Re: [patch 15/17] s390: convert /proc/cio_ignore.

[Cornelia Huck <cornelia.huck@gmail.com>]

2005/12/10, Andrew Morton <akpm@osdl.org>:
> Martin Schwidefsky <schwidefsky@de.ibm.com> wrote:
> >
> > +static void *
> >  +cio_ignore_proc_seq_start(struct seq_file *s, loff_t *offset)
> >  +{
> >  +    struct ccwdev_iter *iter;
> >  +
> >  +    if (*offset > __MAX_SUBCHANNEL)
> >  +            return NULL;
> >  +    iter = kmalloc(sizeof(struct ccwdev_iter), GFP_KERNEL);
> >  +    if (!iter)
> >  +            return ERR_PTR(-ENOMEM);
> >  +    memset(iter, 0, sizeof(struct ccwdev_iter));
>
> kzalloc()

OK, will do.

>
> >  +    iter->devno = *offset;
> >  +    return iter;
> >  +}
> >  +
> >  +static void
> >  +cio_ignore_proc_seq_stop(struct seq_file *s, void *it)
> >  +{
> >  +    if (!IS_ERR(it))
> >  +            kfree(it);
> >  +}
> >  +
> >  +static void *
> >  +cio_ignore_proc_seq_next(struct seq_file *s, void *it, loff_t *offset)
> >  +{
> >  +    struct ccwdev_iter *iter;
> >  +
> >  +    if (*offset > __MAX_SUBCHANNEL)
> >  +            return NULL;
> >  +    iter = (struct ccwdev_iter *)it;
>
> Unneeded (and undesirable) cast of void*.

Some people seem to prefer explicit casts to make the
type more clear. Is there any concensus on this? (I don't care
either way ;))

Cornelia