Re: review of ocfs2

[Pekka Enberg <penberg@gmail.com>]

Hi,

On 5/30/05, Andi Kleen <ak@suse.de> wrote:
> +static int ocfs2_do_request_vote(ocfs_super *osb,
> +                                u64 blkno,
> +                                unsigned int generation,
> +                                enum ocfs2_vote_request type,
> +                                int orphaned_slot,
> +                                struct ocfs2_net_response_cb *callback)
> ...
> +       request = kmalloc(sizeof(*request), GFP_KERNEL);
> +       if (!request) {
> +               status = -ENOMEM;
> +               mlog_errno(status);
> +               goto bail;
> +       }
> +       memset(request, 0, sizeof(*request));
> 
> kcalloc

Actually, the latest preferred form for

       kcalloc(1, sizeof(*p), GFP_KERNEL)

is the following (as suggested by Al Viro):

	request = kmalloc(sizeof(*request), GFP_KERNEL);
	if (!request) {
		status = -ENOMEM;
		mlog_errno(status);
		goto bail;
	}
	*request = (ocfs2_vote_msg) { .md1.v_generic1 = htonl(priv) };
	hdr = &request->v_hdr;

	response_id = ocfs2_new_response_id(osb);

	*hdr = (ocfs2_msg_hdr) {
		.h_response_id = htonl(response_id);
		.h_request = htonl(type);
		.h_blkno = cpu_to_be64(blkno);
		.h_generation = htonl(generation);
		.h_node_num = htonl((unsigned int) osb->node_num)
	};

                   Pekka