[CPUFREQ] Processor Clocking Control interface driver

Processor Clocking Control (PCC) is an interface between the BIOS and OSPM.
Based on the server workload, OSPM can request what frequency it expects
from a logical CPU, and the BIOS will achieve that frequency transparently.

This patch introduces driver support for PCC. OSPM uses the PCC driver to
communicate with the BIOS via the PCC interface.

There is a Documentation file that provides a link to the PCC
Specification, and also provides a summary of the PCC interface.

Currently, certain HP ProLiant platforms implement the PCC interface. However,
any platform whose BIOS implements the PCC Specification, can utilize this
driver.

V2 --> V1 changes (based on Dominik's suggestions):
- Removed the dependency on CPU_FREQ_TABLE
- "cpufreq_stats" will no longer PANIC. Actually, it will not load anymore
because it is not applicable.
- Removed the sanity check for target frequency in the ->target routine.

NOTE: A patch to sanitize the target frequency requested by "ondemand" is
needed to ensure that the target freq < policy->min.

Can this driver be queued up for the 2.6.33 tree?

Signed-off-by: Naga Chumbalkar <[e-mail]>
Signed-off-by: Matthew Garrett <[e-mail]>
Signed-off-by: Thomas Renninger <[e-mail]>
Signed-off-by: Dave Jones <[e-mail]>

