Re: [Qemu-devel] [PATCH v5] Xen PV Device

[Anthony Liguori <anthony@codemonkey.ws>]

Stefano Stabellini <stefano.stabellini@eu.citrix.com> writes:

> On Mon, 8 Jul 2013, Anthony Liguori wrote:
>> Andreas Färber <afaerber@suse.de> writes:
>> 
>> Right, it goes:
>> 
>> 1) Acked-by:
>> 
>> I haven't reviewed the code in detail but the general idea seems sane.
>> 
>> 2) Reviewed-by:
>> 
>> The general idea seems sane, and I have done a thorough review of the
>> patch in question.
>> 
>> 3) Signed-off-by:
>> 
>> All of the above, plus I have ensured that the code is of good quality,
>> does not break things, and the other things expected of a maintainer.
>> This is considered to be a legally binding statement too based on the
>> DCO so be aware of that and ensure you have the right approval to make
>> such a statement.
>
> I don't think that is a good idea to mix up DCO with reviewing
> patches.

It's all a question of patch origin and accounting.  DCO is just one
part of it.

> In fact in the Linux community I think that it's pretty clear that
> Signed-off-by doesn't mean anything other than "at least a portion of
> the changes have been done by me and I am the Copyright owner of
> them".

No, it also means: "I can certify that the person who provided the patch
to me has the appropriate rights to submit the patch."  See section (c)
of the DCO.

It's about establishing a chain of custody.  I'm not making any kind of
judgement when I merge a pull request from you because you've told me
(by adding your Signed-off-by) that all of the code is of appropriate
origin.

Of course, if you are not also saying that the code is of high quality
and does what it's described too, I don't really care about the code
origin in the first place :-)  So this is an important part of it too.

Anyone can add a Signed-off-by.  There's no requirement on authorship.
It's just not all that useful outside of a maintainership context.

If you cherry pick someone's patch from the mailing list and add it to
your series, you should add a Signed-off-by to it even though you aren't
necessarily the maintainer of the area.

> For example Alice writes a patch and goes away, Bob takes it, rewrites
> most of it and then sends it upstream. The patch has Alice and Bob
> Signed-off-by but Alice might not even read Bob's patch.

The ordering of Signed-off-by has significance.  In this case, Alice did
not Signed-off-by Bob's changes and that's expressed in the ordering.

Regards,

Anthony Liguori