| Message ID | 20190827064033.28973-25-peng.fan@nxp.com |
|---|---|
| State | Accepted |
| Commit | e663c703b16b515ce0038f971c5d10d33884869b |
| Delegated to: | Stefano Babic |
| Headers | show |
| Series | i.MX8MM support | expand |
Hi Peng, On 27.08.19 08:25, Peng Fan wrote: > Because we need to get cpu freq in print_cpuinfo at very early stage, > so we need to make sure the ccm be probed. > > Signed-off-by: Peng Fan <peng.fan@nxp.com> > --- > arch/arm/mach-imx/imx8m/soc.c | 17 +++++++++++++++++ > 1 file changed, 17 insertions(+) > > diff --git a/arch/arm/mach-imx/imx8m/soc.c b/arch/arm/mach-imx/imx8m/soc.c > index 3a54db4898..f904049120 100644 > --- a/arch/arm/mach-imx/imx8m/soc.c > +++ b/arch/arm/mach-imx/imx8m/soc.c > @@ -14,6 +14,7 @@ > #include <asm/mach-imx/boot_mode.h> > #include <asm/mach-imx/syscounter.h> > #include <asm/armv8/mmu.h> > +#include <dm/uclass.h> > #include <errno.h> > #include <fdt_support.h> > #include <fsl_wdog.h> > @@ -228,6 +229,22 @@ static void imx_set_wdog_powerdown(bool enable) > writew(enable, &wdog3->wmcr); > } > > +int arch_cpu_init_dm(void) > +{ > + struct udevice *dev; > + int ret; > + > + ret = uclass_get_device_by_name(UCLASS_CLK, > + "clock-controller@30380000", > + &dev); I just upgraded my working tree to v2019.10-rc3 + your v3 MX8MM patches. This somehow breaks things for me. The U-Boot proper hangs when calling uclass_get_device_by_name(). Removing this call makes the boot work again. Do you have any idea what could be wrong? Do I miss something? Thanks, Frieder > + if (ret < 0) { > + printf("Failed to find clock node. Check device tree\n"); > + return ret; > + } > + > + return 0; > +} > + > int arch_cpu_init(void) > { > struct ocotp_regs *ocotp = (struct ocotp_regs *)OCOTP_BASE_ADDR; >
Hi Frieder, > Subject: Re: [PATCH V3 24/27] imx8m: soc: probe clock device in > arch_cpu_init_dm > > Hi Peng, > > On 27.08.19 08:25, Peng Fan wrote: > > Because we need to get cpu freq in print_cpuinfo at very early stage, > > so we need to make sure the ccm be probed. > > > > Signed-off-by: Peng Fan <peng.fan@nxp.com> > > --- > > arch/arm/mach-imx/imx8m/soc.c | 17 +++++++++++++++++ > > 1 file changed, 17 insertions(+) > > > > diff --git a/arch/arm/mach-imx/imx8m/soc.c > > b/arch/arm/mach-imx/imx8m/soc.c index 3a54db4898..f904049120 > 100644 > > --- a/arch/arm/mach-imx/imx8m/soc.c > > +++ b/arch/arm/mach-imx/imx8m/soc.c > > @@ -14,6 +14,7 @@ > > #include <asm/mach-imx/boot_mode.h> > > #include <asm/mach-imx/syscounter.h> > > #include <asm/armv8/mmu.h> > > +#include <dm/uclass.h> > > #include <errno.h> > > #include <fdt_support.h> > > #include <fsl_wdog.h> > > @@ -228,6 +229,22 @@ static void imx_set_wdog_powerdown(bool > enable) > > writew(enable, &wdog3->wmcr); > > } > > > > +int arch_cpu_init_dm(void) > > +{ > > + struct udevice *dev; > > + int ret; > > + > > + ret = uclass_get_device_by_name(UCLASS_CLK, > > + "clock-controller@30380000", > > + &dev); > > I just upgraded my working tree to v2019.10-rc3 + your v3 MX8MM patches. > This somehow breaks things for me. The U-Boot proper