Re: [PATCH 02/47] block-rbd: Less function calls in rbd_header_from_disk() after error detection

From: Ilya Dryomov
Date: Tue Sep 13 2016 - 03:58:48 EST


On Mon, Sep 12, 2016 at 8:43 PM, SF Markus Elfring
<elfring@xxxxxxxxxxxxxxxxxxxxx> wrote:
> From: Markus Elfring <elfring@xxxxxxxxxxxxxxxxxxxxx>
> Date: Sun, 11 Sep 2016 13:18:57 +0200
>
> The functions "ceph_put_snap_context" and "kfree" were called in a few
> cases by the function "rbd_header_from_disk" during error handling
> even if the passed variables contained a null pointer.
>
> Adjust jump targets according to the Linux coding style convention.
>
> Signed-off-by: Markus Elfring <elfring@xxxxxxxxxxxxxxxxxxxxx>
> ---
> drivers/block/rbd.c | 10 ++++++----
> 1 file changed, 6 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c
> index e406c27..f4212e1 100644
> --- a/drivers/block/rbd.c
> +++ b/drivers/block/rbd.c
> @@ -1001,7 +1001,7 @@ static int rbd_header_from_disk(struct rbd_device *rbd_dev,
> snap_count = le32_to_cpu(ondisk->snap_count);
> snapc = ceph_create_snap_context(snap_count, GFP_KERNEL);
> if (!snapc)
> - goto out_err;
> + goto free_prefix;
> snapc->seq = le64_to_cpu(ondisk->snap_seq);
> if (snap_count) {
> struct rbd_image_snap_ondisk *snaps;
> @@ -1013,14 +1013,14 @@ static int rbd_header_from_disk(struct rbd_device *rbd_dev,
> goto out_2big;
> snap_names = kmalloc(snap_names_len, GFP_KERNEL);
> if (!snap_names)
> - goto out_err;
> + goto put_snap_context;
>
> /* ...as well as the array of their sizes. */
> snap_sizes = kmalloc_array(snap_count,
> sizeof(*header->snap_sizes),
> GFP_KERNEL);
> if (!snap_sizes)
> - goto out_err;
> + goto free_names;
>
> /*
> * Copy the names, and fill in each snapshot's id
> @@ -1066,10 +1066,12 @@ static int rbd_header_from_disk(struct rbd_device *rbd_dev,
> return 0;
> out_2big:
> ret = -EIO;
> -out_err:
> kfree(snap_sizes);
> + free_names:
> kfree(snap_names);
> + put_snap_context:
> ceph_put_snap_context(snapc);
> + free_prefix:
> kfree(object_prefix);
>
> return ret;
> --
> 2.10.0
>

Please don't send patches that restructure error handling gotos unless
you've spotted a bug. These patches are easy to get wrong, especially
when done in bulk, and require careful review.

Some people fancy a single sink label approach, others like separate
labels for each cleanup action - as long as the code is working, it's
a matter of taste.

Thanks,

Ilya