On Sat, Apr 10, 2021 at 8:40 PM Linus Walleij <linus.walleij@xxxxxxxxxx> wrote: > > On Sat, Apr 10, 2021 at 11:10 PM Heiko Stübner <heiko@xxxxxxxxx> wrote: > > Am Samstag, 10. April 2021, 20:30:52 CEST schrieb Peter Geis: > > > Good Afternoon, > > > > > > I'm currently working on the rk3566 early bringup support in mainline > > > and encountered an issue with this patch set. > > > Unfortunately in the rk3568/rk3566 the gpio registers switched to the > > > (16 bits write enable / 16 bits config) register format that other > > > rockchip registers use. > > > This differs from previous chips where all 32 bits were used for gpio > > > configuration. > > > The patch fails to account for this, which causes all gpios to fail to function. > > > > > > For clarity, this only affects GPIO_SWPORT_DR and GPIO_SWPORT_DDR. > > > > > > I'm currently working on a patch to fix this issue, but I know that > > > you are planning on breaking out the gpio functions into a separate > > > driver and wanted to make you aware of this immediately. > > > > just pointing to Jianqun's series providing the gpio controller support: > > > > https://lore.kernel.org/r/20210324064704.950104-1-jay.xu@xxxxxxxxxxxxxx > > > > which introduces the necessary writemask-handling but seems to need an > > update, judging by Ezequiel's reply to it. > > Does this mean that if we get a speedy and nice looking respin of the > GPIO series we probably got this problem covered? > > The merge window is getting close but it would be nice to get this in. Thank you for the responses! With a bit of effort and a quick build fix, I've successfully rebased the gpio patch set to the latest linux-next. I now have functional gpio control on the rk3566. I haven't tested them on older devices yet though. With your blessing I can send in the rebased patches in the morning, unless Jianqun wants to handle it. > > Yours, > Linus Walleij