Merge bk://linux-dj.bkbits.net/cpufreq

into ppc970.osdl.org:/home/torvalds/v2.6/linux

arch/arm/Kconfig

@@ -336,9 +336,10 @@ config ZBOOT_ROM_BSS
 	  while the decompressor is running.  Unless you have special requirements,
 	  you should not change this value.
 
+if (ARCH_SA1100 || ARCH_INTEGRATOR)
+
 config CPU_FREQ
-	bool "Support CPU clock change (EXPERIMENTAL)"
-	depends on (ARCH_SA1100 || ARCH_INTEGRATOR) && EXPERIMENTAL
+	bool "Support CPU clock change"
 	help
 	  CPU clock scaling allows you to change the clock speed of the
 	  running CPU on the fly. This is a nice method to save battery power,
@@ -348,20 +349,17 @@ config CPU_FREQ
 	  written) to implement the policy. If you don't understand what this
 	  is all about, it's safe to say 'N'.
 
-
-# CPUfreq on SA11x0 is special -- it _needs_ the userspace governor
+source "drivers/cpufreq/Kconfig"
 
 config CPU_FREQ_SA1100
 	bool
 	depends on CPU_FREQ && SA1100_LART
 	default y
-	select CPU_FREQ_24_API if SYSCTL
 
 config CPU_FREQ_SA1110
 	bool
 	depends on CPU_FREQ && (SA1100_ASSABET || SA1100_CERF || SA1100_PT_SYSTEM3)
 	default y
-	select CPU_FREQ_24_API if SYSCTL
 
 config CPU_FREQ_INTEGRATOR
 	tristate "CPUfreq driver for ARM Integrator CPUs"
@@ -374,10 +372,6 @@ config CPU_FREQ_INTEGRATOR
 
 	  If in doubt, say Y.
 
-if (CPU_FREQ_INTEGRATOR) || (CPU_FREQ_SA1110) || (CPU_FREQ_SA1100)
-
-source "drivers/cpufreq/Kconfig"
-
 endif
 
 source "drivers/pci/Kconfig"

arch/i386/kernel/cpu/cpufreq/Kconfig

@@ -42,16 +42,6 @@ config X86_ACPI_CPUFREQ
 
 	  If in doubt, say N.
 
-config X86_ACPI_CPUFREQ_PROC_INTF
-        bool "/proc/acpi/processor/../performance interface (deprecated)"
-	depends on X86_ACPI_CPUFREQ && PROC_FS
-	help
-	  This enables the deprecated /proc/acpi/processor/../performance 
-	  interface. While it is helpful for debugging, the generic,
-	  cross-architecture cpufreq interfaces should be used.
-
-	  If in doubt, say N.
-
 config ELAN_CPUFREQ
 	tristate "AMD Elan"
 	depends on CPU_FREQ_TABLE && X86_ELAN
@@ -95,7 +85,7 @@ config X86_POWERNOW_K7_ACPI
 
 config X86_POWERNOW_K8
 	tristate "AMD Opteron/Athlon64 PowerNow!"
-	depends on CPU_FREQ && EXPERIMENTAL
+	depends on CPU_FREQ_TABLE && EXPERIMENTAL
 	help
 	  This adds the CPUFreq driver for mobile AMD Opteron/Athlon64 processors.
 
@@ -122,29 +112,40 @@ config X86_GX_SUSPMOD
 config X86_SPEEDSTEP_CENTRINO
 	tristate "Intel Enhanced SpeedStep"
 	depends on CPU_FREQ_TABLE
+	select X86_SPEEDSTEP_CENTRINO_TABLE if (!X86_SPEEDSTEP_CENTRINO_ACPI)
 	help
 	  This adds the CPUFreq driver for Enhanced SpeedStep enabled
-	  mobile CPUs.  This means Intel Pentium M (Centrino) CPUs.
+	  mobile CPUs.  This means Intel Pentium M (Centrino) CPUs. However,
+	  you also need to say Y to "Use ACPI tables to decode..." below
+	  [which might imply enabling ACPI] if you want to use this driver
+	  on non-Banias CPUs.
 	  
 	  For details, take a look at <file:Documentation/cpu-freq/>.
 	  
 	  If in doubt, say N.
 
