Skip to contentSkip to conversation

nvme-cli license clarification

DISCUSSION Daniel Wagner 2 participants 3 messages May 16, 2022 – May 17, 2022 NVMe

lore
Share & tools
lore
#0 lore raw

Hi,

The question pop up about the license of nvme-cli and libnvme. The reporter noticed that there are some inconsistency [1]. First let's just have a look at nvme-cli.

Keith stated that the original license is/was GPL-2.0-or-later.

Most of the files are either not having any copyright notes at the beginning of the file or are sporting the 'or any later version' snipped. So all good here, nothing to do.

But there are some noticeable exceptions though which all have GPL-2.0-only. I assume the general mood is that we want the project to be licensed as GPL-2.0-or-later, so let's try to resolve these issues:

1. fabrics.c

Ming added this file initially with

Copyright (C) 2016 Intel Corporation. All rights reserved. Copyright (c) 2016 HGST, a Western Digital Company. Copyright (c) 2016 Samsung Electronics Co., Ltd.

There are plenty of contributions for this file. E.g the counted lines contributed by an author for the monolithic branch (aka the 1.x version)[2]:

456 Sagi Grimberg
328 Ming Lin
253 Martin Wilck
229 Keith Busch
175 Hannes Reinecke
87 Martin Belanger
85 James Smart
50 Minwoo Im
47 Andy Lutomirski
45 Johannes Thumshirn
41 Christoph Hellwig
39 Martin George
26 Roland Dreier
23 Jay Freyensee
20 Eyal Ben David
11 Chaitanya Kulkarni
10 Duane Grigsby
8 Israel Rukshin
7 Maurizio Lombardi
6 Max Gurtovoy
4 Yair Elharrar
4 Gollu Appalanaidu
3 Yi Zhang
3 Eric Curtin
3 Daniel Wagner
2 Rakesh Pandit
2 Bart Van Assche
1 Tomasz Zawadzki
1 Nitzan Carmi
1 Logan Gunthorpe
1 Gwendal Grignou
1 Guan Junxiong
1 Breno Leitao

and for the master branch (aka 2.x version)

564 Hannes Reinecke
350 Daniel Wagner
127 Martin Belanger
80 Ming Lin
57 Sagi Grimberg
35 Tokunori Ikegami
24 Keith Busch
11 Martin George
8 Christoph Hellwig
7 Wu Bo
6 Roland Dreier
5 Jay Freyensee
3 Eyal Ben David
2 Rakesh Pandit
2 Martin Wilck
2 Gollu Appalanaidu
2 Chaitanya Kulkarni
1 Yi Zhang
1 Vincent Fu
1 Tomas Bzatek
1 Minwoo Im
1 Logan Gunthorpe
1 Breno Leitao
1 Bart Van Assche

Because Hannes and I have moved a lot of code around we listed first.

Not totally sure how to proceed here. I suppose we need for listed the copyright holders an official 'or any later version' change Signed-off-by? Things might be difficult as for example Ming is no longer with Samsung. Any ideas?

2. nvme.h

The situation here is a bit simpler as Keith is 'owning' most of the lines (monolithic and master branch are almost identically)

Copyright (c) 2011-2014, Intel Corporation

104 Keith Busch
26 Hannes Reinecke
6 Steven Seungcheol Lee
3 Sagi Grimberg
3 Christoph Hellwig
2 Scott Bauer
2 Minwoo Im
2 Martin Wilck
2 Daniel Wagner
1 Warner Losh
1 Tomasz Zawadzki
1 Ming Lin
1 Max Gurtovoy
1 Martin Belanger
1 Jeffrey Lien
1 James Smart
1 Brandon Paupore

I suppose someone from Intel needs to signed of the 'or any later version' thing. Keith do you happen to have a contact for this?

3. nvme-rpmb.c

This files seems to be owned mostly by Micron.

Copyright (C) 2020 Micron Techology Inc. All rights reserved.

933 Hanumanthu H
52 Daniel Wagner
35 Hannes Reinecke
22 Gollu Appalanaidu
3 naveen kumar
2 Keith Busch
1 Klaus Jensen
1 Ariadne Conill

Hanumanthu would you be fine to change the license to GPL-2.0-or-later?

4. util/parser.c

This file was provided by Ming and according the commit message it is imported from the Linux kernel 4.5. I don't think we can get that re-licensed under GPL-2.0-or-later. The good news is that it seems we can just remove the file as it is not used.

Thanks
Daniel

[1] https://github.com/linux-nvme/nvme-cli/issues/1508
[2] git blame --line-porcelain FILENAME | sed -n 's/^author //p' | sort -f | uniq -ic | sort -nr

maintainer
#1 lore raw
Quoted 25 lines · Daniel Wagner

2. nvme.h

The situation here is a bit simpler as Keith is 'owning' most of the lines (monolithic and master branch are almost identically)

Copyright (c) 2011-2014, Intel Corporation

104 Keith Busch
26 Hannes Reinecke
6 Steven Seungcheol Lee
3 Sagi Grimberg
3 Christoph Hellwig
2 Scott Bauer
2 Minwoo Im
2 Martin Wilck
2 Daniel Wagner
1 Warner Losh
1 Tomasz Zawadzki
1 Ming Lin
1 Max Gurtovoy
1 Martin Belanger
1 Jeffrey Lien
1 James Smart
1 Brandon Paupore

Quoted 2 lines

I suppose someone from Intel needs to signed of the 'or any later version' thing. Keith do you happen to have a contact for this?

I am not sure of who would be interested at Intel in this anymore. This was initially just a copy of the Linux kernel's header, though, and diverged significantly from there, but I guess we'd need to inherit the original's GPLv2 only license?

#2 lore raw
Quoted 8 lines · Keith Busch
Quoted 7 lines · Daniel Wagner

2. nvme.h

The situation here is a bit simpler as Keith is 'owning' most of the lines (monolithic and master branch are almost identically)

   Copyright (c) 2011-2014, Intel Corporation
Quoted 7 lines
Quoted 2 lines

I suppose someone from Intel needs to signed of the 'or any later version' thing. Keith do you happen to have a contact for this?

I am not sure of who would be interested at Intel in this anymore. This was initially just a copy of the Linux kernel's header, though, and diverged significantly from there, but I guess we'd need to inherit the original's GPLv2 only license?

Indeed, there isn't much left from the uapi header. I see to options.

The first is to accept this situation and mark nvme-cli as GPL-2.0-only or second we reimplement the necessary parts under GPL-2.0-or-later.

Most of the content is just function declarations and the implementation is licensed under GPL-2.0-or-later. Do we really have to consider these declarations as GPL-2.0-only? So this could be moved to a new header file which is licensed under GPL-2.0-or-later, I suppose.

The rest of the file could be reimplemented fairly simple. What you think?