Skip to contentSkip to conversation

Kirkwoode cpufreq driver

PATCH Applied Andrew Lunn 5 participants 17 messages Jan 27, 2013 – May 26, 2013

Subsystems (5)

v3 of 5 Applied Latest is v5 →

lore
Share & tools
lore

Andrew Lunn

author maintainer
#0 lore raw

v3 cover Kirkwoode cpufreq driver

This patchset adds a cpufreq driver for Marvell Kirkwood SoCs.

The changes to kirkwood_defconfig enable it and set the default governor to ondemand.

Changes since v2:

Remove unneeded #include file left over from debugging. WARN_ON(1) instead of dev_err() for unexpected state Various blank lines removed.
Reformat comment.
writel -> writel_relaxed

Changes since v1:

tabify Kconfig.arm entry
Sort order of include files
Remove some unnecassary include files Reformat multiline comment to be coding style conform.

Andrew Lunn (3):
cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs
arm: kirkwood: Instantiate cpufreq driver
arm: kirkwood: Enable cpufreq and ondemand on kirkwood_defconfig

 arch/arm/Kconfig                                  |    1 +
 arch/arm/configs/kirkwood_defconfig               |    3 +
 arch/arm/mach-kirkwood/board-dt.c                 |    3 +-
 arch/arm/mach-kirkwood/common.c                   |   23 ++
 arch/arm/mach-kirkwood/common.h                   |    2 +
 arch/arm/mach-kirkwood/include/mach/bridge-regs.h |    2 +
 drivers/clk/mvebu/clk-gating-ctrl.c               |    1 +
 drivers/cpufreq/Kconfig.arm                       |    6 +
 drivers/cpufreq/Makefile                          |    1 +
 drivers/cpufreq/kirkwood-cpufreq.c                |  264 +++++++++++++++++++++
 10 files changed, 305 insertions(+), 1 deletion(-)
 create mode 100644 drivers/cpufreq/kirkwood-cpufreq.c
Signature · 1 line
1.7.10.4
author maintainer
#1 lore raw

PATCH v3 1/3 cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

The Marvell Kirkwood SoCs have simple cpufreq support in hardware. The CPU can either use the a high speed cpu clock, or the slower DDR clock. Add a driver to swap between these two clock sources.

---
 drivers/clk/mvebu/clk-gating-ctrl.c |    1 +
 drivers/cpufreq/Kconfig.arm         |    6 +
 drivers/cpufreq/Makefile            |    1 +
 drivers/cpufreq/kirkwood-cpufreq.c  |  264 +++++++++++++++++++++++++++++++++++
 4 files changed, 272 insertions(+)
 create mode 100644 drivers/cpufreq/kirkwood-cpufreq.c
drivers/clk/mvebu/clk-gating-ctrl.c +1 -0
index 8fa5408..ebf141d 100644
@@ -193,6 +193,7 @@ static const struct mvebu_soc_descr __initconst kirkwood_gating_descr[] = {
 	{ "runit", NULL, 7 },
 	{ "xor0", NULL, 8 },
 	{ "audio", NULL, 9 },
+	{ "powersave", "cpuclk", 11 },
 	{ "sata0", NULL, 14 },
 	{ "sata1", NULL, 15 },
 	{ "xor1", NULL, 16 },
drivers/cpufreq/Kconfig.arm +6 -0
index a0b3661..80c8229 100644
@@ -77,6 +77,12 @@ config ARM_EXYNOS5250_CPUFREQ
 	  This adds the CPUFreq driver for Samsung EXYNOS5250
 	  SoC.
 
