[linux-cifs-client] Re: [PATCH 08/13] cifs: convert cifs_get_inode_info to use cifs_iget

[Christoph Hellwig]

Looks good.


Reviewed-by: Christoph Hellwig <hch at lst.de>

Two little nitpicks:

On Wed, May 27, 2009 at 08:30:29AM -0400, Jeff Layton wrote:
>  				access_flags_to_mode(ppace[i]->access_req,
>  						     ppace[i]->type,
> -						     &(inode->i_mode),
> +						     &(fattr->cf_mode),
>  						     &user_mask);
>  			if (compare_sids(&(ppace[i]->sid), pgrpsid))
>  				access_flags_to_mode(ppace[i]->access_req,
>  						     ppace[i]->type,
> -						     &(inode->i_mode),
> +						     &(fattr->cf_mode),
>  						     &group_mask);
>  			if (compare_sids(&(ppace[i]->sid), &sid_everyone))
>  				access_flags_to_mode(ppace[i]->access_req,
>  						     ppace[i]->type,
> -						     &(inode->i_mode),
> +						     &(fattr->cf_mode),
>  						     &other_mask);

If you touch these lines please also remove the superflous braces.

> +	cifs_i->delete_pending = fattr->cf_flags & CIFS_FATTR_DELETE_PENDING;
> +
> +	/*
> +	 * Can't safely change the file size here if the client is writing to
> +	 * it due to potential races.
> +	 */
>  	spin_lock(&inode->i_lock);
>  	if (is_size_safe_to_change(cifs_i, fattr->cf_eof)) {
> -		/*
> -		 * We can not safely change the file size here if the client
> -		 * is writing to it due to potential races.
> -		 */

Why isn't this comment introduced in the correct location in the
patch adding it?