-config X86_SPEEDSTEP_CENTRINO_TABLE
-	bool
-	depends on X86_SPEEDSTEP_CENTRINO
-	default y
-
 config X86_SPEEDSTEP_CENTRINO_ACPI
-	bool "Use ACPI tables to decode valid frequency/voltage pairs (EXPERIMENTAL)"
-	depends on EXPERIMENTAL
+	bool "Use ACPI tables to decode valid frequency/voltage pairs"
+	depends on X86_SPEEDSTEP_CENTRINO
 	depends on ((X86_SPEEDSTEP_CENTRINO = "m" && ACPI_PROCESSOR) || (X86_SPEEDSTEP_CENTRINO = "y" && ACPI_PROCESSOR = "y"))
+	default y
 	help
 	  Use primarily the information provided in the BIOS ACPI tables
-	  to determine valid CPU frequency and voltage pairings.
+	  to determine valid CPU frequency and voltage pairings. It is
+	  required for the driver to work on non-Banias CPUs.
 
 	  If in doubt, say Y.
 
+config X86_SPEEDSTEP_CENTRINO_TABLE
+	bool "Built-in tables for Banias CPUs"
+	depends on X86_SPEEDSTEP_CENTRINO
+	default y
+	help
+	  Use built-in tables for Banias CPUs if ACPI encoding
+	  is not available.
+
+	  If in doubt, say N.
+
 config X86_SPEEDSTEP_ICH
 	tristate "Intel Speedstep on ICH-M chipsets (ioport interface)"
 	depends on CPU_FREQ_TABLE
@@ -181,21 +182,6 @@ config X86_P4_CLOCKMOD
 
 	  If in doubt, say N.
 
-config X86_SPEEDSTEP_LIB
-	tristate
-	depends on (X86_SPEEDSTEP_ICH || X86_SPEEDSTEP_SMI || X86_P4_CLOCKMOD)
-	default (X86_SPEEDSTEP_ICH || X86_SPEEDSTEP_SMI || X86_P4_CLOCKMOD)
-
-config X86_SPEEDSTEP_RELAXED_CAP_CHECK
-	bool "Relaxed speedstep capability checks"
-	depends on (X86_SPEEDSTEP_SMI || X86_SPEEDSTEP_ICH)
-	help
-	  Don't perform all checks for a speedstep capable system which would 
-	  normally be done. Some ancient or strange systems, though speedstep 
-	  capable, don't always indicate that they are speedstep capable. This 
-	  option let's the probing code bypass some of those checks if the
-	  parameter "relaxed_check=1" is passed to the module.
-
 config X86_LONGRUN
 	tristate "Transmeta LongRun"
 	depends on CPU_FREQ
@@ -219,4 +205,34 @@ config X86_LONGHAUL
 
 	  If in doubt, say N.
 
+comment "shared options"
+	depends on CPU_FREQ
+
+config X86_ACPI_CPUFREQ_PROC_INTF
+        bool "/proc/acpi/processor/../performance interface (deprecated)"
+	depends on PROC_FS
+	depends on X86_ACPI_CPUFREQ || X86_SPEEDSTEP_CENTRINO_ACPI || X86_POWERNOW_K7_ACPI || X86_POWERNOW_K8_ACPI
+	help
+	  This enables the deprecated /proc/acpi/processor/../performance 
+	  interface. While it is helpful for debugging, the generic,
+	  cross-architecture cpufreq interfaces should be used.
+
+	  If in doubt, say N.
+
+config X86_SPEEDSTEP_LIB
+	tristate
+	depends on (X86_SPEEDSTEP_ICH || X86_SPEEDSTEP_SMI || X86_P4_CLOCKMOD)
+	default (X86_SPEEDSTEP_ICH || X86_SPEEDSTEP_SMI || X86_P4_CLOCKMOD)
+
+config X86_SPEEDSTEP_RELAXED_CAP_CHECK
+	bool "Relaxed speedstep capability checks"
+	depends on (X86_SPEEDSTEP_SMI || X86_SPEEDSTEP_ICH)
+	help
+	  Don't perform all checks for a speedstep capable system which would 
+	  normally be done. Some ancient or strange systems, though speedstep 
+	  capable, don't always indicate that they are speedstep capable. This 
+	  option lets the probing code bypass some of those checks if the
+	  parameter "relaxed_check=1" is passed to the module.
+
+
 endmenu

arch/i386/kernel/cpu/cpufreq/p4-clockmod.c

@@ -110,7 +110,7 @@ static int cpufreq_p4_target(struct cpufreq_policy *policy,
 {
 	unsigned int    newstate = DC_RESV;
 	struct cpufreq_freqs freqs;
-	cpumask_t cpus_allowed, affected_cpu_map;
+	cpumask_t cpus_allowed;
 	int i;
 
 	if (cpufreq_frequency_table_target(policy, &p4clockmod_table[0], target_freq, relation, &newstate))
@@ -122,18 +122,8 @@ static int cpufreq_p4_target(struct cpufreq_policy *policy,
 	if (freqs.new == freqs.old)
 		return 0;
 
