On Fri, Apr 27, 2012 at 08:28:16AM +0200, Jean-Christophe PLAGNIOL-VILLARD wrote: > On 13:48 Fri 27 Apr , Shawn Guo wrote: > > On Thu, Apr 26, 2012 at 05:15:36PM +0200, Jean-Christophe PLAGNIOL-VILLARD wrote: > > > We have on Imx mxc at91 and other SoC controler hich you configure per pin > > > > > > which means one pin have multiple function and the same function is on > > > multiple pins > > > > > > so the groups are just a list of possible pins > > > > > > Instead of re-inventing bindings we do need to come with a common binding whre > > > it's possible > > > > > > So instead I proppose (send in the v2) to use common way to describe the group > > > > > Let's see how many nodes we will have in device tree. For imx6q > > example, there are 332 pins and each pin has up to 8 function selects. > > We will end up with having 332 x 8 = 2656 sub nodes under node > > "functions". Device tree simply cannot afford such a bloating. > device tree can offord it > No. Device tree maintainers has told that. Looking into the clock DT binding discussion, you will find that Grant does not like to have even 100~200 nodes to represent an entire clock tree in the DT. With your proposal (actually this has been proposed long time before), to represent the pins for a 24bit display, it easily consumes 28 nodes on mach-mxs, while my binding only needs one node. So in short, the proposal has been discussed and it's not a sensible one. Regards, Shawn > except you are going to have hundereds of duplicated pinctrl configuration > as different board will have different mux which is impossbile to maintain > either > > and I do not expect we add all the configuration possible but just the common > one > > Best Regards, > J. _______________________________________________ linux-arm-kernel mailing list linux-arm-kernel@xxxxxxxxxxxxxxxxxxx http://lists.infradead.org/mailman/listinfo/linux-arm-kernel