Re: [RFC] PTR_ERR: return 0 if ptr isn't an error value.

[Julia Lawall <julia.lawall@lip6.fr>]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 1696 bytes --]

On Thu, 13 Jun 2013, Michael S. Tsirkin wrote:

> On Thu, Jun 13, 2013 at 02:07:40PM +0930, Rusty Russell wrote:
> > Julia Lawall <julia.lawall@lip6.fr> writes:
> > > On Mon, 3 Jun 2013, Uwe Kleine-König wrote:
> > > For a random example, here is a function that currently uses PTR_RET:
> >
> > Heheh, nice choice: I think I wrote that code originally :)
> >
> > > static int __net_init iptable_raw_net_init(struct net *net)
> > > {
> > >         struct ipt_replace *repl;
> > >
> > >         repl = ipt_alloc_initial_table(&packet_raw);
> > > 	if (repl == NULL)
> > >                 return -ENOMEM;
> > >         net->ipv4.iptable_raw =
> > >                 ipt_register_table(net, &packet_raw, repl);
> > > 	kfree(repl);
> > >         return PTR_RET(net->ipv4.iptable_raw);
> > > }
> > >
> > > If it becomes return PTR_ERR(...); at the end, won't it look like the
> > > function always fails?
> >
> > That is a valid point, though in this case the reader will know that
> > can't be the case.
> >
> > On the other hand, there's an incremental learning curve cost to every
> > convenience function we add.  There are only 50 places where we use
> > PTR_RET(), so it's not saving us very much typing over the clearest
> > solution: open-coding the test.
> >
> > I think using PTR_ERR() is a less bad solution than promoting PTR_RET,
> > which has a non-obvious name.
> >
> > Cheers,
> > Rusty.
>
> Will a longer name make the function more obvious?
> 	PTR_ERR_OR_ZERO() ?
> 	PTR_ERR0() ?
> PTR_ERR() can then stay simple for cases where we know we
> are on the error path.

I was thinking of something along those lines.  And in that case, PTR_ERR
could stay without the additional test.

julia