-	/* switch to physical CPU where state is to be changed*/
-	cpus_allowed = current->cpus_allowed;
-
-	/* only run on CPU to be set, or on its sibling */
-#ifdef CONFIG_SMP
-	affected_cpu_map = cpu_sibling_map[policy->cpu];
-#else
-	affected_cpu_map = cpumask_of_cpu(policy->cpu);
-#endif
-
 	/* notifiers */
-	for_each_cpu_mask(i, affected_cpu_map) {
+	for_each_cpu_mask(i, policy->cpus) {
 		freqs.cpu = i;
 		cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE);
 	}
@@ -141,7 +131,9 @@ static int cpufreq_p4_target(struct cpufreq_policy *policy,
 	/* run on each logical CPU, see section 13.15.3 of IA32 Intel Architecture Software
 	 * Developer's Manual, Volume 3 
 	 */
-	for_each_cpu_mask(i, affected_cpu_map) {
+	cpus_allowed = current->cpus_allowed;
+
+	for_each_cpu_mask(i, policy->cpus) {
 		cpumask_t this_cpu = cpumask_of_cpu(i);
 
 		set_cpus_allowed(current, this_cpu);
@@ -152,7 +144,7 @@ static int cpufreq_p4_target(struct cpufreq_policy *policy,
 	set_cpus_allowed(current, cpus_allowed);
 
 	/* notifiers */
-	for_each_cpu_mask(i, affected_cpu_map) {
+	for_each_cpu_mask(i, policy->cpus) {
 		freqs.cpu = i;
 		cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE);
 	}
@@ -219,6 +211,10 @@ static int cpufreq_p4_cpu_init(struct cpufreq_policy *policy)
 	int cpuid = 0;
 	unsigned int i;
 
+#ifdef CONFIG_SMP
+	policy->cpus = cpu_sibling_map[policy->cpu];
+#endif
+
 	/* Errata workaround */
 	cpuid = (c->x86 << 8) | (c->x86_model << 4) | c->x86_mask;
 	switch (cpuid) {
@@ -260,14 +256,13 @@ static int cpufreq_p4_cpu_exit(struct cpufreq_policy *policy)
 
 static unsigned int cpufreq_p4_get(unsigned int cpu)
 {
-	cpumask_t cpus_allowed, affected_cpu_map;
+	cpumask_t cpus_allowed;
 	u32 l, h;
 
 	cpus_allowed = current->cpus_allowed;
-        affected_cpu_map = cpumask_of_cpu(cpu);
 
-	set_cpus_allowed(current, affected_cpu_map);
-        BUG_ON(!cpu_isset(smp_processor_id(), affected_cpu_map));
+	set_cpus_allowed(current, cpumask_of_cpu(cpu));
+	BUG_ON(smp_processor_id() != cpu);
 
 	rdmsr(MSR_IA32_THERM_CONTROL, l, h);
 
@@ -335,4 +330,3 @@ MODULE_LICENSE ("GPL");
 
 late_initcall(cpufreq_p4_init);
 module_exit(cpufreq_p4_exit);
-

arch/i386/kernel/cpu/cpufreq/powernow-k8.c

@@ -39,7 +39,7 @@
 
 #define PFX "powernow-k8: "
 #define BFX PFX "BIOS error: "
-#define VERSION "version 1.00.09b"
+#define VERSION "version 1.00.09e"
 #include "powernow-k8.h"
 
 /* serialize freq changes  */
@@ -596,6 +596,8 @@ static int find_psb_table(struct powernow_k8_data *data)
 	unsigned int i;
 	u32 mvs;
 	u8 maxvid;
+	u32 cpst = 0;
+	u32 thiscpuid;
 
 	for (i = 0xc0000; i < 0xffff0; i += 0x10) {
 		/* Scan BIOS looking for the signature. */
@@ -634,7 +636,14 @@ static int find_psb_table(struct powernow_k8_data *data)
 		dprintk(KERN_INFO PFX "maximum voltage step: %d - 0x%x\n", mvs, data->vidmvs);
 
 		dprintk(KERN_DEBUG PFX "numpst: 0x%x\n", psb->numpst);
-		if (psb->numpst != 1) {
+		cpst = psb->numpst;
+		if ((psb->cpuid == 0x00000fc0) || (psb->cpuid == 0x00000fe0) ){
+			thiscpuid = cpuid_eax(CPUID_PROCESSOR_SIGNATURE);
+			if ((thiscpuid == 0x00000fc0) || (thiscpuid == 0x00000fe0) ) {
+				cpst = 1;
+			}
+		}
+		if (cpst != 1) {
 			printk(KERN_ERR BFX "numpst must be 1\n");
 			return -ENODEV;
 		}

arch/i386/kernel/cpu/cpufreq/speedstep-centrino.c

@@ -77,7 +77,7 @@ static int centrino_verify_cpu_id(const struct cpuinfo_x86 *c, const struct cpu_
 
