Quoting Krzysztof Kozlowski (2023-01-31 23:12:53) > On 31/01/2023 00:30, Stephen Boyd wrote: > > Quoting Krzysztof Kozlowski (2023-01-26 01:31:55) > >> On 25/01/2023 21:44, Stephen Boyd wrote: > >>> Quoting Krzysztof Kozlowski (2023-01-23 01:49:25) > >>>> There is no point to allow selecting pin-controller drivers for Qualcomm > >>> > >>> pin controllers? > >> > >> Copy-paste, I'll fix it. > >> > >>> > >>>> ARMv7 SoCs when building ARM64 kernel, and vice versa. This makes > >>>> kernel configuration more difficult as many do not remember the Qualcomm > >>>> SoCs model names/numbers. There won't be a single image for ARMv7 and > >>>> ARMv8/9 SoCs, so no features/options are lost. > >>> > >>> Are the drivers used in arm32 emulation mode on these SoCs? I recall > >>> there are some SoCs they run with the arm architecture. > >> > >> I did not add it to the few SoCs which have upstream DTS in ARM and > >> ARM64. I added only to the ones which are in one specific folder. Also > >> my patch does not affect defconfigs (qcom_defconfig and arm64/defconfig). > > > > Cool, thanks for checking. Is it possible to take a dtb from arm64 dts > > directory and boot it on an armv8 CPU running in 32-bit mode? Just > > wondering if even having the dts file exist in the arm64 architecture > > really matters here. > > If DTSI (and/or board DTS) is in arm64, you still need DTS or a link in > arm directory. If such one is added, then the restrictions here can be > removed. Have in mind that I did the same already for pinctrl. > I'm saying that you put the dts file in arch/arm64/boot/dts/, compile the dts to a dtb and stick it on a board that boots the kernel in arm32 mode. If it is possible to boot the dtb on a CPU running in arm32 mode then this restriction should be loosened to depends on ARM || ARM64 || COMPILE_TEST