Documentation/cpu-freq/pcc-cpufreq.txt

···
+/*
+ *  pcc-cpufreq.txt - PCC interface documentation
+ *
+ *  Copyright (C) 2009 Red Hat, Matthew Garrett <[e-mail]>
+ *  Copyright (C) 2009 Hewlett-Packard Development Company, L.P.
+ *      Nagananda Chumbalkar <[e-mail]>
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; version 2 of the License.
+ *
+ *  This program is distributed in the hope that it will be useful, but
+ *  WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or NON
+ *  INFRINGEMENT. See the GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License along
+ *  with this program; if not, write to the Free Software Foundation, Inc.,
+ *  675 Mass Ave, Cambridge, MA 02139, USA.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ */
+
+
+			Processor Clocking Control Driver
+			---------------------------------
+
+Contents:
+---------
+1.	Introduction
+1.1	PCC interface
+1.1.1   Get Average Frequency
+1.1.2	Set Desired Frequency
+1.2	Platforms affected
+2.	Driver and /sys details
+2.1	scaling_available_frequencies
+2.2	cpuinfo_transition_latency
+2.3	cpuinfo_cur_freq
+2.4	related_cpus
+3.	Caveats
+
+1. Introduction:
+----------------
+Processor Clocking Control (PCC) is an interface between the platform
+firmware and OSPM. It is a mechanism for coordinating processor
+performance (ie: frequency) between the platform firmware and the OS.
+
+The PCC driver (pcc-cpufreq) allows OSPM to take advantage of the PCC
+interface.
+
+OS utilizes the PCC interface to inform platform firmware what frequency the
+OS wants for a logical processor. The platform firmware attempts to achieve
+the requested frequency. If the request for the target frequency could not be
+satisfied by platform firmware, then it usually means that power budget
+conditions are in place, and "power capping" is taking place.
+
+1.1 PCC interface:
+------------------
+The complete PCC specification is available here:
+http://www.acpica.org/download/Processor-Clocking-Control-v1p0.pdf
+
+PCC relies on a shared memory region that provides a channel for communication
+between the OS and platform firmware. PCC also implements a "doorbell" that
+is used by the OS to inform the platform firmware that a command has been
+sent.
+
+The ACPI PCCH() method is used to discover the location of the PCC shared
+memory region. The shared memory region header contains the "command" and
+"status" interface. PCCH() also contains details on how to access the platform
+doorbell.
+
+The following commands are supported by the PCC interface:
+* Get Average Frequency
+* Set Desired Frequency
+
+The ACPI PCCP() method is implemented for each logical processor and is
+used to discover the offsets for the input and output buffers in the shared
+memory region.
+
+When PCC mode is enabled, the platform will not expose processor performance
+or throttle states (_PSS, _TSS and related ACPI objects) to OSPM. Therefore,
+the native P-state driver (such as acpi-cpufreq for Intel, powernow-k8 for
+AMD) will not load.
+
+However, OSPM remains in control of policy. The governor (eg: "ondemand")
+computes the required performance for each processor based on server workload.
+The PCC driver fills in the command interface, and the input buffer and
+communicates the request to the platform firmware. The platform firmware is
+responsible for delivering the requested performance.
+
+Each PCC command is "global" in scope and can affect all the logical CPUs in
+the system. Therefore, PCC is capable of performing "group" updates. With PCC
+the OS is capable of getting/setting the frequency of all the logical CPUs in
+the system with a single call to the BIOS.
+
+1.1.1 Get Average Frequency:
+----------------------------
+This command is used by the OSPM to query the running frequency of the
+processor since the last time this command was completed. The output buffer
+indicates the average unhalted frequency of the logical processor expressed as
+a percentage of the nominal (ie: maximum) CPU frequency. The output buffer
+also signifies if the CPU frequency is limited by a power budget condition.
+
+1.1.2 Set Desired Frequency:
+----------------------------
+This command is used by the OSPM to communicate to the platform firmware the
+desired frequency for a logical processor. The output buffer is currently
+ignored by OSPM. The next invocation of "Get Average Frequency" will inform
+OSPM if the desired frequency was achieved or not.
+
+1.2 Platforms affected:
+-----------------------
+The PCC driver will load on any system where the platform firmware:
+* supports the PCC interface, and the associated PCCH() and PCCP() methods
+* assumes responsibility for managing the hardware clocking controls in order
+to deliver the requested processor performance
+
+Currently, certain HP ProLiant platforms implement the PCC interface. On those
+platforms PCC is the "default" choice.
+
+However, it is possible to disable this interface via a BIOS setting. In
+such an instance, as is also the case on platforms where the PCC interface
+is not implemented, the PCC driver will fail to load silently.
+
+2. Driver and /sys details:
+---------------------------
+When the driver loads, it merely prints the lowest and the highest CPU
+frequencies supported by the platform firmware.
+
+The PCC driver loads with a message such as:
+pcc-cpufreq: (v1.00.00) driver loaded with frequency limits: 1600 MHz, 2933
+MHz
+
+This means that the OPSM can request the CPU to run at any frequency in
+between the limits (1600 MHz, and 2933 MHz) specified in the message.
+
+Internally, there is no need for the driver to convert the "target" frequency
+to a corresponding P-state.
+
+The VERSION number for the driver will be of the format v.xy.ab.
+eg: 1.00.02
+   ----- --
+    |    |
+    |    -- this will increase with bug fixes/enhancements to the driver
+    |-- this is the version of the PCC specification the driver adheres to
+
+
+The following is a brief discussion on some of the fields exported via the
+/sys filesystem and how their values are affected by the PCC driver:
+
+2.1 scaling_available_frequencies:
+----------------------------------
+scaling_available_frequencies is not created in /sys. No intermediate
+frequencies need to be listed because the BIOS will try to achieve any
+frequency, within limits, requested by the governor. A frequency does not have
+to be strictly associated with a P-state.
+
+2.2 cpuinfo_transition_latency:
+-------------------------------
+The cpuinfo_transition_latency field is 0. The PCC specification does
+not include a field to expose this value currently.
+
+2.3 cpuinfo_cur_freq:
+---------------------
+A) Often cpuinfo_cur_freq will show a value different than what is declared
+in the scaling_available_frequencies or scaling_cur_freq, or scaling_max_freq.
+This is due to "turbo boost" available on recent Intel processors. If certain
+conditions are met the BIOS can achieve a slightly higher speed than requested
+by OSPM. An example:
+
+scaling_cur_freq	: 2933000
+cpuinfo_cur_freq	: 3196000
+
+B) There is a round-off error associated with the cpuinfo_cur_freq value.
+Since the driver obtains the current frequency as a "percentage" (%) of the
+nominal frequency from the BIOS, sometimes, the values displayed by
+scaling_cur_freq and cpuinfo_cur_freq may not match. An example:
+
+scaling_cur_freq	: 1600000
+cpuinfo_cur_freq	: 1583000
+
+In this example, the nominal frequency is 2933 MHz. The driver obtains the
+current frequency, cpuinfo_cur_freq, as 54% of the nominal frequency:
+
+	54% of 2933 MHz = 1583 MHz
+
+Nominal frequency is the maximum frequency of the processor, and it usually
+corresponds to the frequency of the P0 P-state.
+
+2.4 related_cpus:
+-----------------
+The related_cpus field is identical to affected_cpus.
+
+affected_cpus	: 4
+related_cpus	: 4
+
+Currently, the PCC driver does not evaluate _PSD. The platforms that support
+PCC do not implement SW_ALL. So OSPM doesn't need to perform any coordination
+to ensure that the same frequency is requested of all dependent CPUs.
+
+3. Caveats:
+-----------
+The "cpufreq_stats" module in its present form cannot be loaded and
+expected to work with the PCC driver. Since the "cpufreq_stats" module
+provides information wrt each P-state, it is not applicable to the PCC driver.

