Merge branch 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq

* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq:
  [CPUFREQ] Fix NULL ptr regression in powernow-k8
  [CPUFREQ] Create a blacklist for processors that should not load the acpi-cpufreq module.
  [CPUFREQ] Powernow-k8: Enable more than 2 low P-states
  [CPUFREQ] remove rwsem lock from CPUFREQ_GOV_STOP call (second call site)
  [CPUFREQ] ondemand - Use global sysfs dir for tuning settings
  [CPUFREQ] Introduce global, not per core: /sys/devices/system/cpu/cpufreq
  [CPUFREQ] Bail out of cpufreq_add_dev if the link for a managed CPU got created
  [CPUFREQ] Factor out policy setting from cpufreq_add_dev
  [CPUFREQ] Factor out interface creation from cpufreq_add_dev
  [CPUFREQ] Factor out symlink creation from cpufreq_add_dev
  [CPUFREQ] cleanup up -ENOMEM handling in cpufreq_add_dev
  [CPUFREQ] Reduce scope of cpu_sys_dev in cpufreq_add_dev
  [CPUFREQ] update Doc for cpuinfo_cur_freq and scaling_cur_freq

6 files changed, 350 insertions, 178 deletions

diff --git a/Documentation/cpu-freq/user-guide.txt b/Documentation/cpu-freq/user-guide.txt
index 5d5f5fadd1c2..2a5b850847c0 100644
--- a/Documentation/cpu-freq/user-guide.txt
+++ b/Documentation/cpu-freq/user-guide.txt
@@ -176,7 +176,9 @@ scaling_governor,		and by "echoing" the name of another
                                 work on some specific architectures or
                                 processors.
 
-cpuinfo_cur_freq :              Current speed of the CPU, in KHz.
+cpuinfo_cur_freq :              Current frequency of the CPU as obtained from
+                                the hardware, in KHz. This is the frequency
+                                the CPU actually runs at.
 
 scaling_available_frequencies : List of available frequencies, in KHz.
 
@@ -196,7 +198,10 @@ related_cpus :			List of CPUs that need some sort of frequency
 
 scaling_driver :                Hardware driver for cpufreq.
 
-scaling_cur_freq :              Current frequency of the CPU, in KHz.
+scaling_cur_freq :              Current frequency of the CPU as determined by
+                                the governor and cpufreq core, in KHz. This is
+                                the frequency the kernel thinks the CPU runs
+                                at.
 
 If you have selected the "userspace" governor which allows you to
 set the CPU operating frequency to a specific value, you can read out
diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
index 4109679863c1..7bb676c533aa 100644
--- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
@@ -526,6 +526,21 @@ static const struct dmi_system_id sw_any_bug_dmi_table[] = {
         },
         { }
 };
+
+static int acpi_cpufreq_blacklist(struct cpuinfo_x86 *c)
+{
+        /* http://www.intel.com/Assets/PDF/specupdate/314554.pdf
+         * AL30: A Machine Check Exception (MCE) Occurring during an
+         * Enhanced Intel SpeedStep Technology Ratio Change May Cause
+         * Both Processor Cores to Lock Up when HT is enabled*/
+        if (c->x86_vendor == X86_VENDOR_INTEL) {
+                if ((c->x86 == 15) &&
+                    (c->x86_model == 6) &&
+                    (c->x86_mask == 8) && smt_capable())
+                        return -ENODEV;
+                }
+        return 0;
+}
 #endif
 
 static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy)
@@ -540,6 +555,12 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy)
 
         dprintk("acpi_cpufreq_cpu_init\n");
 
+#ifdef CONFIG_SMP
+        result = acpi_cpufreq_blacklist(c);
+        if (result)
+                return result;
+#endif
+
         data = kzalloc(sizeof(struct acpi_cpufreq_data), GFP_KERNEL);
         if (!data)
                 return -ENOMEM;
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
index 2a50ef891000..6394aa5c7985 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
@@ -605,9 +605,10 @@ static int check_pst_table(struct powernow_k8_data *data, struct pst_s *pst,
         return 0;
 }
 
