breaking DT compatibility (was: Re: [PATCH v4] clk: sunxi: Refactor A31 PLL6 so that it can be reused)

[Mark Rutland]

On Thu, Feb 11, 2016 at 11:00:48AM +0100, Maxime Ripard wrote:
> On Wed, Feb 10, 2016 at 04:30:01PM +0000, Mark Rutland wrote:
> > On Wed, Feb 10, 2016 at 03:37:55PM +0100, Maxime Ripard wrote:
> > > Hi Rob,
> > > 
> > > On Wed, Feb 10, 2016 at 07:42:02AM -0600, Rob Herring wrote:
> > > > On Wed, Feb 10, 2016 at 6:30 AM, Andre Przywara <andre.przywara at arm.com> wrote:
> > > > > Hi,
> > > > >
> > > > > just a ping:
> > > > >
> > > > > Are we really OK with breaking existing DTs in 4.6? (per the code in
> > > > > -next: f7d372ba54ea04d528a291b8dbe34716507bb60b, which is this patch).
> > > > 
> > > > I only warn and make sure people are aware of the issue. I leave that
> > > > up to platform maintainers to decide. It depends on the maturity of
> > > > the platform and users.
> > > 
> > > The impacted SoCs support is really partial. For the most supported
> > > one, big things like display or sound are totally missing, and we
> > > still update them on a regular basis to add support for new
> > > devices. As such, users are very likely to upgrade the DT from one
> > > version to another just because there's new devices available to
> > > them. And the newest SoC impacted just got introduced in 4.5, and only
> > > has the UART and MMC devices available.
> > > 
> > > > If people complain about it then it's their mess. For platforms
> > > > supported in distros such as debian or fedora, I would strongly
> > > > recommend against breaking compatibility.
> > > 
> > > None of them are officially supported:
> > > https://www.debian.org/releases/stable/armhf/ch02s01.html.en
> > > https://fedoraproject.org/wiki/Architectures/ARM#Fedora_23
> > > 
> > > Only the older one are, and they are not affected by this patch.
> > > 
> > > > They do ship dtbs, but it's a chicken and egg problem. If dtbs were
> > > > stable and provided by firmware, then they wouldn't have to provide
> > > > them. If dtbs are unstable, then they have no choice.
> > > 
> > > And while it might work great on platforms that have all the needed
> > > documentation, or a perfect one, which is our case. Almost each
> > > release, we discover that something is not working as it was
> > > documented, when it was documented in the first place.
> > > 
> > > It also seems that even on well documented platforms, supported by the
> > > vendors, the stable ABI dream is not going to happen:
> > > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/Documentation/arm/Atmel/README#n105
> > > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/Documentation/devicetree/bindings/arm/marvell,berlin.txt#n4
> > 
> > To be quite frank, I completely disagree with that stance.
> > 
> > It seems like the only reason DT bindings aren't remaining stable is
> > because people are deliberately ignoring the requirement and reasoning
> > for doing so.
> 
> And for DT maintainers saying on multiple occasions that it's bad but
> ok to break it and / or that they never actually said that it was a
> stable ABI...

Evidently there is a communication failure. Generally, the statement has
been that old DTBs should continue to work. That's even documented, as
Andre pointed out:

Documentation/devicetree/bindings/ABI.txt

There are obviously shades of grey, and _rarely_ it might be necessary
to deliberately break a binding. However, that should be the rare
last-resort case, rather than a crutch for development.

Saying "bad but ok" underplays the "bad" and overplays the "ok".

> I'm guessing it could be a stable ABI if there was bindings
> reviews. Rob actually started to review a significant amount of
> bindings lately, and that's really appreciated, but if you don't
> review all the bindings, then we're going to make mistakes.

I agree that it is unfortunate that we cannot provide the level of
review that we want to.

Mistakes will happen even with review; that doesn't necessitate removing
support for a binding.

> > I agree that it can be painful, and that we cannot predict the future.
> > There will always be bugs.
> 
> In our case, we can't even predict the present.
> 
> > Having code in mainline comes with responsibilities. One of those is to
> > keep said code working for existing users. Otherwise, why bother having
> > it in mainline at all?
> 
> None of our existing users ever complained.

I believe that in this case, Andre was complaining about this particular
breakage, unless I have misunderstood.

To be clear, I'm arguing for the strategy going forward. If no-one has
complained about the stuff broken up to this point, let's not waste time
restoring that.

Going forward we need to keep old DTBs supported.

Thanks,
Mark.