arch/x86/kernel/cpu/cpufreq/Kconfig

···
 
 comment "CPUFreq processor drivers"
 
+config X86_PCC_CPUFREQ
+	tristate "Processor Clocking Control interface driver"
+	depends on ACPI && ACPI_PROCESSOR
+	help
+	  This driver adds support for the PCC interface.
+
+	  For details, take a look at:
+	  <file:Documentation/cpu-freq/pcc-cpufreq.txt>.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called pcc-cpufreq.
+
+	  If in doubt, say N.
+
 config X86_ACPI_CPUFREQ
 	tristate "ACPI Processor P-States driver"
 	select CPU_FREQ_TABLE

arch/x86/kernel/cpu/cpufreq/Makefile

···
 
 obj-$(CONFIG_X86_POWERNOW_K8)		+= powernow-k8.o
 obj-$(CONFIG_X86_ACPI_CPUFREQ)		+= acpi-cpufreq.o
+obj-$(CONFIG_X86_PCC_CPUFREQ)		+= pcc-cpufreq.o
 obj-$(CONFIG_X86_POWERNOW_K6)		+= powernow-k6.o
 obj-$(CONFIG_X86_POWERNOW_K7)		+= powernow-k7.o
 obj-$(CONFIG_X86_LONGHAUL)		+= longhaul.o

arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c

···
+/*
+ *  pcc-cpufreq.c - Processor Clocking Control firmware cpufreq interface
+ *
+ *  Copyright (C) 2009 Red Hat, Matthew Garrett <[e-mail]>
+ *  Copyright (C) 2009 Hewlett-Packard Development Company, L.P.
+ *	Nagananda Chumbalkar <[e-mail]>
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; version 2 of the License.
+ *
+ *  This program is distributed in the hope that it will be useful, but
+ *  WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or NON
+ *  INFRINGEMENT. See the GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License along
+ *  with this program; if not, write to the Free Software Foundation, Inc.,
+ *  675 Mass Ave, Cambridge, MA 02139, USA.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/smp.h>
+#include <linux/sched.h>
+#include <linux/cpufreq.h>
+#include <linux/compiler.h>
+
+#include <linux/acpi.h>
+#include <linux/io.h>
+#include <linux/spinlock.h>
+#include <linux/uaccess.h>
+
+#include <acpi/processor.h>
+
+#define PCC_VERSION 	"1.00.00"
+#define POLL_LOOPS 	300
+
+#define CMD_COMPLETE 	0x1
+#define CMD_GET_FREQ 	0x0
+#define CMD_SET_FREQ 	0x1
+
+#define BUF_SZ		4
+
+#define dprintk(msg...) cpufreq_debug_printk(CPUFREQ_DEBUG_DRIVER,	\
+					     "pcc-cpufreq", msg)
+
+struct pcc_register_resource {
+	u8 descriptor;
+	u16 length;
+	u8 space_id;
+	u8 bit_width;
+	u8 bit_offset;
+	u8 access_size;
+	u64 address;
+} __attribute__ ((packed));
+
+struct pcc_memory_resource {
+	u8 descriptor;
+	u16 length;
+	u8 space_id;
+	u8 resource_usage;
+	u8 type_specific;
+	u64 granularity;
+	u64 minimum;
+	u64 maximum;
+	u64 translation_offset;
+	u64 address_length;
+} __attribute__ ((packed));
+
+static struct cpufreq_driver pcc_cpufreq_driver;
+
+struct pcc_header {
+	u32 signature;
+	u16 length;
+	u8 major;
+	u8 minor;
+	u32 features;
+	u16 command;
+	u16 status;
+	u32 latency;
+	u32 minimum_time;
+	u32 maximum_time;
+	u32 nominal;
+	u32 throttled_frequency;
+	u32 minimum_frequency;
+};
+
+static void __iomem *pcch_virt_addr;
+static struct pcc_header __iomem *pcch_hdr;
+
+static DEFINE_SPINLOCK(pcc_lock);
+
+static struct acpi_generic_address doorbell;
+
+static u64 doorbell_preserve;
+static u64 doorbell_write;
+
+static u8 OSC_UUID[16] = {0x63, 0x9B, 0x2C, 0x9F, 0x70, 0x91, 0x49, 0x1f,
+			  0xBB, 0x4F, 0xA5, 0x98, 0x2F, 0xA1, 0xB5, 0x46};
+
+struct pcc_cpu {
+	u32 input_offset;
+	u32 output_offset;
+};
+
+static struct pcc_cpu *pcc_cpu_info;
+
+static int pcc_cpufreq_verify(struct cpufreq_policy *policy)
+{
+	cpufreq_verify_within_limits(policy, policy->cpuinfo.min_freq,
+				     policy->cpuinfo.max_freq);
+	return 0;
+}
+
+static inline void pcc_cmd(void)
+{
+	u64 doorbell_value;
+	int i;
+
+	acpi_read(&doorbell_value, &doorbell);
+	acpi_write((doorbell_value & doorbell_preserve) | doorbell_write,
+		   &doorbell);
+
+	for (i = 0; i < POLL_LOOPS; i++) {
+		if (ioread16(&pcch_hdr->status) & CMD_COMPLETE)
+			break;
+	}
+}
+
+static inline void pcc_clear_mapping(void)
+{
+	if (pcch_virt_addr)
+		iounmap(pcch_virt_addr);
+	pcch_virt_addr = NULL;
+}
+
+static unsigned int pcc_get_freq(unsigned int cpu)
+{
+	struct pcc_cpu *pcc_cpu_data;
+	unsigned int curr_freq;
+	unsigned int freq_limit;
+	u16 status;
