Merge branch 'release-2.6.27' of git://git.kernel.org/pub/scm/linux/kernel/git/ak...
[pandora-kernel.git] / drivers / cpufreq / cpufreq_conservative.c
index b3ebc8f..fe565ee 100644 (file)
@@ -23,7 +23,6 @@
 #include <linux/fs.h>
 #include <linux/sysfs.h>
 #include <linux/cpu.h>
-#include <linux/sched.h>
 #include <linux/kmod.h>
 #include <linux/workqueue.h>
 #include <linux/jiffies.h>
 #define DEF_FREQUENCY_UP_THRESHOLD             (80)
 #define DEF_FREQUENCY_DOWN_THRESHOLD           (20)
 
-/* 
- * The polling frequency of this governor depends on the capability of 
+/*
+ * The polling frequency of this governor depends on the capability of
  * the processor. Default polling frequency is 1000 times the transition
- * latency of the processor. The governor will work on any processor with 
- * transition latency <= 10mS, using appropriate sampling 
+ * latency of the processor. The governor will work on any processor with
+ * transition latency <= 10mS, using appropriate sampling
  * rate.
- * For CPUs with transition latency > 10mS (mostly drivers with CPUFREQ_ETERNAL)
- * this governor will not work.
+ * For CPUs with transition latency > 10mS (mostly drivers
+ * with CPUFREQ_ETERNAL), this governor will not work.
  * All times here are in uS.
  */
-static unsigned int                            def_sampling_rate;
+static unsigned int def_sampling_rate;
 #define MIN_SAMPLING_RATE_RATIO                        (2)
 /* for correct statistics, we need at least 10 ticks between each measure */
-#define MIN_STAT_SAMPLING_RATE                 (MIN_SAMPLING_RATE_RATIO * jiffies_to_usecs(10))
-#define MIN_SAMPLING_RATE                      (def_sampling_rate / MIN_SAMPLING_RATE_RATIO)
+#define MIN_STAT_SAMPLING_RATE                 \
+                       (MIN_SAMPLING_RATE_RATIO * jiffies_to_usecs(10))
+#define MIN_SAMPLING_RATE                      \
+                       (def_sampling_rate / MIN_SAMPLING_RATE_RATIO)
 #define MAX_SAMPLING_RATE                      (500 * def_sampling_rate)
 #define DEF_SAMPLING_RATE_LATENCY_MULTIPLIER   (1000)
 #define DEF_SAMPLING_DOWN_FACTOR               (1)
 #define MAX_SAMPLING_DOWN_FACTOR               (10)
-#define TRANSITION_LATENCY_LIMIT               (10 * 1000)
+#define TRANSITION_LATENCY_LIMIT               (10 * 1000 * 1000)
 
-static void do_dbs_timer(void *data);
+static void do_dbs_timer(struct work_struct *work);
 
 struct cpu_dbs_info_s {
-       struct cpufreq_policy   *cur_policy;
-       unsigned int            prev_cpu_idle_up;
-       unsigned int            prev_cpu_idle_down;
-       unsigned int            enable;
-       unsigned int            down_skip;
-       unsigned int            requested_freq;
+       struct cpufreq_policy *cur_policy;
+       unsigned int prev_cpu_idle_up;
+       unsigned int prev_cpu_idle_down;
+       unsigned int enable;
+       unsigned int down_skip;
+       unsigned int requested_freq;
 };
 static DEFINE_PER_CPU(struct cpu_dbs_info_s, cpu_dbs_info);
 
@@ -81,35 +82,61 @@ static unsigned int dbs_enable;     /* number of CPUs using this policy */
  * cpu_hotplug lock should be taken before that. Note that cpu_hotplug lock
  * is recursive for the same process. -Venki
  */
-static DEFINE_MUTEX    (dbs_mutex);
-static DECLARE_WORK    (dbs_work, do_dbs_timer, NULL);
+static DEFINE_MUTEX (dbs_mutex);
+static DECLARE_DELAYED_WORK(dbs_work, do_dbs_timer);
 
 struct dbs_tuners {
-       unsigned int            sampling_rate;
-       unsigned int            sampling_down_factor;
-       unsigned int            up_threshold;
-       unsigned int            down_threshold;
-       unsigned int            ignore_nice;
-       unsigned int            freq_step;
+       unsigned int sampling_rate;
+       unsigned int sampling_down_factor;
+       unsigned int up_threshold;
+       unsigned int down_threshold;
+       unsigned int ignore_nice;
+       unsigned int freq_step;
 };
 