hangs when calling > uclass_get_device_by_name(). Removing this call makes the boot work again. > Do you have any idea what could be wrong? Do I miss something? Please enlarge CONFIG_SYS_MALLOC_F_LEN to 0x10000 and have a try. Regards, Peng. > > Thanks, > Frieder > > > + if (ret < 0) { > > + printf("Failed to find clock node. Check device tree\n"); > > + return ret; > > + } > > + > > + return 0; > > +} > > + > > int arch_cpu_init(void) > > { > > struct ocotp_regs *ocotp = (struct ocotp_regs > *)OCOTP_BASE_ADDR; > >
On 29.08.19 03:21, Peng Fan wrote: > Hi Frieder, > >> Subject: Re: [PATCH V3 24/27] imx8m: soc: probe clock device in >> arch_cpu_init_dm >> >> Hi Peng, >> >> On 27.08.19 08:25, Peng Fan wrote: >>> Because we need to get cpu freq in print_cpuinfo at very early stage, >>> so we need to make sure the ccm be probed. >>> >>> Signed-off-by: Peng Fan <peng.fan@nxp.com> >>> --- >>> arch/arm/mach-imx/imx8m/soc.c | 17 +++++++++++++++++ >>> 1 file changed, 17 insertions(+) >>> >>> diff --git a/arch/arm/mach-imx/imx8m/soc.c >>> b/arch/arm/mach-imx/imx8m/soc.c index 3a54db4898..f904049120 >> 100644 >>> --- a/arch/arm/mach-imx/imx8m/soc.c >>> +++ b/arch/arm/mach-imx/imx8m/soc.c >>> @@ -14,6 +14,7 @@ >>> #include <asm/mach-imx/boot_mode.h> >>> #include <asm/mach-imx/syscounter.h> >>> #include <asm/armv8/mmu.h> >>> +#include <dm/uclass.h> >>> #include <errno.h> >>> #include <fdt_support.h> >>> #include <fsl_wdog.h> >>> @@ -228,6 +229,22 @@ static void imx_set_wdog_powerdown(bool >> enable) >>> writew(enable, &wdog3->wmcr); >>> } >>> >>> +int arch_cpu_init_dm(void) >>> +{ >>> + struct udevice *dev; >>> + int ret; >>> + >>> + ret = uclass_get_device_by_name(UCLASS_CLK, >>> + "clock-controller@30380000", >>> + &dev); >> >> I just upgraded my working tree to v2019.10-rc3 + your v3 MX8MM patches. >> This somehow breaks things for me. The U-Boot proper hangs when calling >> uclass_get_device_by_name(). Removing this call makes the boot work again. >> Do you have any idea what could be wrong? Do I miss something? > > Please enlarge CONFIG_SYS_MALLOC_F_LEN to 0x10000 and have a try. That did the trick. Thanks a lot! > > Regards, > Peng. > >> >> Thanks, >> Frieder >> >>> + if (ret < 0) { >>> + printf("Failed to find clock node. Check device tree\n"); >>> + return ret; >>> + } >>> + >>> + return 0; >>> +} >>> + >>> int arch_cpu_init(void) >>> { >>> struct ocotp_regs *ocotp = (struct ocotp_regs >> *)OCOTP_BASE_ADDR; >>>
On Tue, Aug 27, 2019 at 06:25:58AM +0000, Peng Fan wrote: > Because we need to get cpu freq in print_cpuinfo at very early stage, > so we need to make sure the ccm be probed. > > Signed-off-by: Peng Fan <peng.fan@nxp.com> > --- > arch/arm/mach-imx/imx8m/soc.c | 17 +++++++++++++++++ > 1 file changed, 17 insertions(+) > > diff --git a/arch/arm/mach-imx/imx8m/soc.c b/arch/arm/mach-imx/imx8m/soc.c > index 3a54db4898..f904049120 100644 > --- a/arch/arm/mach-imx/imx8m/soc.c > +++ b/arch/arm/mach-imx/imx8m/soc.c > @@ -14,6 +14,7 @@ > #include <asm/mach-imx/boot_mode.h> > #include <asm/mach-imx/syscounter.h> > #include <asm/armv8/mmu.h> > +#include <dm/uclass.h> > #include <errno.h> > #include <fdt_support.h> > #include <fsl_wdog.h> > @@ -228,6 +229,22 @@ static void imx_set_wdog_powerdown(bool enable) > writew(enable, &wdog3->wmcr); > } > > +int arch_cpu_init_dm(void) > +{ > + struct udevice *dev; > + int ret; > + > + ret = uclass_get_device_by_name(UCLASS_CLK, > + "clock-controller@30380000", > + &dev); > + if (ret < 0) { > + printf("Failed to find clock node. Check device tree\n"); > + return ret; > + } > + > + return 0; > +} > + > int arch_cpu_init(void) > { > struct ocotp_regs *ocotp = (struct ocotp_regs *)OCOTP_BASE_ADDR; > -- > 2.16.4 > > _______________________________________________ > U-Boot mailing list > U-Boot@lists.denx.de > https://lists.denx.de/listinfo/u-boot This commit broke my i.MX8MQ board. This is because the i.MX8MQ does not yet have a DM-based clock driver like i.MX8MM, so requiring the existence of such uclass device breaks i.MX8MQ. I would propse the following commit, which make the whole function only available for i.MX8MQ. --- From 644bf63ed0b33c499631be757de077f2059144e7 Mon Sep 17 00:00:00 2001 From: Patrick Wildt <patrick@blueri.se> Date: Fri, 11 Oct 2019 00:28:38 +0200 Subject: [PATCH] imx: require CCM driver only on i.MX8MM Since there is no clock driver like that for i.MX8MQ, we cannot require it on that platform, so make sure it is only built for i.MX8MM. Signed-off-by: Patrick Wildt <patrick@blueri.se> --- arch/arm/mach-imx/imx8m/soc.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/arch/arm/mach-imx/imx8m/soc.c b/arch/arm/mach-imx/imx8m/soc.c index aeca82cdbf..11ca8c86b3 100644 --- a/arch/arm/mach-imx/imx8m/soc.c +++ b/arch/arm/mach-imx/imx8m/soc.c @@ -229,6 +229,7 @@ static void imx_set_wdog_powerdown(bool enable) writew(enable, &wdog3->wmcr); } +#ifdef CONFIG_ARCH_IMX8MM int arch_cpu_init_dm(void) { struct udevice *dev; @@ -244,6 +245,7 @@ int arch_cpu_init_dm(void) return 0; } +#endif int arch_cpu_init(void) {
diff --git a/arch/arm/mach-imx/imx8m/soc.c b/arch/arm/mach-imx/imx8m/soc.c index 3a54db4898..f904049120 100644 --- a/arch/arm/mach-imx/imx8m/soc.c +++ b/arch/arm/mach-imx/imx8m/soc.c @@ -14,6 +14,7 @@ #include <asm/mach-imx/boot_mode.h> #include <asm/mach-imx/syscounter.h> #include <asm/armv8/mmu.h> +#include <dm/uclass.h> #include <errno.h> #include <fdt_support.h> #include <fsl_wdog.h> @@ -228,6 +229,22 @@ static void imx_set_wdog_powerdown(bool enable) writew(enable, &wdog3->wmcr); } +int arch_cpu_init_dm(void) +{ + struct udevice *dev; + int ret; + + ret = uclass_get_device_by_name(UCLASS_CLK, + "clock-controller@30380000", + &dev); + if (ret < 0) { + printf("Failed to find clock node. Check device tree\n"); + return ret; + } + + return 0; +} + int arch_cpu_init(void) { struct ocotp_regs *ocotp = (struct ocotp_regs *)OCOTP_BASE_ADDR;
Because we need to get cpu freq in print_cpuinfo at very early stage, so we need to make sure the ccm be probed. Signed-off-by: Peng Fan <peng.fan@nxp.com> --- arch/arm/mach-imx/imx8m/soc.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+)