On Wed, Feb 22, 2023 at 09:51:21AM +0100, Krzysztof Kozlowski wrote: > On 21/02/2023 18:49, Otto Pflüger wrote: > > This driver provides clocks, resets and power domains needed for various > > components of the MSM8917 SoC and the very similar QM215 SoC. > > > > According to [1] in the downstream kernel, the GPU clock has a different > > source mapping on QM215 (gcc_gfx3d_map vs gcc_gfx3d_map_qm215). > > > > [1]: https://git.codelinaro.org/clo/la/kernel/msm-4.9/-/blob/LF.UM.8.6.2-28000-89xx.0/include/dt-bindings/clock/msm-clocks-hwio-8952.h#L298 > > > > Signed-off-by: Otto Pflüger <otto.pflueger@xxxxxxxxx> > > --- > > drivers/clk/qcom/Kconfig | 8 + > > drivers/clk/qcom/Makefile | 1 + > > drivers/clk/qcom/gcc-msm8917.c | 3283 ++++++++++++++++++++++++++++++++ > > 3 files changed, 3292 insertions(+) > > create mode 100644 drivers/clk/qcom/gcc-msm8917.c > > > > diff --git a/drivers/clk/qcom/Kconfig b/drivers/clk/qcom/Kconfig > > index 70d43f0a8919..3ef58b09385a 100644 > > --- a/drivers/clk/qcom/Kconfig > > +++ b/drivers/clk/qcom/Kconfig > > @@ -196,6 +196,14 @@ config MSM_GCC_8916 > > Say Y if you want to use devices such as UART, SPI i2c, USB, > > SD/eMMC, display, graphics, camera etc. > > > > +config MSM_GCC_8917 > > + tristate "MSM8917 Global Clock Controller" > > I propose to add here: > > depends on ARM64 || COMPILE_TEST Why should this driver only be enabled on ARM64 systems? In my opinion, this should also be supported on 32-bit ARM, just like the other clock drivers for the 64-bit SoCs. Users should be able to boot a 32-bit kernel if they have to, e.g. if their device has broken firmware that prohibits booting 64-bit kernels (there have been such cases with MSM8916) or if they think that 64-bit systems use too much memory (my QM215 device with 512MB RAM shipped with a 32-bit kernel for this very reason). > > > + select QCOM_GDSC > > + help > > + Support for the global clock controller on msm8917 devices. > > + Say Y if you want to use devices such as UART, SPI i2c, USB, > > + SD/eMMC, display, graphics, camera etc. > > + > > (...) > > + > > +static int gcc_msm8917_probe(struct platform_device *pdev) > > +{ > > + struct regmap *regmap; > > + > > + regmap = qcom_cc_map(pdev, &gcc_msm8917_desc); > > + if (IS_ERR(regmap)) > > + return PTR_ERR(regmap); > > + > > + if (of_device_is_compatible(pdev->dev.of_node, "qcom,gcc-qm215")) > > Use data in of_device_id instead. This scales poorly if any new variant > is added here. Unless no new variants will be added? Ever? Thanks. There are a lot of other related SoCs (MSM8920, MSM8937, MSM8940, MSM8952, SDM429 and SDM439) which could get added here as variants, so I'll implement this using data in of_device_id. Regards, Otto Pflüger