+	u32 input_buffer;
+	u32 output_buffer;
+
+	spin_lock(&pcc_lock);
+
+	dprintk("get: get_freq for CPU %d\n", cpu);
+	pcc_cpu_data = per_cpu_ptr(pcc_cpu_info, cpu);
+
+	input_buffer = 0x1;
+	iowrite32(input_buffer,
+			(pcch_virt_addr + pcc_cpu_data->input_offset));
+	iowrite16(CMD_GET_FREQ, &pcch_hdr->command);
+
+	pcc_cmd();
+
+	output_buffer =
+		ioread32(pcch_virt_addr + pcc_cpu_data->output_offset);
+
+	/* Clear the input buffer - we are done with the current command */
+	memset_io((pcch_virt_addr + pcc_cpu_data->input_offset), 0, BUF_SZ);
+
+	status = ioread16(&pcch_hdr->status);
+	if (status != CMD_COMPLETE) {
+		dprintk("get: FAILED: for CPU %d, status is %d\n",
+			cpu, status);
+		goto cmd_incomplete;
+	}
+	iowrite16(0, &pcch_hdr->status);
+	curr_freq = (((ioread32(&pcch_hdr->nominal) * (output_buffer & 0xff))
+			/ 100) * 1000);
+
+	dprintk("get: SUCCESS: (virtual) output_offset for cpu %d is "
+		"0x%x, contains a value of: 0x%x. Speed is: %d MHz\n",
+		cpu, (pcch_virt_addr + pcc_cpu_data->output_offset),
+		output_buffer, curr_freq);
+
+	freq_limit = (output_buffer >> 8) & 0xff;
+	if (freq_limit != 0xff) {
+		dprintk("get: frequency for cpu %d is being temporarily"
+			" capped at %d\n", cpu, curr_freq);
+	}
+
+	spin_unlock(&pcc_lock);
+	return curr_freq;
+
+cmd_incomplete:
+	iowrite16(0, &pcch_hdr->status);
+	spin_unlock(&pcc_lock);
+	return -EINVAL;
+}
+
+static int pcc_cpufreq_target(struct cpufreq_policy *policy,
+			      unsigned int target_freq,
+			      unsigned int relation)
+{
+	struct pcc_cpu *pcc_cpu_data;
+	struct cpufreq_freqs freqs;
+	u16 status;
+	u32 input_buffer;
+	int cpu;
+
+	spin_lock(&pcc_lock);
+	cpu = policy->cpu;
+	pcc_cpu_data = per_cpu_ptr(pcc_cpu_info, cpu);
+
+	dprintk("target: CPU %d should go to target freq: %d "
+		"(virtual) input_offset is 0x%x\n",
+		cpu, target_freq,
+		(pcch_virt_addr + pcc_cpu_data->input_offset));
+
+	freqs.new = target_freq;
+	freqs.cpu = cpu;
+	cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE);
+
+	input_buffer = 0x1 | (((target_freq * 100)
+			       / (ioread32(&pcch_hdr->nominal) * 1000)) << 8);
+	iowrite32(input_buffer,
+			(pcch_virt_addr + pcc_cpu_data->input_offset));
+	iowrite16(CMD_SET_FREQ, &pcch_hdr->command);
+
+	pcc_cmd();
+
+	/* Clear the input buffer - we are done with the current command */
+	memset_io((pcch_virt_addr + pcc_cpu_data->input_offset), 0, BUF_SZ);
+
+	status = ioread16(&pcch_hdr->status);
+	if (status != CMD_COMPLETE) {
+		dprintk("target: FAILED for cpu %d, with status: 0x%x\n",
+			cpu, status);
+		goto cmd_incomplete;
+	}
+	iowrite16(0, &pcch_hdr->status);
+
+	cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE);
+	dprintk("target: was SUCCESSFUL for cpu %d\n", cpu);
+	spin_unlock(&pcc_lock);
+
+	return 0;
+
+cmd_incomplete:
+	iowrite16(0, &pcch_hdr->status);
+	spin_unlock(&pcc_lock);
+	return -EINVAL;
+}
+
+static int pcc_get_offset(int cpu)
+{
+	acpi_status status;
+	struct acpi_buffer buffer = {ACPI_ALLOCATE_BUFFER, NULL};
+	union acpi_object *pccp, *offset;
+	struct pcc_cpu *pcc_cpu_data;
+	struct acpi_processor *pr;
+	int ret = 0;
+
+	pr = per_cpu(processors, cpu);
+	pcc_cpu_data = per_cpu_ptr(pcc_cpu_info, cpu);
+
+	status = acpi_evaluate_object(pr->handle, "PCCP", NULL, &buffer);