 /* Operating points for current CPU */
 static struct cpu_model *centrino_model;
-static int centrino_cpu;
+static const struct cpu_id *centrino_cpu;
 
 #ifdef CONFIG_X86_SPEEDSTEP_CENTRINO_TABLE
 
@@ -289,7 +289,9 @@ static unsigned extract_clock(unsigned msr)
 	 * for centrino, as some DSDTs are buggy.
 	 * Ideally, this can be done using the acpi_data structure.
 	 */
-	if (centrino_cpu) {
+	if ((centrino_cpu == &cpu_ids[CPU_BANIAS]) ||
+	    (centrino_cpu == &cpu_ids[CPU_DOTHAN_A1]) ||
+	    (centrino_cpu == &cpu_ids[CPU_DOTHAN_B0])) {
 		msr = (msr >> 8) & 0xff;
 		return msr * 100000;
 	}
@@ -457,7 +459,7 @@ static int centrino_cpu_init(struct cpufreq_policy *policy)
 			break;
 
 	if (i != N_IDS)
-		centrino_cpu = 1;
+		centrino_cpu = &cpu_ids[i];
 
 	if (centrino_cpu_init_acpi(policy)) {
 		if (policy->cpu != 0)
@@ -567,7 +569,7 @@ static int centrino_target (struct cpufreq_policy *policy,
 	 * Make sure we are running on the CPU that wants to change frequency
 	 */
 	saved_mask = current->cpus_allowed;
-	set_cpus_allowed(current, cpumask_of_cpu(policy->cpu));
+	set_cpus_allowed(current, policy->cpus);
 	if (smp_processor_id() != policy->cpu) {
 		return(-EAGAIN);
 	}

arch/i386/kernel/cpu/cpufreq/speedstep-ich.c

@@ -223,24 +223,23 @@ static unsigned int speedstep_detect_chipset (void)
 	return 0;
 }
 
-static unsigned int speedstep_get(unsigned int cpu)
+static unsigned int _speedstep_get(cpumask_t cpus)
 {
 	unsigned int speed;
-	cpumask_t cpus_allowed,affected_cpu_map;
+	cpumask_t cpus_allowed;
 
-	/* only run on CPU to be set, or on its sibling */
 	cpus_allowed = current->cpus_allowed;
-#ifdef CONFIG_SMP
-	affected_cpu_map = cpu_sibling_map[cpu];
-#else
-	affected_cpu_map = cpumask_of_cpu(cpu);
-#endif
-	set_cpus_allowed(current, affected_cpu_map);
-	speed=speedstep_get_processor_frequency(speedstep_processor);
+	set_cpus_allowed(current, cpus);
+	speed = speedstep_get_processor_frequency(speedstep_processor);
 	set_cpus_allowed(current, cpus_allowed);
 	return speed;
 }
 
+static unsigned int speedstep_get(unsigned int cpu)
+{
+	return _speedstep_get(cpumask_of_cpu(cpu));
+}
+
 /**
  * speedstep_target - set a new CPUFreq policy
  * @policy: new policy
@@ -255,13 +254,13 @@ static int speedstep_target (struct cpufreq_policy *policy,
 {
 	unsigned int newstate = 0;
 	struct cpufreq_freqs freqs;
-	cpumask_t cpus_allowed, affected_cpu_map;
+	cpumask_t cpus_allowed;
 	int i;
 
 	if (cpufreq_frequency_table_target(policy, &speedstep_freqs[0], target_freq, relation, &newstate))
 		return -EINVAL;
 
-	freqs.old = speedstep_get(policy->cpu);
+	freqs.old = _speedstep_get(policy->cpus);
 	freqs.new = speedstep_freqs[newstate].frequency;
 	freqs.cpu = policy->cpu;
 
@@ -271,27 +270,20 @@ static int speedstep_target (struct cpufreq_policy *policy,
 
 	cpus_allowed = current->cpus_allowed;
 
-	/* only run on CPU to be set, or on its sibling */
-#ifdef CONFIG_SMP
-	affected_cpu_map = cpu_sibling_map[policy->cpu];
-#else
-	affected_cpu_map = cpumask_of_cpu(policy->cpu);
-#endif
-
-	for_each_cpu_mask(i, affected_cpu_map) {
+	for_each_cpu_mask(i, policy->cpus) {
 		freqs.cpu = i;
 		cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE);
 	}
 