+config ARM_KIRKWOOD_CPUFREQ
+	def_bool ARCH_KIRKWOOD && OF
+	help
+	  This adds the CPUFreq driver for Marvell Kirkwood
+	  SoCs.
+
 config ARM_SPEAR_CPUFREQ
 	bool "SPEAr CPUFreq support"
 	depends on PLAT_SPEAR
drivers/cpufreq/Makefile +1 -0
index fadc4d4..39a0ffe 100644
@@ -50,6 +50,7 @@ obj-$(CONFIG_ARM_EXYNOS_CPUFREQ)	+= exynos-cpufreq.o
 obj-$(CONFIG_ARM_EXYNOS4210_CPUFREQ)	+= exynos4210-cpufreq.o
 obj-$(CONFIG_ARM_EXYNOS4X12_CPUFREQ)	+= exynos4x12-cpufreq.o
 obj-$(CONFIG_ARM_EXYNOS5250_CPUFREQ)	+= exynos5250-cpufreq.o
+obj-$(CONFIG_ARM_KIRKWOOD_CPUFREQ)	+= kirkwood-cpufreq.o
 obj-$(CONFIG_ARM_OMAP2PLUS_CPUFREQ)     += omap-cpufreq.o
 obj-$(CONFIG_ARM_SPEAR_CPUFREQ)		+= spear-cpufreq.o
 
drivers/cpufreq/kirkwood-cpufreq.c new file +264 -0

264 lines not on this page · raw message

Signature · 1 line
1.7.10.4
maintainer
#2 lore raw

Re: [PATCH v3 1/3] cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

Quoted 5 lines · Andrew Lunn
drivers/cpufreq/kirkwood-cpufreq.c +3 -0
+static void kirkwood_cpufreq_set_cpu_state(unsigned int index)
+{
 
+       if (freqs.old != freqs.new) {

drivers/cpufreq/kirkwood-cpufreq.c line 91

Quoted 10 lines
drivers/cpufreq/kirkwood-cpufreq.c +9 -0
 
+               switch (state) {
+               case STATE_CPU_FREQ:
+                       clk_disable(priv.powersave_clk);
+                       break;
+               case STATE_DDR_FREQ:
+                       clk_enable(priv.powersave_clk);
+                       break;
+               default:
+                       WARN_ON(1);

I still don't feel this case is required :)

Quoted 7 lines
drivers/cpufreq/kirkwood-cpufreq.c +6 -0
+               }
 
+static struct cpufreq_driver kirkwood_cpufreq_driver = {
+       .get    = kirkwood_cpufreq_get_cpu_frequency,
+       .verify = kirkwood_cpufreq_verify,
+       .target = kirkwood_cpufreq_target,
+       .init   = kirkwood_cpufreq_cpu_init,

drivers/cpufreq/kirkwood-cpufreq.c line 170

Quoted 10 lines
drivers/cpufreq/kirkwood-cpufreq.c +10 -0
+       .exit   = kirkwood_cpufreq_cpu_exit,
+       .name   = "kirkwood_freq",
+       .owner  = THIS_MODULE,
+       .attr   = kirkwood_cpufreq_attr,
+};
+
+static int kirkwood_cpufreq_probe(struct platform_device *pdev)
+{
+       struct device_node *np = of_find_compatible_node(
+               NULL, NULL, "marvell,kirkwood-core-clock");

np can be NULL here, want to check?

drivers/cpufreq/kirkwood-cpufreq.c line 214

Quoted 2 lines
drivers/cpufreq/kirkwood-cpufreq.c +2 -0
+       np = of_find_compatible_node(NULL, NULL,
+                                    "marvell,kirkwood-gating-clock");

here too.

drivers/cpufreq/kirkwood-cpufreq.c line 257

Quoted 10 lines
drivers/cpufreq/kirkwood-cpufreq.c +9 -0
+}
 
+static struct platform_driver kirkwood_cpufreq_platform_driver = {
+       .probe = kirkwood_cpufreq_probe,
+       .remove = kirkwood_cpufreq_remove,
+       .driver = {
+               .name = "kirkwood-cpufreq",
+               .owner = THIS_MODULE,
+       },
+};

Two things. Any reason why you created an extra layer of platform_driver? You could have called cpufreq_probe/remove (with names modified) from module init/exit.

