Re: linux-next: Signed-off-by missing for commit in the rdma-fixes tree

[Jason Gunthorpe <jgg@mellanox.com>]

On Thu, Mar 08, 2018 at 09:23:12AM -0800, Joe Perches wrote:
> On Thu, 2018-03-08 at 10:03 -0700, Jason Gunthorpe wrote:
> > On Thu, Mar 08, 2018 at 08:39:38AM -0800, Joe Perches wrote:
> > > On Thu, 2018-03-08 at 09:18 -0700, Jason Gunthorpe wrote:
> > > > On Thu, Mar 08, 2018 at 04:40:58PM +1100, Stephen Rothwell wrote:
> > > > > Hi all,
> > > > > 
> > > > > Commit
> > > > > 
> > > > >   aa0de36a40f4 ("RDMA/mlx5: Fix integer overflow while resizing CQ")
> > > > > 
> > > > > is missing a Signed-off-by from its author.
> > > > 
> > > > Thanks Stephen,
> > > > 
> > > > May I suggest to the checkpatch maintainers that checkpack should look
> > > > for this too?
> > > 
> > > No such commit in current -hext
> > > $ git status HEAD
> > > HEAD detached at next-20180308
> > 
> > https://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma.git/commit/?h=for-rc&id=aa0de36a40f446f5a21a7c1e677b98206e242edb
> > 
> > > checkpatch doesn't try to match email sender to sign-off
> > > as the email sender doesn't have to sign-off on a patch.
> > > 
> > > It currently does:
> > > 
> > > 	if ($is_patch && $has_commit_log && $chk_signoff && $signoff == 0) {
> > > 		ERROR("MISSING_SIGN_OFF",
> > > 		      "Missing Signed-off-by: line(s)\n");
> > > 	}
> > 
> > If I were to suggest an exact check it would be that the commit git
> > author and commit git commiter both have signed-off-by lines,
> 
> The git commit committer can't be known by checkpatch.
> 
> It just looks at patches.

Well, it would have to extract the git From line and look up the
commit ID in the local git to find that information. Obviously not a
work flow for everyone..

> The original author of any patch can't be known either.
> 
> There is a mechanism to add a "From:" line that is
> supposed to represent the sender of the patch, but that
> person may not be the original author.

Well, I would be happy if checkpatch computed the author the same way
git-am did and used that to check the signed-off-by ..

If I recall, git am looks for a 'From ' header at the top of the body
to override the email header?

I only ask becuase I keep seeing these reports from Stephen and
usuaully by the time it gets to him it requires a disruptive rebase to
fix..

Jason