On Thu, Apr 15, 2021 at 05:04:39PM +0800, Yicong Yang wrote: > On 2021/4/15 16:18, Yicong Yang wrote: > > On 2021/4/15 2:06, Geert Uytterhoeven wrote: > >> On Wed, Apr 14, 2021 at 11:24 AM Yicong Yang <yangyicong@xxxxxxxxxxxxx> wrote: > >>> On 2021/4/13 20:26, Geert Uytterhoeven wrote: > >>>> The HiSilicon Kunpeng I2C controller is only present on HiSilicon > >>>> Kunpeng SoCs, and its driver relies on ACPI to probe for its presence. > >>>> Hence add dependencies on ARCH_HISI and ACPI, to prevent asking the user > >>>> about this driver when configuring a kernel without Hisilicon platform > >>>> or ACPI firmware support. > >>> > >>> this is a public IP which doesn't specifically depend on ARCH_HISI. I'm > >>> not sure all the platform this IP on has ARCH_HISI configured. The driver > >>> will not be compiled by default config. This is not correct to have > >>> this dependence. > >> > >> Thanks for your answer! > >> > >> I guess it's still fine to add a dependency on ACPI? > > > > yes. currently we only use this driver through ACPI. So at least > > for this driver, it make sense to keep the dependency. > > > > sorry. i was a little mess about this. I dropped this in [1]. > so just keep it as is. > > [1] https://lore.kernel.org/linux-i2c/YGMntYT2iz72wgrd@xxxxxxxxxxxxxxxxxx/ If you think that ACPI dependency is good to have there, go ahead, not my worries of the consequences. I just consider that as unneeded dependencies. The proper fix would be to have a split in Kbuild infra for compile dependencies and run-time dependencies. +Cc: Masahiro for the discussion, maybe it had already taken place and there is an impediment to do so. -- With Best Regards, Andy Shevchenko
