[dpdk-dev] [dpdk-techboard] DPDK ABI/API Stability

[Bruce Richardson]

On Thu, Apr 04, 2019 at 10:29:19AM +0100, Burakov, Anatoly wrote:
> On 03-Apr-19 4:42 PM, Ray Kinsella wrote:
> > Hi folks,
> > 
> > Recently I started a discussion with the DPDK Technical Board on DPDK
> > ABI/API stability. This was born out informal feedback I had received
> > from a number of users of DPDK about ABI churn. In turn this feedback
> > then prompted an ABI analysis of DPDK using tools from abi-laboratory.
> > 
> > https://abi-laboratory.pro/index.php?view=timeline&l=dpdk
> > 
> > I guess the short story is that DPDK ABI hasn't really settled down as
> > the project has matured. If you take a look at the “Backward Compat.”
> > column which measures ABI compatibility compared to the previous
> > releases, you will see significant churn in the ABI over successive
> > releases since v16.04.
> > 
> > Now compare DPDK to GStreamer as an example of a very mature project
> > with a similar intent, a framework for building applications, and which
> > enjoys a very stable API.
> > 
> > https://abi-laboratory.pro/index.php?view=timeline&l=gstreamer
> > 
> > The DPDK ABI churn has the following affects for users:-
> > 
> > 1. The churn obliges users of DPDK to commit to a constant
> > re-integration and re-validation effort for new versions of DPDK. This
> > effort from their perspective may not add value to their consuming
> > project, particular if they are only updating to "stay current".
> > 2. The churn encourages users of DPDK to slip versions, putting off
> > reintegration to later, building up technical debt and causing their
> > projects to miss support for new hardware or features.
> > 3. It makes DPDK different to almost every other system library and
> > framework that an operating systems might ship. This makes DPDK trickier
> > to dynamically link against, package and maintain for OS maintainers.
> > 
> > In order to address this issue, I have put together the minimal set of
> > concrete proposals below for discussion at the Technical Board next
> > Wednesday.
> > 
> > I wanted to share this, as these might not yet be the right proposals,
> > however I am putting them out there for feedback to start the discussion.
> > 
> > Thanks,
> > 
> > Ray K
> > 
> > 
> > Experimental API
> > 1.	APIs designated as experimental are not considered part of the ABI
> > and may change without warning at any time.
> > 2.	APIs designated as experimental must be marked depreciated for a
> > least one quarterly release before removal.
> > 3.	APIs designated as experimental will no longer automatically graduate
> > to core after one release, they may stay experimental until their author
> > and the maintainer agree that graduation is appropriate.
> > 
> > Core API (non-experimental API)
> > 4.	APIs designated as core must be depreciated for a least two years
> > before removal, to facilitate the continued compatibility with LTS
> > releases. A final removal notice will be published to the DPDK Mailing
> > List, and if there are no strong objections only then an API may be
> > removed.
> > 5.	APIs designated as core may be changed as follows:-
> > 5.a	The change proposer must demonstrated that the change has a
> > supporting use case and could not be achieved in any other way.
> > 5.b	ABI version compatibility must be retained, as described below.
> 
> Hi Ray,
> 
> My somewhat rambly 2 cents :)
> 
> While i think some solution has to be found for the situation, we also have
> to balance this against speed of development and new features rollout.
> 
> For example, let's consider what i am intimately familiar with - the memory
> rework. I have made enormous efforts to ensure that pre-18.05 and post-18.05
> remain as ABI/API compatible as possible, but there were a couple of API
> calls that were removed, and there couldn't have been any replacements
> (these API's were exposing internal structures that shouldn't have been
> exposed in the first place), and 18.05 also broke the ABI compatibility,
> because there was no way to do it without it (shared internal structures
> needed to change in part to support multiprocess).
> 
> So, if i understand your proposal correctly, assuming a 2-year waiting
> period for the deprecation of core API's, you would essentially still be
> waiting for the memory rework to land for a year more. Moreover, even
> *after* it has landed, there was a continuous stream of improvements and
> bugfixes, some of which has broke ABI compatibility as well. Some of them
> were my fault (as in, i could've foreseen the need for those changes, but
> didn't), but others came as a result of people using these new features in
> the wild and reporting issues/problems/suggestions - i am but one man, after
> all. Plus, you know, there's only 24 hours in a day, and some stuff takes
> time to implement :)
> 
> Since this rework goes right at the heart of DPDK (arguably there isn't a
> more "core" API than memory!), there is no (sane) way in the universe to 1)
> keep backwards compatibility for this, or 2) keep two parallel versions of
> it. We also need to test all that, and, to be honest, one validation cycle
> for a release wouldn't be enough to figure out all of the kinks and
> implications of such a case. It was really great that memory rework has
> landed in 18.05 and we had time to improve and prepare it for an 18.11 LTS -
> i think everyone can say that it's in much better shape in 18.11 than it was
> in 18.05, but if we couldn't do an ABI break here or there, this rate of
> improvements would have slowed down significantly.
> 
> Now, i understand that this is probably a highly exceptional case, but i'm
> sure that maintainers of other parts of DPDK will have their own examples of
> similar things happening.
> 
> I have no idea what a proper solution would look like. Any "splitting" of
> the trees into "experimental" vs. "stable" will end up causing the same
> issue - people choose to use stable over experimental because, well, it's
> more stable, and new/experimental features don't get tested as much because
> no one runs the thing in the first place.
> 
> TL;DR we have to be careful not to constrain the pace of
> development/bugfixing just for the sake of having a stable API/ABI :)
> 

Actually, I think we *do* need to constrain the pace of development for the
sake of ABI stability. At this stage DPDK has been around for quite a
number of years and so should be considered a fairly mature project - it
should just start acting like it.

Now, in terms of features like the memory rework, that is indeed a case
that there was no alternative other than a massive ABI break. However, for
that rework there was a strong need for improvement in that area that we
can make the case for an ABI break to support it - and it is of a scale
that nothing other than an ABI change would do. For other areas and
examples, I doubt there are many in the last couple of years that are of
that scale.

My thoughts on the matter are:
1. I think we really need to do work to start hiding more of our data
structures - like what Stephen's latest RFC does. This hiding should reduce
the scope for ABI breaks.
2. Once done, I think we should commit to having an ABI break only in the
rarest of circumstances, and only with very large justification. I want us
to get to the point where DPDK releases can immediately be picked up by all
linux distros and rolled out because they are ABI compatible.

I'm not sure I like the idea of planned ABI break releases - that strikes
me as a plan where we end up with the same number of ABI breaks as before,
just balled into one release.

Question for Kevin, Luca and others who look at distro-packaging: is it the
case that each distro will only ship one version of DPDK, or is it possible
that if we have ABI breaks, a distro will provide two copies of DPDK
simultaneously, e.g. a 19.11 ABI version and a 20.11 ABI version?


So, in short, I'm generally in favour of a zero-tolerance approach for DPDK
ABI breaks, and having ABI breaks as a major event reserved only for
massive rework changes, such as major mbuf changes, or new memory layout or
similar.

Regards,
/Bruce