-static void invalidate_entry(struct powernow_k8_data *data, unsigned int entry)
+static void invalidate_entry(struct cpufreq_frequency_table *powernow_table,
+                unsigned int entry)
 {
-        data->powernow_table[entry].frequency = CPUFREQ_ENTRY_INVALID;
+        powernow_table[entry].frequency = CPUFREQ_ENTRY_INVALID;
 }
 
 static void print_basics(struct powernow_k8_data *data)
@@ -854,6 +855,10 @@ static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data)
                 goto err_out;
         }
 
+        /* fill in data */
+        data->numps = data->acpi_data.state_count;
+        powernow_k8_acpi_pst_values(data, 0);
+
         if (cpu_family == CPU_HW_PSTATE)
                 ret_val = fill_powernow_table_pstate(data, powernow_table);
         else
@@ -866,11 +871,8 @@ static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data)
         powernow_table[data->acpi_data.state_count].index = 0;
         data->powernow_table = powernow_table;
 
-        /* fill in data */
-        data->numps = data->acpi_data.state_count;
         if (cpumask_first(cpu_core_mask(data->cpu)) == data->cpu)
                 print_basics(data);
-        powernow_k8_acpi_pst_values(data, 0);
 
         /* notify BIOS that we exist */
         acpi_processor_notify_smm(THIS_MODULE);
@@ -914,13 +916,13 @@ static int fill_powernow_table_pstate(struct powernow_k8_data *data,
                                         "bad value %d.\n", i, index);
                         printk(KERN_ERR PFX "Please report to BIOS "
                                         "manufacturer\n");
-                        invalidate_entry(data, i);
+                        invalidate_entry(powernow_table, i);
                         continue;
                 }
                 rdmsr(MSR_PSTATE_DEF_BASE + index, lo, hi);
                 if (!(hi & HW_PSTATE_VALID_MASK)) {
                         dprintk("invalid pstate %d, ignoring\n", index);
-                        invalidate_entry(data, i);
+                        invalidate_entry(powernow_table, i);
                         continue;
                 }
 
@@ -941,7 +943,6 @@ static int fill_powernow_table_fidvid(struct powernow_k8_data *data,
                 struct cpufreq_frequency_table *powernow_table)
 {
         int i;
-        int cntlofreq = 0;
 
         for (i = 0; i < data->acpi_data.state_count; i++) {
                 u32 fid;
@@ -970,7 +971,7 @@ static int fill_powernow_table_fidvid(struct powernow_k8_data *data,
                 /* verify frequency is OK */
                 if ((freq > (MAX_FREQ * 1000)) || (freq < (MIN_FREQ * 1000))) {
                         dprintk("invalid freq %u kHz, ignoring\n", freq);
-                        invalidate_entry(data, i);
+                        invalidate_entry(powernow_table, i);
                         continue;
                 }
 
@@ -978,38 +979,17 @@ static int fill_powernow_table_fidvid(struct powernow_k8_data *data,
                  * BIOSs are using "off" to indicate invalid */
                 if (vid == VID_OFF) {
                         dprintk("invalid vid %u, ignoring\n", vid);
-                        invalidate_entry(data, i);
+                        invalidate_entry(powernow_table, i);
                         continue;
                 }
 
-                /* verify only 1 entry from the lo frequency table */
-                if (fid < HI_FID_TABLE_BOTTOM) {
-                        if (cntlofreq) {
-                                /* if both entries are the same,
-                                 * ignore this one ... */
-                                if ((freq != powernow_table[cntlofreq].frequency) ||
-                                    (index != powernow_table[cntlofreq].index)) {
-                                        printk(KERN_ERR PFX
-                                                "Too many lo freq table "
-                                                "entries\n");
-                                        return 1;
-                                }
-
-                                dprintk("double low frequency table entry, "
-                                                "ignoring it.\n");
-                                invalidate_entry(data, i);
-                                continue;
-                        } else
-                                cntlofreq = i;
-                }
-
                 if (freq != (data->acpi_data.states[i].core_frequency * 1000)) {
                         printk(KERN_INFO PFX "invalid freq entries "
                                 "%u kHz vs. %u kHz\n", freq,
                                 (unsigned int)
                                 (data->acpi_data.states[i].core_frequency
                                  * 1000));
-                        invalidate_entry(data, i);
+                        invalidate_entry(powernow_table, i);
                         continue;
                 }
         }
diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index 2968ed6a9c49..3938c7817095 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -61,6 +61,8 @@ static DEFINE_SPINLOCK(cpufreq_driver_lock);
  *   are concerned with are online after they get the lock.
  * - Governor routines that can be called in cpufreq hotplug path should not
  *   take this sem as top level hotplug notifier handler takes this.
+ * - Lock should not be held across
+ *     __cpufreq_governor(data, CPUFREQ_GOV_STOP);
  */
 static DEFINE_PER_CPU(int, policy_cpu);
 static DEFINE_PER_CPU(struct rw_semaphore, cpu_policy_rwsem);
@@ -686,6 +688,9 @@ static struct attribute *default_attrs[] = {
         NULL
 };
 
