On Thu, 2020-06-04 at 20:35 +0300, Dan Carpenter wrote:
> On Thu, Jun 04, 2020 at 09:08:44AM -0700, Joe Perches wrote:
> > On Thu, 2020-06-04 at 15:30 +0300, Dan Carpenter wrote:
> > > On Thu, Jun 04, 2020 at 01:42:12PM +0200, Julia Lawall wrote:
> > > > OK, I recall a discussion with Dan where he suggested that some things
> > > > that were not actually bug fixes could also merit a Fixes tag. But it's
> > > > probably better if he weighs in directly.
> > >
> > > I generally think Fixes should only be used for "real bug" fixes.
> > >
> > > The one exception is when I'm reviewing a patch that fixes an "unused
> > > assignment" static checker warning is that I know which commit
> > > introduced the warning.
Sometimes those warnings are introduced by new compiler
versions.
That's why I don't care for -Werror use in Makefiles.
> > > I don't have strong feelings if it's in the
> > > Fixes tag or if it's just mentioned in the commit message.
> >
> > My view is that changes that silence compiler warnings are
> > not fixing bugs and that these changes should generally not
> > be backported.
> >
> The Fixes tag is useful for backports but that's not whole the point of
> it. It's also for collecting metrics.
Hmm, how are these metrics used?
> Also sometimes we fix the bug
> before the kernel is released so the Fixes tag means we can automatically
> ignore those ones when we look at which patches to backport.
>
> I don't care if the "unused assignment" patches use a Fixes tag or just
> mention the commit. Either way the information is there for when I
> review the patch.
Perhaps there could/should be some distinction between
"real bug" fixes and trivialities like "unused assignment"
Maybe something like:
Updates: <commit> ("commit description")
vs
Fixes: <commit> ("commit description")