+	if (ACPI_FAILURE(status))
+		return -ENODEV;
+
+	pccp = buffer.pointer;
+	if (!pccp || pccp->type != ACPI_TYPE_PACKAGE) {
+		ret = -ENODEV;
+		goto out_free;
+	};
+
+	offset = &(pccp->package.elements[0]);
+	if (!offset || offset->type != ACPI_TYPE_INTEGER) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+
+	pcc_cpu_data->input_offset = offset->integer.value;
+
+	offset = &(pccp->package.elements[1]);
+	if (!offset || offset->type != ACPI_TYPE_INTEGER) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+
+	pcc_cpu_data->output_offset = offset->integer.value;
+
+	memset_io((pcch_virt_addr + pcc_cpu_data->input_offset), 0, BUF_SZ);
+	memset_io((pcch_virt_addr + pcc_cpu_data->output_offset), 0, BUF_SZ);
+
+	dprintk("pcc_get_offset: for CPU %d: pcc_cpu_data "
+		"input_offset: 0x%x, pcc_cpu_data output_offset: 0x%x\n",
+		cpu, pcc_cpu_data->input_offset, pcc_cpu_data->output_offset);
+out_free:
+	kfree(buffer.pointer);
+	return ret;
+}
+
+static int __init pcc_cpufreq_do_osc(acpi_handle *handle)
+{
+	acpi_status status;
+	struct acpi_object_list input;
+	struct acpi_buffer output = {ACPI_ALLOCATE_BUFFER, NULL};
+	union acpi_object in_params[4];
+	union acpi_object *out_obj;
+	u32 capabilities[2];
+	u32 errors;
+	u32 supported;
+	int ret = 0;
+
+	input.count = 4;
+	input.pointer = in_params;
+	input.count = 4;
+	input.pointer = in_params;
+	in_params[0].type               = ACPI_TYPE_BUFFER;
+	in_params[0].buffer.length      = 16;
+	in_params[0].buffer.pointer     = OSC_UUID;
+	in_params[1].type               = ACPI_TYPE_INTEGER;
+	in_params[1].integer.value      = 1;
+	in_params[2].type               = ACPI_TYPE_INTEGER;
+	in_params[2].integer.value      = 2;
+	in_params[3].type               = ACPI_TYPE_BUFFER;
+	in_params[3].buffer.length      = 8;
+	in_params[3].buffer.pointer     = (u8 *)&capabilities;
+
+	capabilities[0] = OSC_QUERY_ENABLE;
+	capabilities[1] = 0x1;
+
+	status = acpi_evaluate_object(*handle, "_OSC", &input, &output);
+	if (ACPI_FAILURE(status))
+		return -ENODEV;
+
+	if (!output.length)
+		return -ENODEV;
+
+	out_obj = output.pointer;
+	if (out_obj->type != ACPI_TYPE_BUFFER) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+
+	errors = *((u32 *)out_obj->buffer.pointer) & ~(1 << 0);
+	if (errors) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+
+	supported = *((u32 *)(out_obj->buffer.pointer + 4));
+	if (!(supported & 0x1)) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+
+	kfree(output.pointer);
+	capabilities[0] = 0x0;
+	capabilities[1] = 0x1;
+
+	status = acpi_evaluate_object(*handle, "_OSC", &input, &output);
+	if (ACPI_FAILURE(status))
+		return -ENODEV;
+
+	if (!output.length)
+		return -ENODEV;
+
+	out_obj = output.pointer;
+	if (out_obj->type != ACPI_TYPE_BUFFER) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+
+	errors = *((u32 *)out_obj->buffer.pointer) & ~(1 << 0);
+	if (errors) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+
+	supported = *((u32 *)(out_obj->buffer.pointer + 4));
+	if (!(supported & 0x1)) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+
+out_free:
+	kfree(output.pointer);
+	return ret;
+}
+
+static int __init pcc_cpufreq_probe(void)
+{
+	acpi_status status;
+	struct acpi_buffer output = {ACPI_ALLOCATE_BUFFER, NULL};