+struct kobject *cpufreq_global_kobject;
+EXPORT_SYMBOL(cpufreq_global_kobject);
+
 #define to_policy(k) container_of(k, struct cpufreq_policy, kobj)
 #define to_attr(a) container_of(a, struct freq_attr, attr)
 
@@ -756,92 +761,20 @@ static struct kobj_type ktype_cpufreq = {
         .release        = cpufreq_sysfs_release,
 };
 
-
-/**
- * cpufreq_add_dev - add a CPU device
- *
- * Adds the cpufreq interface for a CPU device.
- *
- * The Oracle says: try running cpufreq registration/unregistration concurrently
- * with with cpu hotplugging and all hell will break loose. Tried to clean this
- * mess up, but more thorough testing is needed. - Mathieu
+/*
+ * Returns:
+ *   Negative: Failure
+ *   0:        Success
+ *   Positive: When we have a managed CPU and the sysfs got symlinked
  */
-static int cpufreq_add_dev(struct sys_device *sys_dev)
+int cpufreq_add_dev_policy(unsigned int cpu, struct cpufreq_policy *policy,
+                struct sys_device *sys_dev)
 {
-        unsigned int cpu = sys_dev->id;
         int ret = 0;
-        struct cpufreq_policy new_policy;
-        struct cpufreq_policy *policy;
-        struct freq_attr **drv_attr;
-        struct sys_device *cpu_sys_dev;
+#ifdef CONFIG_SMP
         unsigned long flags;
         unsigned int j;
 
-        if (cpu_is_offline(cpu))
-                return 0;
-
-        cpufreq_debug_disable_ratelimit();
-        dprintk("adding CPU %u\n", cpu);
-
-#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)) {
-                cpufreq_cpu_put(policy);
-                cpufreq_debug_enable_ratelimit();
-                return 0;
-        }
-#endif
-
-        if (!try_module_get(cpufreq_driver->owner)) {
-                ret = -EINVAL;
-                goto module_out;
-        }
-
-        policy = kzalloc(sizeof(struct cpufreq_policy), GFP_KERNEL);
-        if (!policy) {
-                ret = -ENOMEM;
-                goto nomem_out;
-        }
-        if (!alloc_cpumask_var(&policy->cpus, GFP_KERNEL)) {
-                ret = -ENOMEM;
-                goto err_free_policy;
-        }
-        if (!zalloc_cpumask_var(&policy->related_cpus, GFP_KERNEL)) {
-                ret = -ENOMEM;
-                goto err_free_cpumask;
-        }
-
-        policy->cpu = cpu;
-        cpumask_copy(policy->cpus, cpumask_of(cpu));
-
-        /* Initially set CPU itself as the policy_cpu */
-        per_cpu(policy_cpu, cpu) = cpu;
-        ret = (lock_policy_rwsem_write(cpu) < 0);
-        WARN_ON(ret);
-
-        init_completion(&policy->kobj_unregister);
-        INIT_WORK(&policy->update, handle_update);
-
-        /* Set governor before ->init, so that driver could check it */
-        policy->governor = CPUFREQ_DEFAULT_GOVERNOR;
-        /* call driver. From then on the cpufreq must be able
-         * to accept all calls to ->verify and ->setpolicy for this CPU
-         */
-        ret = cpufreq_driver->init(policy);
-        if (ret) {
-                dprintk("initialization failed\n");
-                goto err_unlock_policy;
-        }
-        policy->user_policy.min = policy->min;
-        policy->user_policy.max = policy->max;
-
-        blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
-                                     CPUFREQ_START, policy);
-
-#ifdef CONFIG_SMP
-
 #ifdef CONFIG_HOTPLUG_CPU
         if (per_cpu(cpufreq_cpu_governor, cpu)) {
                 policy->governor = per_cpu(cpufreq_cpu_governor, cpu);
@@ -872,9 +805,8 @@ static int cpufreq_add_dev(struct sys_device *sys_dev)
                                 /* Should not go through policy unlock path */
                                 if (cpufreq_driver->exit)
                                         cpufreq_driver->exit(policy);
-                                ret = -EBUSY;
                                 cpufreq_cpu_put(managed_policy);
-                                goto err_free_cpumask;
+                                return -EBUSY;
                         }
 