Also, in case you want to keep it (which i don't feel is required), then would make
sense to keep same names in .name field for both structures.

author maintainer
#3 lore raw

Re: [PATCH v3 1/3] cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

Quoted 17 lines · Viresh Kumar
Quoted 5 lines · Andrew Lunn
drivers/cpufreq/kirkwood-cpufreq.c +3 -0
+static void kirkwood_cpufreq_set_cpu_state(unsigned int index)
+{
 
+       if (freqs.old != freqs.new) {

drivers/cpufreq/kirkwood-cpufreq.c line 91

Quoted 10 lines
drivers/cpufreq/kirkwood-cpufreq.c +9 -0
 
+               switch (state) {
+               case STATE_CPU_FREQ:
+                       clk_disable(priv.powersave_clk);
+                       break;
+               case STATE_DDR_FREQ:
+                       clk_enable(priv.powersave_clk);
+                       break;
+               default:
+                       WARN_ON(1);
Quoted 1 line

I still don't feel this case is required :)

O.K, i will take it out.

Quoted 4 lines
Quoted 4 lines
drivers/cpufreq/kirkwood-cpufreq.c +4 -0
+static struct platform_driver kirkwood_cpufreq_platform_driver = {
+       .probe = kirkwood_cpufreq_probe,
+       .remove = kirkwood_cpufreq_remove,
+       .driver = {
Quoted 8 lines

drivers/cpufreq/kirkwood-cpufreq.c line 257

Quoted 4 lines
drivers/cpufreq/kirkwood-cpufreq.c +4 -0
+               .name = "kirkwood-cpufreq",
+               .owner = THIS_MODULE,
+       },
+};

Two things. Any reason why you created an extra layer of platform_driver? You could have called cpufreq_probe/remove (with names modified) from module init/exit.

And how would the module get loaded? There is no hardware anchor to make the module load. No enumeration of some bus causing it to be loaded. In the ARM world, platform drivers are the norm.

Andrew

maintainer
#4 lore raw

Re: [PATCH v3 1/3] cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

Quoted 3 lines · Andrew Lunn

And how would the module get loaded? There is no hardware anchor to make the module load. No enumeration of some bus causing it to be loaded. In the ARM world, platform drivers are the norm.

The way you do it now is by creating a platform device for it in your arch/arm/mach-* directory. And you are passing an IOMEM resource too.

I believe, normally we don't require any DT node or platform device from arch/arm/mach-* for cpufreq drivers. This is something which should always be initialized once it is selected in .config.

Because we will call the _probe() (or renamed as kirkwood_cpufreq_init) from module_init() now, it will get initialized directly. No need to any enumeration at all. Now, in case there are multiple values of IOMEM resource depending on machine type, you can create properties in cpu node. Otherwise just embed the address directly into driver as nobody else is going to use it.

Makes sense?

--
viresh

author maintainer
#5 lore raw

Re: [PATCH v3 1/3] cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

Quoted 11 lines · Viresh Kumar
Quoted 3 lines · Andrew Lunn

And how would the module get loaded? There is no hardware anchor to make the module load. No enumeration of some bus causing it to be loaded. In the ARM world, platform drivers are the norm.

The way you do it now is by creating a platform device for it in your arch/arm/mach-* directory. And you are passing an IOMEM resource too.

I believe, normally we don't require any DT node or platform device from arch/arm/mach-* for cpufreq drivers. This is something which should always be initialized once it is selected in .config.

What current happens is that most of the drivers use a late_initcall():

linux/drivers/cpufreq$ grep late_initcall * acpi-cpufreq.c:late_initcall(acpi_cpufreq_init); cpufreq-cpu0.c:late_initcall(cpu0_cpufreq_driver_init); exynos-cpufreq.c:late_initcall(exynos_cpufreq_init); longhaul.c:late_initcall(longhaul_init); p4-clockmod.c:late_initcall(cpufreq_p4_init); pcc-cpufreq.c:late_initcall(pcc_cpufreq_init); powernow-k7.c:late_initcall(powernow_init); powernow-k8.c:late_initcall(powernowk8_init); s5pv210-cpufreq.c:late_initcall(s5pv210_cpufreq_init); spear-cpufreq.c:late_initcall(spear_cpufreq_driver_init); speedstep-centrino.c:late_initcall(centrino_init);

So when we have a multiplatform kernel with many of these drivers built in, all but one are going to notice they are not on the hardware they support, and return -ENODEV.