+	struct pcc_memory_resource *mem_resource;
+	struct pcc_register_resource *reg_resource;
+	union acpi_object *out_obj, *member;
+	acpi_handle handle, osc_handle;
+	int ret = 0;
+
+	status = acpi_get_handle(NULL, "\\_SB", &handle);
+	if (ACPI_FAILURE(status))
+		return -ENODEV;
+
+	status = acpi_get_handle(handle, "_OSC", &osc_handle);
+	if (ACPI_SUCCESS(status)) {
+		ret = pcc_cpufreq_do_osc(&osc_handle);
+		if (ret)
+			dprintk("probe: _OSC evaluation did not succeed\n");
+		/* Firmware's use of _OSC is optional */
+		ret = 0;
+	}
+
+	status = acpi_evaluate_object(handle, "PCCH", NULL, &output);
+	if (ACPI_FAILURE(status))
+		return -ENODEV;
+
+	out_obj = output.pointer;
+	if (out_obj->type != ACPI_TYPE_PACKAGE) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+
+	member = &out_obj->package.elements[0];
+	if (member->type != ACPI_TYPE_BUFFER) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+
+	mem_resource = (struct pcc_memory_resource *)member->buffer.pointer;
+
+	dprintk("probe: mem_resource descriptor: 0x%x,"
+		" length: %d, space_id: %d, resource_usage: %d,"
+		" type_specific: %d, granularity: 0x%llx,"
+		" minimum: 0x%llx, maximum: 0x%llx,"
+		" translation_offset: 0x%llx, address_length: 0x%llx\n",
+		mem_resource->descriptor, mem_resource->length,
+		mem_resource->space_id, mem_resource->resource_usage,
+		mem_resource->type_specific, mem_resource->granularity,
+		mem_resource->minimum, mem_resource->maximum,
+		mem_resource->translation_offset,
+		mem_resource->address_length);
+
+	if (mem_resource->space_id != ACPI_ADR_SPACE_SYSTEM_MEMORY) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+
+	pcch_virt_addr = ioremap_nocache(mem_resource->minimum,
+					mem_resource->address_length);
+	if (pcch_virt_addr == NULL) {
+		dprintk("probe: could not map shared mem region\n");
+		goto out_free;
+	}
+	pcch_hdr = pcch_virt_addr;
+
+	dprintk("probe: PCCH header (virtual) addr: 0x%llx\n",
+		(u64)pcch_hdr);
+	dprintk("probe: PCCH header is at physical address: 0x%llx,"
+		" signature: 0x%x, length: %d bytes, major: %d, minor: %d,"
+		" supported features: 0x%x, command field: 0x%x,"
+		" status field: 0x%x, nominal latency: %d us\n",
+		mem_resource->minimum, ioread32(&pcch_hdr->signature),
+		ioread16(&pcch_hdr->length), ioread8(&pcch_hdr->major),
+		ioread8(&pcch_hdr->minor), ioread32(&pcch_hdr->features),
+		ioread16(&pcch_hdr->command), ioread16(&pcch_hdr->status),
+		ioread32(&pcch_hdr->latency));
+
+	dprintk("probe: min time between commands: %d us,"
+		" max time between commands: %d us,"
+		" nominal CPU frequency: %d MHz,"
+		" minimum CPU frequency: %d MHz,"
+		" minimum CPU frequency without throttling: %d MHz\n",
+		ioread32(&pcch_hdr->minimum_time),
+		ioread32(&pcch_hdr->maximum_time),
+		ioread32(&pcch_hdr->nominal),
+		ioread32(&pcch_hdr->throttled_frequency),
+		ioread32(&pcch_hdr->minimum_frequency));
+
+	member = &out_obj->package.elements[1];
+	if (member->type != ACPI_TYPE_BUFFER) {
+		ret = -ENODEV;
+		goto pcch_free;
+	}
+
+	reg_resource = (struct pcc_register_resource *)member->buffer.pointer;
+
+	doorbell.space_id = reg_resource->space_id;
+	doorbell.bit_width = reg_resource->bit_width;
+	doorbell.bit_offset = reg_resource->bit_offset;