                         spin_lock_irqsave(&cpufreq_driver_lock, flags);
@@ -893,17 +825,62 @@ static int cpufreq_add_dev(struct sys_device *sys_dev)
                          * Call driver->exit() because only the cpu parent of
                          * the kobj needed to call init().
                          */
-                        goto out_driver_exit; /* call driver->exit() */
+                        if (cpufreq_driver->exit)
+                                cpufreq_driver->exit(policy);
+
+                        if (!ret)
+                                return 1;
+                        else
+                                return ret;
                 }
         }
 #endif
-        memcpy(&new_policy, policy, sizeof(struct cpufreq_policy));
+        return ret;
+}
+
+
+/* symlink affected CPUs */
+int cpufreq_add_dev_symlink(unsigned int cpu, struct cpufreq_policy *policy)
+{
+        unsigned int j;
+        int ret = 0;
+
+        for_each_cpu(j, policy->cpus) {
+                struct cpufreq_policy *managed_policy;
+                struct sys_device *cpu_sys_dev;
+
+                if (j == cpu)
+                        continue;
+                if (!cpu_online(j))
+                        continue;
+
+                dprintk("CPU %u already managed, adding link\n", j);
+                managed_policy = cpufreq_cpu_get(cpu);
+                cpu_sys_dev = get_cpu_sysdev(j);
+                ret = sysfs_create_link(&cpu_sys_dev->kobj, &policy->kobj,
+                                        "cpufreq");
+                if (ret) {
+                        cpufreq_cpu_put(managed_policy);
+                        return ret;
+                }
+        }
+        return ret;
+}
+
+int cpufreq_add_dev_interface(unsigned int cpu, struct cpufreq_policy *policy,
+                struct sys_device *sys_dev)
+{
+        struct cpufreq_policy new_policy;
+        struct freq_attr **drv_attr;
+        unsigned long flags;
+        int ret = 0;
+        unsigned int j;
 
         /* prepare interface data */
-        ret = kobject_init_and_add(&policy->kobj, &ktype_cpufreq, &sys_dev->kobj,
-                                   "cpufreq");
+        ret = kobject_init_and_add(&policy->kobj, &ktype_cpufreq,
+                                   &sys_dev->kobj, "cpufreq");
         if (ret)
-                goto out_driver_exit;
+                return ret;
 
         /* set up files for this cpu device */
         drv_attr = cpufreq_driver->attr;
@@ -926,35 +903,20 @@ static int cpufreq_add_dev(struct sys_device *sys_dev)
 
         spin_lock_irqsave(&cpufreq_driver_lock, flags);
         for_each_cpu(j, policy->cpus) {
-                if (!cpu_online(j))
-                        continue;
+        if (!cpu_online(j))
+                continue;
                 per_cpu(cpufreq_cpu_data, j) = policy;
                 per_cpu(policy_cpu, j) = policy->cpu;
         }
         spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
-        /* symlink affected CPUs */
-        for_each_cpu(j, policy->cpus) {
-                struct cpufreq_policy *managed_policy;
-
-                if (j == cpu)
-                        continue;
-                if (!cpu_online(j))
-                        continue;
-
-                dprintk("CPU %u already managed, adding link\n", j);
-                managed_policy = cpufreq_cpu_get(cpu);
-                cpu_sys_dev = get_cpu_sysdev(j);
-                ret = sysfs_create_link(&cpu_sys_dev->kobj, &policy->kobj,
-                                        "cpufreq");
-                if (ret) {
-                        cpufreq_cpu_put(managed_policy);
-                        goto err_out_unregister;
-                }
-        }
+        ret = cpufreq_add_dev_symlink(cpu, policy);
+        if (ret)
+                goto err_out_kobj_put;
 
