linux-kernel - Re: [PATCH v2 2/2] gpiolib: check the 'ngpios' property in core gpiolib code

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <YZt9DVihGaq0kjKG@smile.fi.intel.com>
Date:   Mon, 22 Nov 2021 13:20:45 +0200
From:   Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
To:     Bartosz Golaszewski <brgl@...ev.pl>
Cc:     Linus Walleij <linus.walleij@...aro.org>,
        "open list:GPIO SUBSYSTEM" <linux-gpio@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2 2/2] gpiolib: check the 'ngpios' property in core
 gpiolib code

On Fri, Nov 19, 2021 at 08:35:33PM +0100, Bartosz Golaszewski wrote:
> On Thu, Nov 18, 2021 at 10:16 PM Andy Shevchenko
> <andriy.shevchenko@...ux.intel.com> wrote:
> > On Thu, Nov 18, 2021 at 09:12:59PM +0100, Bartosz Golaszewski wrote:
> > > On Thu, Nov 18, 2021 at 6:06 PM Andy Shevchenko
> > > <andriy.shevchenko@...ux.intel.com> wrote:
> > > > On Thu, Nov 18, 2021 at 02:23:17PM +0100, Bartosz Golaszewski wrote:

...

> > > > >       if (gc->ngpio == 0) {
> > > > > -             chip_err(gc, "tried to insert a GPIO chip with zero lines\n");
> > > > > -             ret = -EINVAL;
> > > > > -             goto err_free_descs;
> > > > > +             ret = device_property_read_u32(&gdev->dev, "ngpios", &ngpios);
> > > > > +             if (ret) {
> > > > > +                     chip_err(gc, "tried to insert a GPIO chip with zero lines\n");
> > > > > +                     ret = -EINVAL;
> > > > > +                     goto err_free_descs;
> > > > > +             }
> > > > > +
> > > > > +             gc->ngpio = ngpios;
> > > > >       }
> > > >
> > > > This should be
> > > >
> > > >         if (gc->ngpio == 0) {
> > > >                 ret = device_property_read_u32(&gdev->dev, "ngpios", &ngpios);
> > > >                 if (ret)
> > > >                         return ret;
> > >
> > > But device_property_read_u32() returning -ENODATA means there's no
> > > such property, which should actually be converted to -EINVAL as the
> > > caller wanting to create the chip provided invalid configuration - in
> > > this case: a chip with 0 lines. In case of the non-array variant of
> > > read_u32 that's also the only error that can be returned so this bit
> > > looks right to me.
> >
> > So, what is so special about -EINVAL? Why -ENODATA is not good enough which
> > will exactly explain to the caller what's going on, no?
> >
> 
> Let's imagine the user sets gc->ngpio = 0 incorrectly thinking it'll
> make gpiolib set it to some sane default. Then gpiochip_add_data()
> returns -ENODATA (No data available). This is confusing IMO. But if we
> convert it to -EINVAL, it now says "Invalid value" which points to the
> wrong configuration.
> 
> ENODATA means "device tree property is not present" in this case but
> the problem is that user supplies the gpiolib with invalid
> configuration. EINVAL is the right error here.

Then be explicit, don't shadow other error codes from fwnode API.

	if (ret && ret != -ENODATA)

> > > >                 gc->ngpio = ngpios;
> > > >         }
> > > >
> > > >         if (gc->ngpio == 0) {
> > > >                 chip_err(gc, "tried to insert a GPIO chip with zero lines\n");
> > > >                 ret = -EINVAL;
> > > >                 goto err_free_descs;
> >
> > When the caller intended to create a chip with 0 GPIOs they will get an error
> > as you wish with an error message.
> 
> Yes, as it was before.

-- 
With Best Regards,
Andy Shevchenko