 static struct dbs_tuners dbs_tuners_ins = {
-       .up_threshold           = DEF_FREQUENCY_UP_THRESHOLD,
-       .down_threshold         = DEF_FREQUENCY_DOWN_THRESHOLD,
-       .sampling_down_factor   = DEF_SAMPLING_DOWN_FACTOR,
-       .ignore_nice            = 0,
-       .freq_step              = 5,
+       .up_threshold = DEF_FREQUENCY_UP_THRESHOLD,
+       .down_threshold = DEF_FREQUENCY_DOWN_THRESHOLD,
+       .sampling_down_factor = DEF_SAMPLING_DOWN_FACTOR,
+       .ignore_nice = 0,
+       .freq_step = 5,
 };
 
 static inline unsigned int get_cpu_idle_time(unsigned int cpu)
 {
-       return  kstat_cpu(cpu).cpustat.idle +
+       unsigned int add_nice = 0, ret;
+
+       if (dbs_tuners_ins.ignore_nice)
+               add_nice = kstat_cpu(cpu).cpustat.nice;
+
+       ret = kstat_cpu(cpu).cpustat.idle +
                kstat_cpu(cpu).cpustat.iowait +
-               ( dbs_tuners_ins.ignore_nice ?
-                 kstat_cpu(cpu).cpustat.nice :
-                 0);
+               add_nice;
+
+       return ret;
+}
+
+/* keep track of frequency transitions */
+static int
+dbs_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
+                    void *data)
+{
+       struct cpufreq_freqs *freq = data;
+       struct cpu_dbs_info_s *this_dbs_info = &per_cpu(cpu_dbs_info,
+                                                       freq->cpu);
+
+       if (!this_dbs_info->enable)
+               return 0;
+
+       this_dbs_info->requested_freq = freq->new;
+
+       return 0;
 }
 
+static struct notifier_block dbs_cpufreq_notifier_block = {
+       .notifier_call = dbs_cpufreq_notifier
+};
+
 /************************** sysfs interface ************************/
 static ssize_t show_sampling_rate_max(struct cpufreq_policy *policy, char *buf)
 {
@@ -121,8 +148,8 @@ static ssize_t show_sampling_rate_min(struct cpufreq_policy *policy, char *buf)
        return sprintf (buf, "%u\n", MIN_SAMPLING_RATE);
 }
 
-#define define_one_ro(_name)                                   \
-static struct freq_attr _name =                                \
+#define define_one_ro(_name)                           \
+static struct freq_attr _name =                                \
 __ATTR(_name, 0444, show_##_name, NULL)
 
 define_one_ro(sampling_rate_max);
@@ -142,7 +169,7 @@ show_one(down_threshold, down_threshold);
 show_one(ignore_nice_load, ignore_nice);
 show_one(freq_step, freq_step);
 
-static ssize_t store_sampling_down_factor(struct cpufreq_policy *unused, 
+static ssize_t store_sampling_down_factor(struct cpufreq_policy *unused,
                const char *buf, size_t count)
 {
        unsigned int input;
@@ -158,7 +185,7 @@ static ssize_t store_sampling_down_factor(struct cpufreq_policy *unused,
        return count;
 }
 
-static ssize_t store_sampling_rate(struct cpufreq_policy *unused, 
+static ssize_t store_sampling_rate(struct cpufreq_policy *unused,
                const char *buf, size_t count)
 {
        unsigned int input;
@@ -177,7 +204,7 @@ static ssize_t store_sampling_rate(struct cpufreq_policy *unused,
        return count;
 }
 
-static ssize_t store_up_threshold(struct cpufreq_policy *unused, 
+static ssize_t store_up_threshold(struct cpufreq_policy *unused,
                const char *buf, size_t count)
 {
        unsigned int input;
@@ -196,7 +223,7 @@ static ssize_t store_up_threshold(struct cpufreq_policy *unused,
        return count;
 }
 
-static ssize_t store_down_threshold(struct cpufreq_policy *unused, 
+static ssize_t store_down_threshold(struct cpufreq_policy *unused,
                const char *buf, size_t count)
 {
        unsigned int input;
@@ -222,16 +249,16 @@ static ssize_t store_ignore_nice_load(struct cpufreq_policy *policy,
        int ret;
 
        unsigned int j;
-       
-       ret = sscanf (buf, "%u", &input);
-       if ( ret != 1 )
+
+       ret = sscanf(buf, "%u", &input);
+       if (ret != 1)
                return -EINVAL;
 
