From: Fuzzey, Martin <martin.fuzzey@flowbird.group> Sent: Saturday, May 23, 2020 2:03 AM > Hi Andy, > > On Fri, 22 May 2020, 03:01 Andy Duan, <fugang.duan@xxxxxxx> wrote: > > > > Andrew, many customers require the wol feature, NXP NPI release always > > support the wol feature to match customers requirement. > > > > And some customers' board only design one ethernet instance based on > > imx6sx/imx7d/ > > Imx8 serial, but which instance we never know, maybe enet1, maybe > > enet2. So we should supply different values for gpr. > > > > So, it is very necessary to support wol feature for multiple instances. > > > > Yes, I don't think anyone is saying otherwise. > > The problem is just that there are already .dtsi files for i.MX chips having > multiple ethernet interfaces in the mainline kernel (at least imx6ui.dtsi, > imx6sx.dts, imx7d.dtsi) but that this patch series does not modify those files > to use the new DT format. > For imx6ul/imx6sx/imx7d/imx8mq/imx8mm/imx8mn chips to support wol, I plan to submit another dts patch after the patch set is accepted. If you think add the dts patch appending to the patch set, I will add it in v2. > It currently only modifies the dts files that are already supported by > hardcoded values in the driver. > > As to not knowing which instance it shouldn't matter. > The base dtsi can declare both/all ethernet interfaces with the appropriate > GPR bits. > Both set to status = "disabled". > > Then the board specific dts file sets status="okay" and activates wol by adding > " > "fsl,magic-packet" if the hardaware supports it (because that depends on > things beyond the SoC, like how the ethernet PHY is clocked and powered.) > > Martin
