]> err.no Git - linux-2.6/commitdiff
[CPUFREQ] Misc cleanups in ondemand.
authorVenkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Wed, 28 Jun 2006 20:52:18 +0000 (13:52 -0700)
committerDave Jones <davej@redhat.com>
Fri, 30 Jun 2006 05:36:40 +0000 (01:36 -0400)
Misc cleanups in ondemand. Should have zero functional impact.
Also adding Alexey as author.

Signed-off-by: Alexey Starikovskiy <alexey.y.starikovskiy@intel.com>
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Dave Jones <davej@redhat.com>
drivers/cpufreq/cpufreq_ondemand.c

index 18b016ea5f4877cb10cc663ef8752b430079bc8c..87299924e73508accd8961191cc4baf99832927b 100644 (file)
 
 #include <linux/kernel.h>
 #include <linux/module.h>
-#include <linux/smp.h>
 #include <linux/init.h>
-#include <linux/interrupt.h>
-#include <linux/ctype.h>
 #include <linux/cpufreq.h>
-#include <linux/sysctl.h>
-#include <linux/types.h>
-#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>
 #include <linux/kernel_stat.h>
-#include <linux/percpu.h>
 #include <linux/mutex.h>
 
 /*
@@ -79,8 +68,7 @@ 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 struct workqueue_struct *kondemand_wq;
 
@@ -142,7 +130,7 @@ static ssize_t store_sampling_rate(struct cpufreq_policy *unused,
 {
        unsigned int input;
        int ret;
-       ret = sscanf (buf, "%u", &input);
+       ret = sscanf(buf, "%u", &input);
 
        mutex_lock(&dbs_mutex);
        if (ret != 1 || input > MAX_SAMPLING_RATE || input < MIN_SAMPLING_RATE) {
@@ -161,7 +149,7 @@ static ssize_t store_up_threshold(struct cpufreq_policy *unused,
 {
        unsigned int input;
        int ret;
-       ret = sscanf (buf, "%u", &input);
+       ret = sscanf(buf, "%u", &input);
 
        mutex_lock(&dbs_mutex);
        if (ret != 1 || input > MAX_FREQUENCY_UP_THRESHOLD ||
@@ -184,7 +172,7 @@ static ssize_t store_ignore_nice_load(struct cpufreq_policy *policy,
 
        unsigned int j;
 
-       ret = sscanf (buf, "%u", &input);
+       ret = sscanf(buf, "%u", &input);
        if ( ret != 1 )
                return -EINVAL;
 
@@ -349,8 +337,7 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
 
        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 >
@@ -424,13 +411,13 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
                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);
+                       __cpufreq_driver_target(this_dbs_info->cur_policy,
+                                               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);
+                       __cpufreq_driver_target(this_dbs_info->cur_policy,
+                                               policy->min,
+                                               CPUFREQ_RELATION_L);
                mutex_unlock(&dbs_mutex);
                unlock_cpu_hotplug();
                break;
@@ -439,9 +426,9 @@ static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
 }
 
 static struct cpufreq_governor cpufreq_gov_dbs = {
-       .name           = "ondemand",
-       .governor       = cpufreq_governor_dbs,
-       .owner          = THIS_MODULE,
+       .name = "ondemand",
+       .governor = cpufreq_governor_dbs,
+       .owner = THIS_MODULE,
 };
 
 static int __init cpufreq_gov_dbs_init(void)
@@ -455,10 +442,11 @@ static void __exit cpufreq_gov_dbs_exit(void)
 }
 
 
-MODULE_AUTHOR ("Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>");
-MODULE_DESCRIPTION ("'cpufreq_ondemand' - A dynamic cpufreq governor for "
-               "Low Latency Frequency Transition capable processors");
-MODULE_LICENSE ("GPL");
+MODULE_AUTHOR("Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>");
+MODULE_AUTHOR("Alexey Starikovskiy <alexey.y.starikovskiy@intel.com>");
+MODULE_DESCRIPTION("'cpufreq_ondemand' - A dynamic cpufreq governor for "
+                   "Low Latency Frequency Transition capable processors");
+MODULE_LICENSE("GPL");
 
 module_init(cpufreq_gov_dbs_init);
 module_exit(cpufreq_gov_dbs_exit);