-        policy->governor = NULL; /* to assure that the starting sequence is
-                                  * run in cpufreq_set_policy */
+        memcpy(&new_policy, policy, sizeof(struct cpufreq_policy));
+        /* assure that the starting sequence is run in __cpufreq_set_policy */
+        policy->governor = NULL;
 
         /* set default policy */
         ret = __cpufreq_set_policy(policy, &new_policy);
@@ -963,8 +925,107 @@ static int cpufreq_add_dev(struct sys_device *sys_dev)
 
         if (ret) {
                 dprintk("setting policy failed\n");
-                goto err_out_unregister;
+                if (cpufreq_driver->exit)
+                        cpufreq_driver->exit(policy);
+        }
+        return ret;
+
+err_out_kobj_put:
+        kobject_put(&policy->kobj);
+        wait_for_completion(&policy->kobj_unregister);
+        return ret;
+}
+
+
+/**
+ * cpufreq_add_dev - add a CPU device
+ *
+ * Adds the cpufreq interface for a CPU device.
+ *
+ * The Oracle says: try running cpufreq registration/unregistration concurrently
+ * with with cpu hotplugging and all hell will break loose. Tried to clean this
+ * mess up, but more thorough testing is needed. - Mathieu
+ */
+static int cpufreq_add_dev(struct sys_device *sys_dev)
+{
+        unsigned int cpu = sys_dev->id;
+        int ret = 0;
+        struct cpufreq_policy *policy;
+        unsigned long flags;
+        unsigned int j;
+
+        if (cpu_is_offline(cpu))
+                return 0;
+
+        cpufreq_debug_disable_ratelimit();
+        dprintk("adding CPU %u\n", cpu);
+
+#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)) {
+                cpufreq_cpu_put(policy);
+                cpufreq_debug_enable_ratelimit();
+                return 0;
+        }
+#endif
+
+        if (!try_module_get(cpufreq_driver->owner)) {
+                ret = -EINVAL;
+                goto module_out;
+        }
+
+        ret = -ENOMEM;
+        policy = kzalloc(sizeof(struct cpufreq_policy), GFP_KERNEL);
+        if (!policy)
+                goto nomem_out;
+
+        if (!alloc_cpumask_var(&policy->cpus, GFP_KERNEL))
+                goto err_free_policy;
+
+        if (!zalloc_cpumask_var(&policy->related_cpus, GFP_KERNEL))
+                goto err_free_cpumask;
+
+        policy->cpu = cpu;
+        cpumask_copy(policy->cpus, cpumask_of(cpu));
+
+        /* Initially set CPU itself as the policy_cpu */
+        per_cpu(policy_cpu, cpu) = cpu;
+        ret = (lock_policy_rwsem_write(cpu) < 0);
+        WARN_ON(ret);
+
+        init_completion(&policy->kobj_unregister);
+        INIT_WORK(&policy->update, handle_update);
+
+        /* Set governor before ->init, so that driver could check it */
+        policy->governor = CPUFREQ_DEFAULT_GOVERNOR;
+        /* call driver. From then on the cpufreq must be able
+         * to accept all calls to ->verify and ->setpolicy for this CPU
+         */
+        ret = cpufreq_driver->init(policy);
+        if (ret) {
+                dprintk("initialization failed\n");
+                goto err_unlock_policy;
         }
+        policy->user_policy.min = policy->min;
+        policy->user_policy.max = policy->max;
+
+        blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
+                                     CPUFREQ_START, policy);
+
+        ret = cpufreq_add_dev_policy(cpu, policy, sys_dev);
+        if (ret) {
+                if (ret > 0)
+                        /* This is a managed cpu, symlink created,
+                           exit with 0 */
+                        ret = 0;
+                goto err_unlock_policy;
+        }
+
+        ret = cpufreq_add_dev_interface(cpu, policy, sys_dev);
+        if (ret)
+                goto err_out_unregister;
 
         unlock_policy_rwsem_write(cpu);
 
@@ -982,14 +1043,9 @@ err_out_unregister:
                 per_cpu(cpufreq_cpu_data, j) = NULL;
         spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
-err_out_kobj_put:
         kobject_put(&policy->kobj);
         wait_for_completion(&policy->kobj_unregister);
 
