[alsa-devel] Missing return check of of_property_read_*()

[Mark Brown]

On Wed, Sep 09, 2015 at 06:01:22PM +0200, Takashi Iwai wrote:
> Mark Brown wrote:

> > It'll be because nobody is checking !OF ASoC configurations and paying
> > attention to warnings - with !OF the functions are inline in the header
> > and the compiler can tell nothing happens.

> Yes, but non-OF ASoC gets more popular on x86 nowadays, so this should
> be covered regularly, too.

Hopefully Intel will be paying attention here!  :P

> > > I'm considering whether adding __must_check to of_property_read*()
> > > would make sense.  Then it'll be caught no matter which gcc version is
> > > used.  But this would hit too many spots.

> > Well, if they're all buggy...

> Not all of them are buggy, of course.  Some codes set the default
> fallback value beforehand.  But, if thinking of future use, it might
> be better to make the check mandatory.  No strong opinion here,
> though.

I don't know how clear it is that a failed read won't overwrite existing
data in the value variable.  The other option would be to have Kconfig
able to generate all*config with select options (like OF) disabled and
then get the build bots to start covering those as a standard
configuration.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 473 bytes
Desc: Digital signature
URL: <http://mailman.alsa-project.org/pipermail/alsa-devel/attachments/20150909/84ee94b9/attachment.sig>