On Wed, Jan 23, 2019 at 09:19:59PM +0300, Sergei Shtylyov wrote: > Hello! > > On 01/21/2019 03:01 PM, Magnus Damm wrote: > > >> On Mon, Jan 21, 2019 at 12:49 PM Magnus Damm <[email protected]> wrote: > >>> Remove undocumented IMR-LX4 device nodes > >>> > >>> [PATCH/RFC 01/02] arm64: dts: renesas: r8a7795: Remove IMR-LX4 device > >>> nodes > >>> [PATCH/RFC 02/02] arm64: dts: renesas: r8a7796: Remove IMR-LX4 device > >>> nodes > >>> > >>> These patches take the easy way out and simply remove the undocumented > >>> IMR-LX4 device nodes from the upstream tree. Good or bad, let me know! > >>> > >>> So perhaps this is a bit overly aggressive but since the DT bindings seem > >>> undocumented and no driver exists in upstream my gut feeling says these DT > >>> nodes were part of an upstreaming attempt that got suspended half-way > >>> through. > >>> > >>> In case DT binding documentation is in-flight and queued up somewhere > >>> (ideally together with a driver) then feel free to ignore this series. > >>> > >>> Instead of removing nodes we could also document the DT bindings for the > >>> IMR-LX4 devices. It would also make sense to add device nodes to other > >>> more recent SoCs than just H3 and M3-W. But blindly adding more DT nodes > >>> with a DT binding but without a driver seems a bit suboptimal compared to > >>> testing against an actual driver. > >> > >> [PATCH v5] media: platform: Renesas IMR driver > >> https://lore.kernel.org/linux-renesas-soc/[email protected]/ > > > > Thanks, but that seems to be from 2017! =) > > I dropped the ball there, as I was tasked with upstreaming V3x support... > The last thing done about the IMR driver was talking to Hans in Prague in > 2017. I'm planning to return to the driver after I'm done with the > HyperFlash driver.
Hi Sergei, I appreciate that we are not always in control of our own priorities, indeed I sympathise with that predicament. However, we shouldn't really be in a situation where DT is making use of undocumented bindings. I would like to ask for the bindings to be documented in the upstream kernel in the near future. And if that is not possible I believe we should consider temporarily removing their use in DT in the upstream kernel.