-out_driver_exit:
-        if (cpufreq_driver->exit)
-                cpufreq_driver->exit(policy);
-
 err_unlock_policy:
         unlock_policy_rwsem_write(cpu);
 err_free_cpumask:
@@ -1653,8 +1709,17 @@ static int __cpufreq_set_policy(struct cpufreq_policy *data,
                         dprintk("governor switch\n");
 
                         /* end old governor */
-                        if (data->governor)
+                        if (data->governor) {
+                                /*
+                                 * Need to release the rwsem around governor
+                                 * stop due to lock dependency between
+                                 * cancel_delayed_work_sync and the read lock
+                                 * taken in the delayed work handler.
+                                 */
+                                unlock_policy_rwsem_write(data->cpu);
                                 __cpufreq_governor(data, CPUFREQ_GOV_STOP);
+                                lock_policy_rwsem_write(data->cpu);
+                        }
 
                         /* start new governor */
                         data->governor = policy->governor;
@@ -1884,7 +1949,11 @@ static int __init cpufreq_core_init(void)
                 per_cpu(policy_cpu, cpu) = -1;
                 init_rwsem(&per_cpu(cpu_policy_rwsem, cpu));
         }
+
+        cpufreq_global_kobject = kobject_create_and_add("cpufreq",
+                                                &cpu_sysdev_class.kset.kobj);
+        BUG_ON(!cpufreq_global_kobject);
+
         return 0;
 }
-
 core_initcall(cpufreq_core_init);
diff --git a/drivers/cpufreq/cpufreq_ondemand.c b/drivers/cpufreq/cpufreq_ondemand.c
index d7a528c80de8..071699de50ee 100644
--- a/drivers/cpufreq/cpufreq_ondemand.c
+++ b/drivers/cpufreq/cpufreq_ondemand.c
@@ -55,6 +55,18 @@ static unsigned int min_sampling_rate;
 #define TRANSITION_LATENCY_LIMIT                (10 * 1000 * 1000)
 
 static void do_dbs_timer(struct work_struct *work);
+static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
+                                unsigned int event);
+
+#ifndef CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND
+static
+#endif
+struct cpufreq_governor cpufreq_gov_ondemand = {
+       .name                   = "ondemand",
+       .governor               = cpufreq_governor_dbs,
+       .max_transition_latency = TRANSITION_LATENCY_LIMIT,
+       .owner                  = THIS_MODULE,
+};
 
 /* Sampling types */
 enum {DBS_NORMAL_SAMPLE, DBS_SUB_SAMPLE};
@@ -207,20 +219,23 @@ static void ondemand_powersave_bias_init(void)
 }
 
 /************************** sysfs interface ************************/
-static ssize_t show_sampling_rate_max(struct cpufreq_policy *policy, char *buf)
+
+static ssize_t show_sampling_rate_max(struct kobject *kobj,
+                                      struct attribute *attr, char *buf)
 {
         printk_once(KERN_INFO "CPUFREQ: ondemand sampling_rate_max "
                "sysfs file is deprecated - used by: %s\n", current->comm);
         return sprintf(buf, "%u\n", -1U);
 }
 
-static ssize_t show_sampling_rate_min(struct cpufreq_policy *policy, char *buf)
+static ssize_t show_sampling_rate_min(struct kobject *kobj,
+                                      struct attribute *attr, char *buf)
 {
         return sprintf(buf, "%u\n", min_sampling_rate);
 }
 
 #define define_one_ro(_name)            \
-static struct freq_attr _name =         \
+static struct global_attr _name =       \
 __ATTR(_name, 0444, show_##_name, NULL)
 
 define_one_ro(sampling_rate_max);
@@ -229,7 +244,7 @@ define_one_ro(sampling_rate_min);
 /* cpufreq_ondemand Governor Tunables */
 #define show_one(file_name, object)                                     \
 static ssize_t show_##file_name                                         \
-(struct cpufreq_policy *unused, char *buf)                              \
+(struct kobject *kobj, struct attribute *attr, char *buf)              \
 {                                                                       \
         return sprintf(buf, "%u\n", dbs_tuners_ins.object);             \
 }
@@ -238,8 +253,38 @@ show_one(up_threshold, up_threshold);
 show_one(ignore_nice_load, ignore_nice);
 show_one(powersave_bias, powersave_bias);
 
