On Tue, Aug 12, 2014 at 07:21:35PM +0200, Javier Martinez Canillas wrote: > On 08/12/2014 06:58 PM, Mark Brown wrote: > > On Tue, Aug 12, 2014 at 06:44:23PM +0200, Javier Martinez Canillas wrote: > >> The tps65090 is a Power Management Unit (PMU) used in several > >> boards so the same information is described on different DTS. > >> It is better to create a .dtsi fragment that can be included. > > Why is it better to do this? > Is better IMHO because we have a single place where the tps65090 information can > be updated instead of duplicating the same definition on each DTS. But there is no real information in this file. > This appears to be the current trend to better manage shared DTS snippet across > different boards. Others examples are arch/arm/boot/dts/omap-gpmc-smsc911x.dtsi > and arch/arm/boot/dts/twl6030.dtsi. In the smsc911x case that's a block from a reference design that's commonly repeated over multiple systems and is therefore similar to the reference design elements that have been factored out for Chromebooks. The twl6030 fragment is just broken - the regulator section is actively harmful and should be removed. > > >> + regulators { > >> + tps65090_dcdc1: dcdc1 { > >> + }; > >> + > > > > It appears to be largely content free, exactly the same effect should be > > achieved by removing the entire regulators node. > > > > Yes it's content free but later "[PATCH 6/6] ARM: dts: Add tps65090 FETs > constraints" [0] fills the FETs constraints [0]. This is a preparatory patch. > > Also, having all the regulators allows DTS files to reference the node by the > label if they want to add other properties. I can squash this patch and 06/06 if > you think that is better but I thought that the split makes it easier to review. > > Best regards, > Javier > > [0]: https://lkml.org/lkml/2014/8/12/377 >
Attachment:
signature.asc
Description: Digital signature