Re: [patch -next] ath9k: fix ath9k_hw_nvram_check_version()

[Dan Carpenter <dan.carpenter@oracle.com>]

On Sun, Dec 20, 2015 at 11:05:55AM -0800, Joe Perches wrote:
> On Sun, 2015-12-20 at 20:00 +0100, Julia Lawall wrote:
> > On Sun, 20 Dec 2015, Joe Perches wrote:
> > 
> > > On Sat, 2015-12-19 at 13:59 +0300, Dan Carpenter wrote:
> > > > There is a type bug so it always returns success.
> > > 
> > > How many false positives do you have to sift
> > > through to find this sort of error?
> > 
> > The return type is thoughtfully bool, so it should be easy in this case.  
> > The function has a return -EINVAL and a return true, so even without the 
> > return type it would be locally apparent that there is an inconsistency.
> 
> True, -EINVAL is a non-bool constant, but
> bool returns can be a variable int.
> 
> Dan, was the check any constant non-bool?

This warning is for returning negative with unsigned types smaller than
int.  It's doing cross function flow analysis so it could warning about
other things besides constants, but it's 90% constants.  It warns
about ds2482_w1_triplet() for example.

There aren't many false positives.

regards,
dan carpenter