 	/* switch to physical CPU where state is to be changed */
-	set_cpus_allowed(current, affected_cpu_map);
+	set_cpus_allowed(current, policy->cpus);
 
 	speedstep_set_state(newstate);
 
 	/* allow to be run on all CPUs */
 	set_cpus_allowed(current, cpus_allowed);
 
-	for_each_cpu_mask(i, affected_cpu_map) {
+	for_each_cpu_mask(i, policy->cpus) {
 		freqs.cpu = i;
 		cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE);
 	}
@@ -317,21 +309,15 @@ static int speedstep_cpu_init(struct cpufreq_policy *policy)
 {
 	int result = 0;
 	unsigned int speed;
-	cpumask_t cpus_allowed,affected_cpu_map;
-
-
-	/* capability check */
-	if (policy->cpu != 0) /* FIXME: better support for SMT in cpufreq core. Up until then, it's better to register only one CPU */
-		return -ENODEV;
+	cpumask_t cpus_allowed;
 
 	/* only run on CPU to be set, or on its sibling */
-	cpus_allowed = current->cpus_allowed;
 #ifdef CONFIG_SMP
-	affected_cpu_map = cpu_sibling_map[policy->cpu];
-#else
-	affected_cpu_map = cpumask_of_cpu(policy->cpu);
+	policy->cpus = cpu_sibling_map[policy->cpu];
 #endif
-	set_cpus_allowed(current, affected_cpu_map);
+
+	cpus_allowed = current->cpus_allowed;
+	set_cpus_allowed(current, policy->cpus);
 
 	/* detect low and high frequency */
 	result = speedstep_get_freqs(speedstep_processor,
@@ -343,7 +329,7 @@ static int speedstep_cpu_init(struct cpufreq_policy *policy)
 		return result;
 
 	/* get current speed setting */
-	speed = speedstep_get(policy->cpu);
+	speed = _speedstep_get(policy->cpus);
 	if (!speed)
 		return -EIO;
 

arch/i386/kernel/cpu/cpufreq/speedstep-smi.c

@@ -260,12 +260,6 @@ static int speedstep_cpu_init(struct cpufreq_policy *policy)
 	if (result) {
 		/* fall back to speedstep_lib.c dection mechanism: try both states out */
 		dprintk(KERN_INFO PFX "could not detect low and high frequencies by SMI call.\n");
-		if (!speedstep_processor)
-			speedstep_processor = speedstep_detect_processor();
-
-		if (!speedstep_processor)
-			return -ENODEV;
-
 		result = speedstep_get_freqs(speedstep_processor,
 				&speedstep_freqs[SPEEDSTEP_LOW].frequency,
 				&speedstep_freqs[SPEEDSTEP_HIGH].frequency,
@@ -310,10 +304,6 @@ static unsigned int speedstep_get(unsigned int cpu)
 {
 	if (cpu)
 		return -ENODEV;
-	if (!speedstep_processor)
-		speedstep_processor = speedstep_detect_processor();
-	if (!speedstep_processor)
-		return 0;
 	return speedstep_get_processor_frequency(speedstep_processor);
 }
 
@@ -354,10 +344,19 @@ static struct cpufreq_driver speedstep_driver = {
  */
 static int __init speedstep_init(void)
 {
-    struct cpuinfo_x86 *c = cpu_data;
+	speedstep_processor = speedstep_detect_processor();
+
+	switch (speedstep_processor) {
+	case SPEEDSTEP_PROCESSOR_PIII_T:
+	case SPEEDSTEP_PROCESSOR_PIII_C:
+	case SPEEDSTEP_PROCESSOR_PIII_C_EARLY:
+		break;
+	default:
+		speedstep_processor = 0;
+	}
 
-    if (c->x86_vendor != X86_VENDOR_INTEL) {
-		printk (KERN_INFO PFX "No Intel CPU detected.\n");
+	if (!speedstep_processor) {
+		dprintk (KERN_INFO PFX "No supported Intel CPU detected.\n");
 		return -ENODEV;
 	}
 

arch/x86_64/kernel/cpufreq/Kconfig

@@ -48,7 +48,7 @@ config X86_POWERNOW_K8_ACPI
 
 config X86_SPEEDSTEP_CENTRINO
 	tristate "Intel Enhanced SpeedStep"
-	depends on CPU_FREQ_TABLE
+	depends on CPU_FREQ_TABLE && ACPI_PROCESSOR
 	help
 	  This adds the CPUFreq driver for Enhanced SpeedStep enabled
 	  mobile CPUs.  This means Intel Pentium M (Centrino) CPUs
@@ -58,21 +58,11 @@ config X86_SPEEDSTEP_CENTRINO
 
 	  If in doubt, say N.
 