-static ssize_t store_sampling_rate(struct cpufreq_policy *unused,
-                const char *buf, size_t count)
+/*** delete after deprecation time ***/
+
+#define DEPRECATION_MSG(file_name)                                      \
+        printk_once(KERN_INFO "CPUFREQ: Per core ondemand sysfs "       \
+                    "interface is deprecated - " #file_name "\n");
+
+#define show_one_old(file_name)                                         \
+static ssize_t show_##file_name##_old                                   \
+(struct cpufreq_policy *unused, char *buf)                              \
+{                                                                       \
+        printk_once(KERN_INFO "CPUFREQ: Per core ondemand sysfs "       \
+                    "interface is deprecated - " #file_name "\n");      \
+        return show_##file_name(NULL, NULL, buf);                       \
+}
+show_one_old(sampling_rate);
+show_one_old(up_threshold);
+show_one_old(ignore_nice_load);
+show_one_old(powersave_bias);
+show_one_old(sampling_rate_min);
+show_one_old(sampling_rate_max);
+
+#define define_one_ro_old(object, _name)       \
+static struct freq_attr object =               \
+__ATTR(_name, 0444, show_##_name##_old, NULL)
+
+define_one_ro_old(sampling_rate_min_old, sampling_rate_min);
+define_one_ro_old(sampling_rate_max_old, sampling_rate_max);
+
+/*** delete after deprecation time ***/
+
+static ssize_t store_sampling_rate(struct kobject *a, struct attribute *b,
+                                   const char *buf, size_t count)
 {
         unsigned int input;
         int ret;
@@ -254,8 +299,8 @@ static ssize_t store_sampling_rate(struct cpufreq_policy *unused,
         return count;
 }
 
-static ssize_t store_up_threshold(struct cpufreq_policy *unused,
-                const char *buf, size_t count)
+static ssize_t store_up_threshold(struct kobject *a, struct attribute *b,
+                                  const char *buf, size_t count)
 {
         unsigned int input;
         int ret;
@@ -273,8 +318,8 @@ static ssize_t store_up_threshold(struct cpufreq_policy *unused,
         return count;
 }
 
-static ssize_t store_ignore_nice_load(struct cpufreq_policy *policy,
-                const char *buf, size_t count)
+static ssize_t store_ignore_nice_load(struct kobject *a, struct attribute *b,
+                                      const char *buf, size_t count)
 {
         unsigned int input;
         int ret;
@@ -310,8 +355,8 @@ static ssize_t store_ignore_nice_load(struct cpufreq_policy *policy,
         return count;
 }
 
-static ssize_t store_powersave_bias(struct cpufreq_policy *unused,
-                const char *buf, size_t count)
+static ssize_t store_powersave_bias(struct kobject *a, struct attribute *b,
+                                    const char *buf, size_t count)
 {
         unsigned int input;
         int ret;
@@ -332,7 +377,7 @@ static ssize_t store_powersave_bias(struct cpufreq_policy *unused,
 }
 
 #define define_one_rw(_name) \
-static struct freq_attr _name = \
+static struct global_attr _name = \
 __ATTR(_name, 0644, show_##_name, store_##_name)
 
 define_one_rw(sampling_rate);
@@ -355,6 +400,47 @@ static struct attribute_group dbs_attr_group = {
         .name = "ondemand",
 };
 
+/*** delete after deprecation time ***/
+
+#define write_one_old(file_name)                                        \
+static ssize_t store_##file_name##_old                                  \
+(struct cpufreq_policy *unused, const char *buf, size_t count)          \
+{                                                                       \
+       printk_once(KERN_INFO "CPUFREQ: Per core ondemand sysfs "        \
+                   "interface is deprecated - " #file_name "\n");       \
+       return store_##file_name(NULL, NULL, buf, count);                \
+}
+write_one_old(sampling_rate);
+write_one_old(up_threshold);
+write_one_old(ignore_nice_load);
+write_one_old(powersave_bias);
+
+#define define_one_rw_old(object, _name)       \
+static struct freq_attr object =               \
+__ATTR(_name, 0644, show_##_name##_old, store_##_name##_old)
+
+define_one_rw_old(sampling_rate_old, sampling_rate);
+define_one_rw_old(up_threshold_old, up_threshold);
+define_one_rw_old(ignore_nice_load_old, ignore_nice_load);
+define_one_rw_old(powersave_bias_old, powersave_bias);
+
+static struct attribute *dbs_attributes_old[] = {
+       &sampling_rate_max_old.attr,
+       &sampling_rate_min_old.attr,
+       &sampling_rate_old.attr,
+       &up_threshold_old.attr,
+       &ignore_nice_load_old.attr,
+       &powersave_bias_old.attr,
+       NULL
+};
+
+static struct attribute_group dbs_attr_group_old = {
+       .attrs = dbs_attributes_old,
+       .name = "ondemand",
+};
+
+/*** delete after deprecation time ***/
+
 /************************** sysfs end ************************/
 