By making it a platform driver, the kirkwood cpufreq driver will only get loaded on kirkwood systems, and won't slow down the boot for everybody else.

Andrew

maintainer
#6 lore raw

Re: [PATCH v3 1/3] cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

Quoted 7 lines · Andrew Lunn

So when we have a multiplatform kernel with many of these drivers built in, all but one are going to notice they are not on the hardware they support, and return -ENODEV.

By making it a platform driver, the kirkwood cpufreq driver will only get loaded on kirkwood systems, and won't slow down the boot for everybody else.

I tried to grep platform_driver in drivers/cpufreq/ and got only:
drivers/cpufreq/dbx500-cpufreq.c

Now, the counter argument to your explanation is, multiplatform kernel would be built only for testing purpose and not for real product. So, even this kind of delay is really not a big issue. On the other hand with platform driver too, we will hit lot of other code and would consume some extra memory for keeping its structures :)

--
viresh

author maintainer
#7 lore raw

Re: [PATCH v3 1/3] cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

Quoted 9 lines · Viresh Kumar
Quoted 7 lines · Andrew Lunn

So when we have a multiplatform kernel with many of these drivers built in, all but one are going to notice they are not on the hardware they support, and return -ENODEV.

By making it a platform driver, the kirkwood cpufreq driver will only get loaded on kirkwood systems, and won't slow down the boot for everybody else.

Quoted 5 lines

I tried to grep platform_driver in drivers/cpufreq/ and got only:
drivers/cpufreq/dbx500-cpufreq.c

Now, the counter argument to your explanation is, multiplatform kernel would be built only for testing purpose and not for real product.

I expect Debian, Fedora etc will strongly disagree with you there. They want one kernel that will run on as many products they support as possible. Kirkwood is mostly used in NAS boxes and is supported by all these distributions.

Now for a SoC used in a deeply embedded system, using a custom distribution, buildroot, etc, multiplatform is probably not wanted. But the products i've seen using Kirkwood don't fit this use case.

Quoted 3 lines

So, even this kind of delay is really not a big issue. On the other hand with platform driver too, we will hit lot of other code and would consume some extra memory for keeping its structures

Kirkwood has many platform drivers, all this code is already pulled in, lots of structures already exist. The incremental costs of another platform device is minimal.

Andrew

maintainer
#8 lore raw

Re: [PATCH v3 1/3] cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

Quoted 9 lines · Andrew Lunn

I expect Debian, Fedora etc will strongly disagree with you there. They want one kernel that will run on as many products they support as possible. Kirkwood is mostly used in NAS boxes and is supported by all these distributions.

Now for a SoC used in a deeply embedded system, using a custom distribution, buildroot, etc, multiplatform is probably not wanted. But the products i've seen using Kirkwood don't fit this use case.

Accepted. I was wrong with my comment here :)

Quoted 7 lines
Quoted 3 lines

So, even this kind of delay is really not a big issue. On the other hand with platform driver too, we will hit lot of other code and would consume some extra memory for keeping its structures

Kirkwood has many platform drivers, all this code is already pulled in, lots of structures already exist. The incremental costs of another platform device is minimal.

Its about delay executing other code too for platform device.

Other than that, there is one more issue with this style. It against DT philosophy. :(

We really shouldn't add any devices from arch/arm/mach-* for any new drivers. And when we have DT support in driver, then it doesn't make any sense. So, you really need to pass this from cpu node in DT.

--
viresh

#9 lore raw

Re: [PATCH v3 1/3] cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

In reply to Andrew Lunn

drivers/cpufreq/kirkwood-cpufreq.c line 31

Quoted 8 lines · Andrew Lunn
drivers/cpufreq/kirkwood-cpufreq.c new file +8 -0
+static struct priv
+{
+	struct clk *cpu_clk;
+	struct clk *ddr_clk;
+	struct clk *powersave_clk;
+	struct device *dev;
+	void __iomem *base;
+} priv;

I guess you probably think that the compiler will do something special with this, like reference these all through a base address and offset, but you'd be wrong there. This is no different from declaring each as an individual static variable.

