linux-kernel - Re: [PATCH v3 6/6] leds: Add support for RTL8231 LED scan matrix

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

Date:   Mon, 24 May 2021 17:30:30 +0200
From:   Sander Vanheule <sander@...nheule.net>
To:     Andy Shevchenko <andy.shevchenko@...il.com>
Cc:     Pavel Machek <pavel@....cz>, Rob Herring <robh+dt@...nel.org>,
        Lee Jones <lee.jones@...aro.org>,
        Mark Brown <broonie@...nel.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        "Rafael J . Wysocki" <rafael@...nel.org>,
        Michael Walle <michael@...le.cc>,
        Linus Walleij <linus.walleij@...aro.org>,
        Bartosz Golaszewski <bgolaszewski@...libre.com>,
        Linux LED Subsystem <linux-leds@...r.kernel.org>,
        devicetree <devicetree@...r.kernel.org>,
        "open list:GPIO SUBSYSTEM" <linux-gpio@...r.kernel.org>,
        Andrew Lunn <andrew@...n.ch>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v3 6/6] leds: Add support for RTL8231 LED scan matrix

On Mon, 2021-05-24 at 15:47 +0300, Andy Shevchenko wrote:
> On Mon, May 24, 2021 at 3:04 PM Sander Vanheule <sander@...nheule.net> wrote:
> > On Mon, 2021-05-24 at 13:24 +0300, Andy Shevchenko wrote:
> > > On Mon, May 24, 2021 at 1:34 AM Sander Vanheule <sander@...nheule.net>
> > > wrote:
> 
> ...
> 
> > > > +       if (ret != 2)
> > > > +               return -ENODEV;
> > > 
> > > I would say -EINVAL, but -ENODEV is similarly okay.
> > 
> > Any specific reason you think EINVAL is more appropriate than ENODEV?
> 
> My logic is that the initial values (from resource provider) are incorrect.
> But as I said, I'm fine with either.

Ok, that makes sense. Actually, I'm already using "address invalid" in the error
messages when reading the address fails, so I'll change to EINVAL for
consistency.


> 
> > > > +       int err;
> > > 
> > > ret or err? Be consistent across a single driver.
> > 
> > I had first used 'err' for both fwnode_property_count_u32() and
> > fwnode_property_read_u32_array(). The former returns "actual count or error
> > code", while the latter is only "error code". And I found it weird to read
> > the
> > code as "does error code equal 2", if I used 'err' as variable name.
> > 
> > I've split this up:
> >  * addr_count for fwnode_property_count_u32's result
> >  * err for fwnode_property_read_u32_array's result
> > 
> > Since addr_count is only used before err is touched, I guess the compiler
> > will
> > optimize this out anyway?
> 
> Usually we do this pattern (and it seems you missed the point, name of
> variable is ret in some functions and err in the rest):
> 
> err /* ret */ = foo();
> if (err < 0)
>   return err;
> count = err;

I had only used 'ret' specifically in this one function, because I didn't like 
"if (err != 2)" (and I apparently decided that I disliked that more than the
inconsistency introduced by using 'ret'). I'll stick to calling the variable
'err', and change the clause to (err != ARRAY_SIZE(addr)) to make it more
obvious that 2 isn't just some random return value.


Best,
Sander