-       if ( input > 1 )
+       if (input > 1)
                input = 1;
-       
+
        mutex_lock(&dbs_mutex);
-       if ( input == dbs_tuners_ins.ignore_nice ) { /* nothing to do */
+       if (input == dbs_tuners_ins.ignore_nice) { /* nothing to do */
                mutex_unlock(&dbs_mutex);
                return count;
        }
@@ -255,14 +282,14 @@ static ssize_t store_freq_step(struct cpufreq_policy *policy,
        unsigned int input;
        int ret;
 
-       ret = sscanf (buf, "%u", &input);
+       ret = sscanf(buf, "%u", &input);
 
-       if ( ret != 1 )
+       if (ret != 1)
                return -EINVAL;
 
-       if ( input > 100 )
+       if (input > 100)
                input = 100;
-       
+
        /* no need to test here if freq_step is zero as the user might actually
         * want this, they would be crazy though :) */
        mutex_lock(&dbs_mutex);
@@ -316,18 +343,18 @@ static void dbs_check_cpu(int cpu)
 
        policy = this_dbs_info->cur_policy;
 
-       /* 
-        * The default safe range is 20% to 80% 
+       /*
+        * The default safe range is 20% to 80%
         * Every sampling_rate, we check
-        *      - If current idle time is less than 20%, then we try to 
-        *        increase frequency
+        *      - If current idle time is less than 20%, then we try to
+        *        increase frequency
         * Every sampling_rate*sampling_down_factor, we check
-        *      - If current idle time is more than 80%, then we try to
-        *        decrease frequency
+        *      - If current idle time is more than 80%, then we try to
+        *        decrease frequency
         *
-        * Any frequency increase takes it to the maximum frequency. 
-        * Frequency reduction happens at minimum steps of 
-        * 5% (default) of max_frequency 
+        * Any frequency increase takes it to the maximum frequency.
+        * Frequency reduction happens at minimum steps of
+        * 5% (default) of max_frequency
         */
 
        /* Check for frequency increase */
@@ -355,13 +382,13 @@ static void dbs_check_cpu(int cpu)
                /* if we are already at full speed then break out early */
                if (this_dbs_info->requested_freq == policy->max)
                        return;
-               
+
                freq_step = (dbs_tuners_ins.freq_step * policy->max) / 100;
 
                /* max freq cannot be less than 100. But who knows.... */
                if (unlikely(freq_step == 0))
                        freq_step = 5;
-               
+
                this_dbs_info->requested_freq += freq_step;
                if (this_dbs_info->requested_freq > policy->max)
                        this_dbs_info->requested_freq = policy->max;
@@ -420,22 +447,19 @@ static void dbs_check_cpu(int cpu)
        }
 }
 
-static void do_dbs_timer(void *data)
-{ 
+static void do_dbs_timer(struct work_struct *work)
+{
        int i;
-       lock_cpu_hotplug();
        mutex_lock(&dbs_mutex);
        for_each_online_cpu(i)
                dbs_check_cpu(i);
-       schedule_delayed_work(&dbs_work, 
+       schedule_delayed_work(&dbs_work,
                        usecs_to_jiffies(dbs_tuners_ins.sampling_rate));
        mutex_unlock(&dbs_mutex);
-       unlock_cpu_hotplug();
-} 
+}
 
 static inline void dbs_timer_init(void)
 {
-       INIT_WORK(&dbs_work, do_dbs_timer, NULL);
        schedule_delayed_work(&dbs_work,
                        usecs_to_jiffies(dbs_tuners_ins.sampling_rate));
        return;
@@ -453,27 +477,31 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
        unsigned int cpu = policy->cpu;
        struct cpu_dbs_info_s *this_dbs_info;
        unsigned int j;
+       int rc;
 
        this_dbs_info = &per_cpu(cpu_dbs_info, cpu);
 
        switch (event) {
        case CPUFREQ_GOV_START:
-               if ((!cpu_online(cpu)) || 
-                   (!policy->cur))
+               if ((!cpu_online(cpu)) || (!policy->cur))
                        return -EINVAL;
 