-config X86_SPEEDSTEP_CENTRINO_TABLE
+config X86_SPEEDSTEP_CENTRINO_ACPI
 	bool
 	depends on X86_SPEEDSTEP_CENTRINO
 	default y
 
-config X86_SPEEDSTEP_CENTRINO_ACPI
-	bool "Use ACPI tables to decode valid frequency/voltage pairs (EXPERIMENTAL)"
-	depends on EXPERIMENTAL
-	depends on ((X86_SPEEDSTEP_CENTRINO = "m" && ACPI_PROCESSOR) || (X86_SPEEDSTEP_CENTRINO = "y" && ACPI_PROCESSOR = "y"))
-	help
-	  Use primarily the information provided in the BIOS ACPI tables
-	  to determine valid CPU frequency and voltage pairings.
-
-	  If in doubt, say Y.
-
 config X86_ACPI_CPUFREQ
 	tristate "ACPI Processor P-States driver"
 	depends on CPU_FREQ_TABLE && ACPI_PROCESSOR
@@ -84,9 +74,13 @@ config X86_ACPI_CPUFREQ
 
 	  If in doubt, say N.
 
+comment "shared options"
+	depends on CPU_FREQ
+
 config X86_ACPI_CPUFREQ_PROC_INTF
         bool "/proc/acpi/processor/../performance interface (deprecated)"
-	depends on X86_ACPI_CPUFREQ && PROC_FS
+	depends on PROC_FS
+	depends on X86_ACPI_CPUFREQ || X86_SPEEDSTEP_CENTRINO_ACPI || X86_POWERNOW_K8_ACPI
 	help
 	  This enables the deprecated /proc/acpi/processor/../performance
 	  interface. While it is helpful for debugging, the generic,

drivers/cpufreq/Kconfig

@@ -33,7 +33,7 @@ config CPU_FREQ_DEFAULT_GOV_USERSPACE
 	help
 	  Use the CPUFreq governor 'userspace' as default. This allows
 	  you to set the CPU frequency manually or when an userspace 
-	  programm shall be able to set the CPU dynamically without having
+	  program shall be able to set the CPU dynamically without having
 	  to enable the userspace governor manually.
 
 endchoice
@@ -42,7 +42,7 @@ config CPU_FREQ_GOV_PERFORMANCE
        tristate "'performance' governor"
        depends on CPU_FREQ
        help
-	  This cpufreq governors set the frequency statically to the
+	  This cpufreq governor sets the frequency statically to the
 	  highest available CPU frequency.
 
 	  If in doubt, say Y.
@@ -51,7 +51,7 @@ config CPU_FREQ_GOV_POWERSAVE
        tristate "'powersave' governor"
        depends on CPU_FREQ
        help
-	  Theis cpufreq governors set the frequency statically to the
+	  This cpufreq governor sets the frequency statically to the
 	  lowest available CPU frequency.
 
 	  If in doubt, say Y.
@@ -61,7 +61,7 @@ config CPU_FREQ_GOV_USERSPACE
        depends on CPU_FREQ 
        help
 	  Enable this cpufreq governor when you either want to set the
-	  CPU frequency manually or when an userspace programm shall
+	  CPU frequency manually or when an userspace program shall
 	  be able to set the CPU dynamically, like on LART 
 	  <http://www.lart.tudelft.nl/>
 
@@ -69,6 +69,21 @@ config CPU_FREQ_GOV_USERSPACE
 
 	  If in doubt, say Y.
 
+config CPU_FREQ_24_API
+	bool "/proc/sys/cpu/ interface (2.4. / OLD)"
+	depends on CPU_FREQ_GOV_USERSPACE
+	depends on SYSCTL
+	help
+	  This enables the /proc/sys/cpu/ sysctl interface for controlling
+	  the CPUFreq,"userspace" governor. This is the same interface
+	  as known from the 2.4.-kernel patches for CPUFreq, and offers
+	  the same functionality as long as "userspace" is the
+	  selected governor for the specified CPU.
+	
+	  For details, take a look at <file:Documentation/cpu-freq/>.
+
+	  If in doubt, say N.
+
 config CPU_FREQ_GOV_ONDEMAND
 	tristate "'ondemand' cpufreq policy governor"
 	depends on CPU_FREQ
@@ -83,17 +98,3 @@ config CPU_FREQ_GOV_ONDEMAND
 	  For details, take a look at linux/Documentation/cpu-freq.
 