drivers/cpufreq/kirkwood-cpufreq.c line 53

Quoted 3 lines
drivers/cpufreq/kirkwood-cpufreq.c new file +3 -0
+static unsigned int kirkwood_cpufreq_get_cpu_frequency(unsigned int cpu)
+{
+	if (__clk_is_enabled(priv.powersave_clk))

This looks to me to be a layering violation.

drivers/cpufreq/kirkwood-cpufreq.c line 170

Quoted 4 lines
drivers/cpufreq/kirkwood-cpufreq.c new file +4 -0
+static int kirkwood_cpufreq_probe(struct platform_device *pdev)
+{
+	struct device_node *np = of_find_compatible_node(
+		NULL, NULL, "marvell,kirkwood-core-clock");

What if np is NULL?

Quoted 6 lines
drivers/cpufreq/kirkwood-cpufreq.c new file +6 -0
+
+	struct of_phandle_args clkspec;
+	struct resource *res;
+	int err;
+
+	priv.dev = &pdev->dev;

drivers/cpufreq/kirkwood-cpufreq.c line 186

Quoted 10 lines
drivers/cpufreq/kirkwood-cpufreq.c new file +10 -0
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (!res) {
+		dev_err(&pdev->dev, "Cannot get memory resource\n");
+		return -ENODEV;
+	}
+	priv.base = devm_request_and_ioremap(&pdev->dev, res);
+	if (!priv.base) {
+		dev_err(&pdev->dev, "Cannot ioremap\n");
+		return -ENOMEM;

A number of people have been pointing out that the right return value for this is -EADDRNOTAVAIL as documented against devm_request_and_ioremap().

drivers/cpufreq/kirkwood-cpufreq.c line 193

Quoted 7 lines
drivers/cpufreq/kirkwood-cpufreq.c new file +7 -0
+	}
+
+	clkspec.np = np;
+	clkspec.args_count = 1;
+	clkspec.args[0] = 1;
+
+	priv.cpu_clk = of_clk_get_from_provider(&clkspec);

Oh, yet another way to get clocks...

author maintainer
#10 lore raw

Re: [PATCH v3 1/3] cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

Quoted 7 lines

drivers/cpufreq/kirkwood-cpufreq.c line 193

Quoted 5 lines
drivers/cpufreq/kirkwood-cpufreq.c new file +5 -0
+	clkspec.np = np;
+	clkspec.args_count = 1;
+	clkspec.args[0] = 1;
+
+	priv.cpu_clk = of_clk_get_from_provider(&clkspec);

Oh, yet another way to get clocks...

Yep. I didn't like it, but could not find a better way. It has been argued that cpufreq drivers should not have nodes in DT. So the normal of_clk_get() does not work. Since the clocks themselves are instantiated from DT, there are no clkdev alias, so plain clk_get() also does not work.

Do you know of a better way to do this?

Thanks
Andrew

#11 lore raw

Re: [PATCH v3 1/3] cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

Quoted 8 lines · Andrew Lunn
Quoted 7 lines

drivers/cpufreq/kirkwood-cpufreq.c line 193

Quoted 5 lines
drivers/cpufreq/kirkwood-cpufreq.c new file +5 -0
+	clkspec.np = np;
+	clkspec.args_count = 1;
+	clkspec.args[0] = 1;
+
+	priv.cpu_clk = of_clk_get_from_provider(&clkspec);

Oh, yet another way to get clocks...

Quoted 6 lines

Yep. I didn't like it, but could not find a better way. It has been argued that cpufreq drivers should not have nodes in DT. So the normal of_clk_get() does not work. Since the clocks themselves are instantiated from DT, there are no clkdev alias, so plain clk_get() also does not work.

But you should have a cpu node in DT, and it will be perfectly fine to have cpu node consume that core_clk.