-               if (policy->cpuinfo.transition_latency >
-                               (TRANSITION_LATENCY_LIMIT * 1000))
-                       return -EINVAL;
                if (this_dbs_info->enable) /* Already enabled */
                        break;
-                
+
                mutex_lock(&dbs_mutex);
-               for_each_cpu_mask(j, policy->cpus) {
+
+               rc = sysfs_create_group(&policy->kobj, &dbs_attr_group);
+               if (rc) {
+                       mutex_unlock(&dbs_mutex);
+                       return rc;
+               }
+
+               for_each_cpu_mask_nr(j, policy->cpus) {
                        struct cpu_dbs_info_s *j_dbs_info;
                        j_dbs_info = &per_cpu(cpu_dbs_info, j);
                        j_dbs_info->cur_policy = policy;
-               
+
                        j_dbs_info->prev_cpu_idle_up = get_cpu_idle_time(cpu);
                        j_dbs_info->prev_cpu_idle_down
                                = j_dbs_info->prev_cpu_idle_up;
@@ -481,7 +509,7 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
                this_dbs_info->enable = 1;
                this_dbs_info->down_skip = 0;
                this_dbs_info->requested_freq = policy->cur;
-               sysfs_create_group(&policy->kobj, &dbs_attr_group);
+
                dbs_enable++;
                /*
                 * Start the timerschedule work, when this governor
@@ -503,8 +531,11 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
                        dbs_tuners_ins.sampling_rate = def_sampling_rate;
 
                        dbs_timer_init();
+                       cpufreq_register_notifier(
+                                       &dbs_cpufreq_notifier_block,
+                                       CPUFREQ_TRANSITION_NOTIFIER);
                }
-               
+
                mutex_unlock(&dbs_mutex);
                break;
 
@@ -517,40 +548,44 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
                 * Stop the timerschedule work, when this governor
                 * is used for first time
                 */
-               if (dbs_enable == 0) 
+               if (dbs_enable == 0) {
                        dbs_timer_exit();
-               
+                       cpufreq_unregister_notifier(
+                                       &dbs_cpufreq_notifier_block,
+                                       CPUFREQ_TRANSITION_NOTIFIER);
+               }
+
                mutex_unlock(&dbs_mutex);
 
                break;
 
        case CPUFREQ_GOV_LIMITS:
-               lock_cpu_hotplug();
                mutex_lock(&dbs_mutex);
                if (policy->max < this_dbs_info->cur_policy->cur)
                        __cpufreq_driver_target(
                                        this_dbs_info->cur_policy,
-                                       policy->max, CPUFREQ_RELATION_H);
+                                       policy->max, CPUFREQ_RELATION_H);
                else if (policy->min > this_dbs_info->cur_policy->cur)
                        __cpufreq_driver_target(
                                        this_dbs_info->cur_policy,
-                                       policy->min, CPUFREQ_RELATION_L);
+                                       policy->min, CPUFREQ_RELATION_L);
                mutex_unlock(&dbs_mutex);
-               unlock_cpu_hotplug();
                break;
        }
        return 0;
 }
 
-static struct cpufreq_governor cpufreq_gov_dbs = {
-       .name           = "conservative",
-       .governor       = cpufreq_governor_dbs,
-       .owner          = THIS_MODULE,
+struct cpufreq_governor cpufreq_gov_conservative = {
+       .name                   = "conservative",
+       .governor               = cpufreq_governor_dbs,
+       .max_transition_latency = TRANSITION_LATENCY_LIMIT,
+       .owner                  = THIS_MODULE,
 };
+EXPORT_SYMBOL(cpufreq_gov_conservative);
 
 static int __init cpufreq_gov_dbs_init(void)
 {
-       return cpufreq_register_governor(&cpufreq_gov_dbs);
+       return cpufreq_register_governor(&cpufreq_gov_conservative);
 }
 
 static void __exit cpufreq_gov_dbs_exit(void)
@@ -558,7 +593,7 @@ static void __exit cpufreq_gov_dbs_exit(void)
        /* Make sure that the scheduled work is indeed not running */
        flush_scheduled_work();
 
-       cpufreq_unregister_governor(&cpufreq_gov_dbs);
+       cpufreq_unregister_governor(&cpufreq_gov_conservative);
 }
 
 
@@ -568,5 +603,9 @@ MODULE_DESCRIPTION ("'cpufreq_conservative' - A dynamic cpufreq governor for "
                "optimised for use in a battery environment");
 MODULE_LICENSE ("GPL");
 
+#ifdef CONFIG_CPU_FREQ_DEFAULT_GOV_CONSERVATIVE
+fs_initcall(cpufreq_gov_dbs_init);
+#else
 module_init(cpufreq_gov_dbs_init);
+#endif
 module_exit(cpufreq_gov_dbs_exit);