Re: [PATCH] coccinelle: put_device: reduce false positives

[Dan Carpenter]

On Tue, Mar 26, 2019 at 10:38:43AM +0100, Julia Lawall wrote:
> 
> 
> On Tue, 26 Mar 2019, Dan Carpenter wrote:
> 
> > On Sat, Mar 23, 2019 at 09:06:54PM +0100, Julia Lawall wrote:
> > >
> > >
> > > On Sat, 23 Mar 2019, Markus Elfring wrote:
> > >
> > > > > Don't complain about a return when this function returns "&pdev->dev".
> > > >
> > > > Would this information qualify to add the tag âFixesâ to the commit message?
> > >
> > > Fixes tags relate to stable kernels, so that one can see which stable
> > > kernels a particular patch should be propagated to.  There is no need to
> > > propagate patches on semantic patches to stable kernels.  People who run
> > > stable kernels are interested in their behavior, not the bug finding
> > > rules that they contain.
> >
> > The Fixes tag is not just about stable...  For example, we use them for
> > statistics to see how quickly bugs get fixed etc.
> 
> OK.  But still do we need fixes tags for bug finding rules?  Perhaps if
> the previous version was really broken, and it would be really undesirable
> to use it.

It's not worth resending a patch for that, but I probably would use the
fixes tag.  It depends on your definition of "bug" really...  I tell
people not to use Fixes for spelling mistakes and unused variables.  But
I do use the Fixes tag for things like "an off by one in a sanity check
which doesn't affect run time because the index is always correct".

regards,
dan carpenter