+	doorbell.access_width = 64;
+	doorbell.address = reg_resource->address;
+
+	dprintk("probe: doorbell: space_id is %d, bit_width is %d, "
+		"bit_offset is %d, access_width is %d, address is 0x%llx\n",
+		doorbell.space_id, doorbell.bit_width, doorbell.bit_offset,
+		doorbell.access_width, reg_resource->address);
+
+	member = &out_obj->package.elements[2];
+	if (member->type != ACPI_TYPE_INTEGER) {
+		ret = -ENODEV;
+		goto pcch_free;
+	}
+
+	doorbell_preserve = member->integer.value;
+
+	member = &out_obj->package.elements[3];
+	if (member->type != ACPI_TYPE_INTEGER) {
+		ret = -ENODEV;
+		goto pcch_free;
+	}
+
+	doorbell_write = member->integer.value;
+
+	dprintk("probe: doorbell_preserve: 0x%llx,"
+		" doorbell_write: 0x%llx\n",
+		doorbell_preserve, doorbell_write);
+
+	pcc_cpu_info = alloc_percpu(struct pcc_cpu);
+	if (!pcc_cpu_info) {
+		ret = -ENOMEM;
+		goto pcch_free;
+	}
+
+	printk(KERN_DEBUG "pcc-cpufreq: (v%s) driver loaded with frequency"
+	       " limits: %d MHz, %d MHz\n", PCC_VERSION,
+	       ioread32(&pcch_hdr->minimum_frequency),
+	       ioread32(&pcch_hdr->nominal));
+	kfree(output.pointer);
+	return ret;
+pcch_free:
+	pcc_clear_mapping();
+out_free:
+	kfree(output.pointer);
+	return ret;
+}
+
+static int pcc_cpufreq_cpu_init(struct cpufreq_policy *policy)
+{
+	unsigned int cpu = policy->cpu;
+	unsigned int result = 0;
+
+	if (!pcch_virt_addr) {
+		result = -1;
+		goto pcch_null;
+	}
+
+	result = pcc_get_offset(cpu);
+	if (result) {
+		dprintk("init: PCCP evaluation failed\n");
+		goto free;
+	}
+
+	policy->max = policy->cpuinfo.max_freq =
+		ioread32(&pcch_hdr->nominal) * 1000;
+	policy->min = policy->cpuinfo.min_freq =
+		ioread32(&pcch_hdr->minimum_frequency) * 1000;
+	policy->cur = pcc_get_freq(cpu);
+
+	dprintk("init: policy->max is %d, policy->min is %d\n",
+		policy->max, policy->min);
+
+	return 0;
+free:
+	pcc_clear_mapping();
+	free_percpu(pcc_cpu_info);
+pcch_null:
+	return result;
+}
+
+static int pcc_cpufreq_cpu_exit(struct cpufreq_policy *policy)
+{
+	return 0;
+}
+
+static struct cpufreq_driver pcc_cpufreq_driver = {
+	.flags = CPUFREQ_CONST_LOOPS,
+	.get = pcc_get_freq,
+	.verify = pcc_cpufreq_verify,
+	.target = pcc_cpufreq_target,
+	.init = pcc_cpufreq_cpu_init,
+	.exit = pcc_cpufreq_cpu_exit,
+	.name = "pcc-cpufreq",
+	.owner = THIS_MODULE,
+};
+
+static int __init pcc_cpufreq_init(void)
+{
+	int ret;
+
+	if (acpi_disabled)
+		return 0;
+
+	ret = pcc_cpufreq_probe();
+	if (ret) {
+		dprintk("pcc_cpufreq_init: PCCH evaluation failed\n");
+		return ret;
+	}
+
+	ret = cpufreq_register_driver(&pcc_cpufreq_driver);
+
+	return ret;
+}
+
+static void __exit pcc_cpufreq_exit(void)
+{
+	cpufreq_unregister_driver(&pcc_cpufreq_driver);
+
+	pcc_clear_mapping();
+
+	free_percpu(pcc_cpu_info);
+}
+
+MODULE_AUTHOR("Matthew Garrett, Naga Chumbalkar");
+MODULE_VERSION(PCC_VERSION);
+MODULE_DESCRIPTION("Processor Clocking Control interface driver");
+MODULE_LICENSE("GPL");
+
+late_initcall(pcc_cpufreq_init);
+module_exit(pcc_cpufreq_exit);

drivers/acpi/processor_core.c

···
 #endif
 
 DEFINE_PER_CPU(struct acpi_processor *, processors);
+EXPORT_PER_CPU_SYMBOL(processors);
+
 struct acpi_processor_errata errata __read_mostly;
 static int set_no_mwait(const struct dmi_system_id *id)
 {