Original Message -----
> | Hi,
> | 
> | That looks much better, but why not create & pass a file_ra_state in
> | the
> | NFS getname function so that you don't have to check for it being
> | NULL?
> | Since that function always reads everything up until the name it is
> | looking for, it is an ideal place to do readahead as the access will
> | always be sequential.
> | 
> | Otherwise, it looks good to me,
> | 
> | Steve.
> 
> Hi,
> 
> There isn't a file struct passed in to the nfs get_name function, but
> I can do something like the following.  Is this what you mean?
> 
> diff --git a/fs/gfs2/export.c b/fs/gfs2/export.c
> index a581cd2..70ba891 100644
> --- a/fs/gfs2/export.c
> +++ b/fs/gfs2/export.c
> @@ -99,6 +99,7 @@ static int gfs2_get_name(struct dentry *parent, char *name,
>  	struct gfs2_holder gh;
>  	u64 offset = 0;
>  	int error;
> +	struct file_ra_state f_ra = { .start = 0 };
>  
>  	if (!dir)
>  		return -EINVAL;
> @@ -118,7 +119,7 @@ static int gfs2_get_name(struct dentry *parent, char *name,
>  	if (error)
>  		return error;
>  
> -	error = gfs2_dir_read(dir, &offset, &gnfd, get_name_filldir, NULL);
> +	error = gfs2_dir_read(dir, &offset, &gnfd, get_name_filldir, &f_ra);
>  
>  	gfs2_glock_dq_uninit(&gh);
>  
> Regards,
> 
> Bob Peterson
> Red Hat File Systems

Yes, thats the kind of things. Looks like we can use
file_ra_state_init() since it is exported too,

Steve.