Do we have people interested in device tree janitoring / cleanup?

Dave Martin Dave.Martin at
Thu Jul 25 07:06:34 EDT 2013

On Wed, Jul 24, 2013 at 08:27:13AM -0700, Olof Johansson wrote:
> Every now and then I come across a binding that's just done Wrong(tm),
> merged through a submaintainer tree and hasn't seen proper review --
> if it had, it wouldn't look the way it does. It's something we're
> starting to address now since there's more people stepping up to be
> maintainers, but there's a backlog of bad bindings already merged.
> Often they are produced by translating the platform_data structures
> directly over into device-tree properties without consideration to
> describing the hardware or usual conventions, using key/value pairs
> instead of boolean properties, etc.
> Getting involved in cleaning up these kind of bindings is a great way
> to learn "the ways of device tree" for someone that has interest in
> that.
> Latest find in this area is the Maxim 8925 bindings, that I came
> across since they caused a compile warning on some defconfig. I'll
> post a patch to address the warning but if someone else feels like
> fixing the bindings on top of it that would be appreciated!

DT bindings (even poorly conceived, ad-hoc and/or undocumented ones)
are ABI.

A review/merge process which _allows_ junk into an ABI is the real
problem we need to solve here, but once it's there we can't just magic
it away.

Do we plan on having a proper deprecation path for the junk, so that
the old, superseded bindings continue to work for a limited time,
preferably with a big fat warning somewhere?


More information about the linux-arm-kernel mailing list