Shawn

author maintainer
#12 lore raw

Re: [PATCH v3 1/3] cpufreq: kirkwood: Add a cpufreq driver for Marvell Kirkwood SoCs

In reply to Russell King - ARM Linux
Quoted 12 lines · Russell King - ARM Linux

drivers/cpufreq/kirkwood-cpufreq.c line 31

Quoted 8 lines · Andrew Lunn
drivers/cpufreq/kirkwood-cpufreq.c new file +8 -0
+static struct priv
+{
+	struct clk *cpu_clk;
+	struct clk *ddr_clk;
+	struct clk *powersave_clk;
+	struct device *dev;
+	void __iomem *base;
+} priv;

I guess you probably think that the compiler will do something special with this

Nope, i expect nothing at all from the compiler. I just know i need only one of these structures so i statically allocated it. I could allocate it dynamically, probably get the cleanup wrong in the error path and get shouted at by Russel King. Oh well...

Quoted 5 lines

drivers/cpufreq/kirkwood-cpufreq.c line 53

Quoted 3 lines
drivers/cpufreq/kirkwood-cpufreq.c new file +3 -0
+static unsigned int kirkwood_cpufreq_get_cpu_frequency(unsigned int cpu)
+{
+	if (__clk_is_enabled(priv.powersave_clk))

This looks to me to be a layering violation.

Possibly is. Not sure. This is a gated clock, so clk_get_rate() returns the rate of the parent clock independent of if the gated clock is enabled or not. So that does not help me. The cpufreq driver needs to cause a transition on this clock, either disabled->enabled, or enabled->disabled, then do a WFI, and once the system is stable it wakes up. If there is no transition, it was already enabled and all that clk_enable() does is increment the count, the WFI never exits and the system sleeps forever. I'm assuming here that no other driver is using this clock, which i think is a sensible assumption. But i've no idea of the initial state of the clock. Did uboot enable it or not?

Thanks
Andrew

author maintainer
#13 lore raw

PATCH v3 2/3 arm: kirkwood: Instantiate cpufreq driver

In reply to Andrew Lunn

Register a platform driver structure for the cpufreq driver.

---
 arch/arm/Kconfig                                  |    1 +
 arch/arm/mach-kirkwood/board-dt.c                 |    3 ++-
 arch/arm/mach-kirkwood/common.c                   |   23 +++++++++++++++++++++
 arch/arm/mach-kirkwood/common.h                   |    2 ++
 arch/arm/mach-kirkwood/include/mach/bridge-regs.h |    2 ++
 5 files changed, 30 insertions(+), 1 deletion(-)
arch/arm/Kconfig +1 -0
index 67874b8..830975b 100644
@@ -543,6 +543,7 @@ config ARCH_DOVE
 
 config ARCH_KIRKWOOD
 	bool "Marvell Kirkwood"
+	select ARCH_HAS_CPUFREQ
 	select ARCH_REQUIRE_GPIOLIB
 	select CPU_FEROCEON
 	select GENERIC_CLOCKEVENTS
arch/arm/mach-kirkwood/board-dt.c +2 -1
index de4fd2b..fab541d 100644
@@ -70,7 +70,6 @@ static void __init kirkwood_legacy_clk_init(void)
 	clkspec.args[0] = CGC_BIT_SDIO;
 	orion_clkdev_add(NULL, "mvsdio",
 			 of_clk_get_from_provider(&clkspec));
-
 }
 
 static void __init kirkwood_of_clk_init(void)
@@ -95,6 +94,8 @@ static void __init kirkwood_dt_init(void)
 
 	kirkwood_l2_init();
 
+	kirkwood_cpufreq_init();
+
 	/* Setup root of clk tree */
 	kirkwood_of_clk_init();
 
arch/arm/mach-kirkwood/common.c +23 -0
index bac21a5..a3dc21c 100644
@@ -584,6 +584,29 @@ void __init kirkwood_audio_init(void)
 }
 