 static void dbs_check_cpu(struct cpu_dbs_info_s *this_dbs_info)
@@ -545,7 +631,7 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
 
                 mutex_lock(&dbs_mutex);
 
-                rc = sysfs_create_group(&policy->kobj, &dbs_attr_group);
+                rc = sysfs_create_group(&policy->kobj, &dbs_attr_group_old);
                 if (rc) {
                         mutex_unlock(&dbs_mutex);
                         return rc;
@@ -566,13 +652,20 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
                 }
                 this_dbs_info->cpu = cpu;
                 ondemand_powersave_bias_init_cpu(cpu);
-                mutex_init(&this_dbs_info->timer_mutex);
                 /*
                  * Start the timerschedule work, when this governor
                  * is used for first time
                  */
                 if (dbs_enable == 1) {
                         unsigned int latency;
+
+                        rc = sysfs_create_group(cpufreq_global_kobject,
+                                                &dbs_attr_group);
+                        if (rc) {
+                                mutex_unlock(&dbs_mutex);
+                                return rc;
+                        }
+
                         /* policy latency is in nS. Convert it to uS first */
                         latency = policy->cpuinfo.transition_latency / 1000;
                         if (latency == 0)
@@ -586,6 +679,7 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
                 }
                 mutex_unlock(&dbs_mutex);
 
+                mutex_init(&this_dbs_info->timer_mutex);
                 dbs_timer_init(this_dbs_info);
                 break;
 
@@ -593,10 +687,13 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
                 dbs_timer_exit(this_dbs_info);
 
                 mutex_lock(&dbs_mutex);
-                sysfs_remove_group(&policy->kobj, &dbs_attr_group);
+                sysfs_remove_group(&policy->kobj, &dbs_attr_group_old);
                 mutex_destroy(&this_dbs_info->timer_mutex);
                 dbs_enable--;
                 mutex_unlock(&dbs_mutex);
+                if (!dbs_enable)
+                        sysfs_remove_group(cpufreq_global_kobject,
+                                           &dbs_attr_group);
 
                 break;
 
@@ -614,16 +711,6 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
         return 0;
 }
 
-#ifndef CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND
-static
-#endif
-struct cpufreq_governor cpufreq_gov_ondemand = {
-        .name                   = "ondemand",
-        .governor               = cpufreq_governor_dbs,
-        .max_transition_latency = TRANSITION_LATENCY_LIMIT,
-        .owner                  = THIS_MODULE,
-};
-
 static int __init cpufreq_gov_dbs_init(void)
 {
         int err;
diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h
index 161042746afc..44717eb47639 100644
--- a/include/linux/cpufreq.h
+++ b/include/linux/cpufreq.h
@@ -65,6 +65,9 @@ static inline int cpufreq_unregister_notifier(struct notifier_block *nb,
 
 struct cpufreq_governor;
 
+/* /sys/devices/system/cpu/cpufreq: entry point for global variables */
+extern struct kobject *cpufreq_global_kobject;
+
 #define CPUFREQ_ETERNAL                 (-1)
 struct cpufreq_cpuinfo {
         unsigned int            max_freq;
@@ -274,6 +277,13 @@ struct freq_attr {
         ssize_t (*store)(struct cpufreq_policy *, const char *, size_t count);
 };
 
+struct global_attr {
+        struct attribute attr;
+        ssize_t (*show)(struct kobject *kobj,
+                        struct attribute *attr, char *buf);
+        ssize_t (*store)(struct kobject *a, struct attribute *b,
+                         const char *c, size_t count);
+};
 
 /*********************************************************************
  *                        CPUFREQ 2.6. INTERFACE                     *