 	  If in doubt, say N.
-
-config CPU_FREQ_24_API
-	bool "/proc/sys/cpu/ interface (2.4. / OLD)"
-	depends on CPU_FREQ && SYSCTL && CPU_FREQ_GOV_USERSPACE
-	help
-	  This enables the /proc/sys/cpu/ sysctl interface for controlling
-	  the CPUFreq,"userspace" governor. This is the same interface
-	  as known from the 2.4.-kernel patches for CPUFreq, and offers
-	  the same functionality as long as "userspace" is the
-	  selected governor for the specified CPU.
-	
-	  For details, take a look at <file:Documentation/cpu-freq/>.
-
-	  If in doubt, say N.

drivers/cpufreq/cpufreq.c

@@ -33,6 +33,14 @@ static struct cpufreq_driver   	*cpufreq_driver;
 static struct cpufreq_policy	*cpufreq_cpu_data[NR_CPUS];
 static spinlock_t		cpufreq_driver_lock = SPIN_LOCK_UNLOCKED;
 
+
+/* we keep a copy of all ->add'ed CPU's struct sys_device here;
+ * as it is only accessed in ->add and ->remove, no lock or reference
+ * count is necessary.
+ */
+static struct sys_device	*cpu_sys_devices[NR_CPUS];
+
+
 /* internal prototypes */
 static int __cpufreq_governor(struct cpufreq_policy *policy, unsigned int event);
 static void handle_update(void *data);
@@ -154,6 +162,7 @@ void cpufreq_notify_transition(struct cpufreq_freqs *freqs, unsigned int state)
 		 */
 		if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
 			if ((likely(cpufreq_cpu_data[freqs->cpu])) &&
+			    (likely(cpufreq_cpu_data[freqs->cpu]->cpu == freqs->cpu)) &&
 			    (likely(cpufreq_cpu_data[freqs->cpu]->cur)) &&
 			    (unlikely(freqs->old != cpufreq_cpu_data[freqs->cpu]->cur)))
 			{
@@ -168,7 +177,8 @@ void cpufreq_notify_transition(struct cpufreq_freqs *freqs, unsigned int state)
 	case CPUFREQ_POSTCHANGE:
 		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
 		notifier_call_chain(&cpufreq_transition_notifier_list, CPUFREQ_POSTCHANGE, freqs);
-		if (likely(cpufreq_cpu_data[freqs->cpu]))
+		if ((likely(cpufreq_cpu_data[freqs->cpu])) && 
+		    (likely(cpufreq_cpu_data[freqs->cpu]->cpu == freqs->cpu)))
 			cpufreq_cpu_data[freqs->cpu]->cur = freqs->new;
 		break;
 	}
@@ -456,6 +466,18 @@ static int cpufreq_add_dev (struct sys_device * sys_dev)
 	struct cpufreq_policy *policy;
 	struct freq_attr **drv_attr;
 	unsigned long flags;
+	unsigned int j;
+
+#ifdef CONFIG_SMP
+	/* check whether a different CPU already registered this
+	 * CPU because it is in the same boat. */
+	policy = cpufreq_cpu_get(cpu);
+	if (unlikely(policy)) {
+		cpu_sys_devices[cpu] = sys_dev;
+		sysfs_create_link(&sys_dev->kobj, &policy->kobj, "cpufreq");
+		return 0;
+	}
+#endif
 
 	if (!try_module_get(cpufreq_driver->owner))
 		return -EINVAL;
@@ -468,6 +490,8 @@ static int cpufreq_add_dev (struct sys_device * sys_dev)
 	memset(policy, 0, sizeof(struct cpufreq_policy));
 
 	policy->cpu = cpu;
+	policy->cpus = cpumask_of_cpu(cpu);
+
 	init_MUTEX_LOCKED(&policy->lock);
 	init_completion(&policy->kobj_unregister);
 	INIT_WORK(&policy->update, handle_update, (void *)(long)cpu);
@@ -502,7 +526,8 @@ static int cpufreq_add_dev (struct sys_device * sys_dev)
 		sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
 
 	spin_lock_irqsave(&cpufreq_driver_lock, flags);
-	cpufreq_cpu_data[cpu] = policy;
+	for_each_cpu_mask(j, policy->cpus)
+		cpufreq_cpu_data[j] = policy;
 	spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 	policy->governor = NULL; /* to assure that the starting sequence is
 				  * run in cpufreq_set_policy */
@@ -515,12 +540,14 @@ static int cpufreq_add_dev (struct sys_device * sys_dev)
 		goto err_out_unregister;
 
 	module_put(cpufreq_driver->owner);
+	cpu_sys_devices[cpu] = sys_dev;
 	return 0;
 