 /*****************************************************************************
+ * CPU Frequency
+ ****************************************************************************/
+static struct resource kirkwood_cpufreq_resources[] = {
+	[0] = {
+		.start  = CPU_CONTROL_PHYS,
+		.end    = CPU_CONTROL_PHYS + 3,
+		.flags  = IORESOURCE_MEM,
+	},
+};
+
+static struct platform_device kirkwood_cpufreq_device = {
+	.name		= "kirkwood-cpufreq",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(kirkwood_cpufreq_resources),
+	.resource	= kirkwood_cpufreq_resources,
+};
+
+void __init kirkwood_cpufreq_init(void)
+{
+	platform_device_register(&kirkwood_cpufreq_device);
+}
+
+/*****************************************************************************
  * General
  ****************************************************************************/
 /*
arch/arm/mach-kirkwood/common.h +2 -0
index 5ffa57f..9ede04b 100644
@@ -50,6 +50,8 @@ void kirkwood_nand_init(struct mtd_partition *parts, int nr_parts, int delay);
 void kirkwood_nand_init_rnb(struct mtd_partition *parts, int nr_parts,
 			    int (*dev_ready)(struct mtd_info *));
 void kirkwood_audio_init(void);
+void kirkwood_cpufreq_init(void);
+
 void kirkwood_restart(char, const char *);
 void kirkwood_clk_init(void);
 
arch/arm/mach-kirkwood/include/mach/bridge-regs.h +2 -0
index 5c82b7d..d4cbe5e 100644
@@ -17,6 +17,7 @@
 #define CPU_CONFIG_ERROR_PROP	0x00000004
 
 #define CPU_CONTROL		(BRIDGE_VIRT_BASE + 0x0104)
+#define CPU_CONTROL_PHYS	(BRIDGE_PHYS_BASE + 0x0104)
 #define CPU_RESET		0x00000002
 
 #define RSTOUTn_MASK		(BRIDGE_VIRT_BASE + 0x0108)
@@ -69,6 +70,7 @@
 #define CGC_RUNIT		(1 << 7)
 #define CGC_XOR0		(1 << 8)
 #define CGC_AUDIO		(1 << 9)
+#define CGC_POWERSAVE           (1 << 11)
 #define CGC_SATA0		(1 << 14)
 #define CGC_SATA1		(1 << 15)
 #define CGC_XOR1		(1 << 16)
Signature · 1 line
1.7.10.4
#14 lore raw

Re: [PATCH v3 2/3] arm: kirkwood: Instantiate cpufreq driver

Quoted 10 lines · Andrew Lunn

Register a platform driver structure for the cpufreq driver.

---
 arch/arm/Kconfig                                  |    1 +
 arch/arm/mach-kirkwood/board-dt.c                 |    3 ++-
 arch/arm/mach-kirkwood/common.c                   |   23 +++++++++++++++++++++
 arch/arm/mach-kirkwood/common.h                   |    2 ++
 arch/arm/mach-kirkwood/include/mach/bridge-regs.h |    2 ++
 5 files changed, 30 insertions(+), 1 deletion(-)

Thanks again to Andrew for digging this one up. Applied to mvebu/soc

thx,

Jason.

author maintainer
#15 lore raw

PATCH v3 3/3 arm: kirkwood: Enable cpufreq and ondemand on kirkwood_defconfig

In reply to Andrew Lunn

Now that we have a cpufreq driver for kirkwood, enable it in kirkwood_defconfig and set the default governer to ondemand.

---
 arch/arm/configs/kirkwood_defconfig |    3 +++
 1 file changed, 3 insertions(+)
arch/arm/configs/kirkwood_defconfig +3 -0
index 93f3794..6ecb7de 100644
@@ -55,6 +55,9 @@ CONFIG_AEABI=y
 # CONFIG_OABI_COMPAT is not set
 CONFIG_ZBOOT_ROM_TEXT=0x0
 CONFIG_ZBOOT_ROM_BSS=0x0
+CONFIG_CPU_FREQ=y
+CONFIG_CPU_FREQ_STAT_DETAILS=y
+CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y
 CONFIG_CPU_IDLE=y
 CONFIG_NET=y
 CONFIG_PACKET=y
Signature · 1 line
1.7.10.4
Applied
#16 lore raw

Re: [PATCH v3 3/3] arm: kirkwood: Enable cpufreq and ondemand on kirkwood_defconfig

Quoted 7 lines · Andrew Lunn

Now that we have a cpufreq driver for kirkwood, enable it in kirkwood_defconfig and set the default governer to ondemand.

---
 arch/arm/configs/kirkwood_defconfig |    3 +++
 1 file changed, 3 insertions(+)

Applied to mvebu/defconfig

thx,

Jason.