On Mon, Apr 03, 2017 at 12:32:08AM +0800, Icenowy Zheng wrote: > > > On Sun, 2 Apr 2017, Maxime Ripard wrote: > > > Hi, > > > > On Wed, Mar 29, 2017 at 06:42:41PM +0800, Icenowy Zheng wrote: > > > Allwinner SoCs after sun6i-a31 nearly all have a R_CCU in PRCM part. > > > (V3s and R40 do not have it, as they have even no PRCM) > > > > > > This patch adds support for the ones on H3/A64. > > > > > > Some clock/reset values are reserved for easier extending the support to > > > A31/A23, but for this I think some changes to the PRCM MFD should be made, > > > see [1] (Although this is only a sketch). > > > > > > The r_pio device node is also added for A64, as the driver is already > > > merged, and its depends (r_ccu) is now met. > > > > > > [1] https://github.com/wens/linux/commits/sunxi-ng-prcm > > > > It looks good now, thanks. > > > > Beside from the minor fix you've mentionned, please don't use the > > defined IDs (yet) in your DT patches (3 to 5). > > Thanks! (I think it's the same question we meet when merging V3s patchset, > right? :-) ) Exactly. When doing that kind of conversion, in order to not break the compilation, we would take the DT patches through the clock tree. However, this doesn't work very well when we also have DT patches that conflicts with those clock conversion ones that would need to get through arm-soc. And in this case, this is even more complex since you have the H3/H5 split patches that would need to go through yet another branch. Not using the defines for now, and sending a fix later fixes all these issues. Maxime -- Maxime Ripard, Free Electrons Embedded Linux and Kernel engineering http://free-electrons.com
Attachment:
signature.asc
Description: PGP signature