 
 err_out_unregister:
 	spin_lock_irqsave(&cpufreq_driver_lock, flags);
-	cpufreq_cpu_data[cpu] = NULL;
+	for_each_cpu_mask(j, policy->cpus)
+		cpufreq_cpu_data[j] = NULL;
 	spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
 	kobject_unregister(&policy->kobj);
@@ -545,19 +572,69 @@ static int cpufreq_remove_dev (struct sys_device * sys_dev)
 	unsigned int cpu = sys_dev->id;
 	unsigned long flags;
 	struct cpufreq_policy *data;
+#ifdef CONFIG_SMP
+	unsigned int j;
+#endif
 
 	spin_lock_irqsave(&cpufreq_driver_lock, flags);
 	data = cpufreq_cpu_data[cpu];
 
 	if (!data) {
 		spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
+		cpu_sys_devices[cpu] = NULL;
 		return -EINVAL;
 	}
 	cpufreq_cpu_data[cpu] = NULL;
+
+
+#ifdef CONFIG_SMP
+	/* if this isn't the CPU which is the parent of the kobj, we
+	 * only need to unlink, put and exit 
+	 */
+	if (unlikely(cpu != data->cpu)) {
 		spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
+		sysfs_remove_link(&sys_dev->kobj, "cpufreq");
+		cpu_sys_devices[cpu] = NULL;
+		cpufreq_cpu_put(data);
+		return 0;
+	}
+#endif
 
-	if (!kobject_get(&data->kobj))
+	cpu_sys_devices[cpu] = NULL;
+
+	if (!kobject_get(&data->kobj)) {
+		spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
  		return -EFAULT;
+	}
+
+#ifdef CONFIG_SMP
+	/* if we have other CPUs still registered, we need to unlink them,
+	 * or else wait_for_completion below will lock up. Clean the
+	 * cpufreq_cpu_data[] while holding the lock, and remove the sysfs
+	 * links afterwards.
+	 */
+	if (unlikely(cpus_weight(data->cpus) > 1)) {
+		for_each_cpu_mask(j, data->cpus) {
+			if (j == cpu)
+				continue;
+			cpufreq_cpu_data[j] = NULL;
+		}
+	}
+
+	spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
+
+	if (unlikely(cpus_weight(data->cpus) > 1)) {
+		for_each_cpu_mask(j, data->cpus) {
+			if (j == cpu)
+				continue;
+			
+			sysfs_remove_link(&cpu_sys_devices[j]->kobj, "cpufreq");
+			cpufreq_cpu_put(data);
+		}
+	}
+#else
+	spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
+#endif
 
 	if (cpufreq_driver->target)
 		__cpufreq_governor(data, CPUFREQ_GOV_STOP);
@@ -676,6 +753,12 @@ static int cpufreq_resume(struct sys_device * sysdev)
 	if (!cpu_policy)
 		return -EINVAL;
 
+	/* only handle each CPU group once */
+	if (unlikely(cpu_policy->cpu != cpu)) {
+		cpufreq_cpu_put(cpu_policy);
+		return 0;
+	}
+
 	if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
 		unsigned int cur_freq = 0;
 

drivers/cpufreq/cpufreq_userspace.c

@@ -515,7 +515,7 @@ static int cpufreq_governor_userspace(struct cpufreq_policy *policy,
 	unsigned int cpu = policy->cpu;
 	switch (event) {
 	case CPUFREQ_GOV_START:
-		if ((!cpu_online(cpu)) || (!try_module_get(THIS_MODULE)))
+		if (!cpu_online(cpu))
 			return -EINVAL;
 		BUG_ON(!policy->cur);
 		down(&userspace_sem);
@@ -534,7 +534,6 @@ static int cpufreq_governor_userspace(struct cpufreq_policy *policy,
 		cpu_max_freq[cpu] = 0;
 		sysfs_remove_file (&policy->kobj, &freq_attr_scaling_setspeed.attr);
 		up(&userspace_sem);
-		module_put(THIS_MODULE);
 		break;
 	case CPUFREQ_GOV_LIMITS:
 		down(&userspace_sem);

include/linux/cpufreq.h

@@ -22,6 +22,7 @@
 #include <linux/sysfs.h>
 #include <linux/completion.h>
 #include <linux/workqueue.h>
+#include <linux/cpumask.h>
 
 #define CPUFREQ_NAME_LEN 16
 
@@ -69,7 +70,8 @@ struct cpufreq_real_policy {
 };
 
 struct cpufreq_policy {
-	unsigned int		cpu;    /* cpu nr */
+	cpumask_t		cpus;	/* affected CPUs */
+	unsigned int		cpu;    /* cpu nr of registered CPU */
 	struct cpufreq_cpuinfo	cpuinfo;/* see above */
 
 	unsigned int		min;    /* in kHz */