From ae0fb4b72c8db7e6c4ef32bc58a43a759ad414b9 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Tue, 23 Oct 2012 01:09:12 +0200 Subject: PM / QoS: Introduce PM QoS device flags support Modify the device PM QoS core code to support PM QoS flags requests. First, add a new field of type struct pm_qos_flags called "flags" to struct dev_pm_qos for representing the list of PM QoS flags requests for the given device. Accordingly, add a new "type" field to struct dev_pm_qos_request (along with an enum for representing request types) and a new member called "flr" to its data union for representig flags requests. Second, modify dev_pm_qos_add_request(), dev_pm_qos_update_request(), the internal routine apply_constraint() used by them and their existing callers to cover flags requests as well as latency requests. In particular, dev_pm_qos_add_request() gets a new argument called "type" for specifying the type of a request to be added. Finally, introduce two routines, __dev_pm_qos_flags() and dev_pm_qos_flags(), allowing their callers to check which PM QoS flags have been requested for the given device (the caller is supposed to pass the mask of flags to check as the routine's second argument and examine its return value for the result). Signed-off-by: Rafael J. Wysocki Reviewed-by: Jean Pihet Reviewed-by: mark gross --- Documentation/power/pm_qos_interface.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'Documentation') diff --git a/Documentation/power/pm_qos_interface.txt b/Documentation/power/pm_qos_interface.txt index 17e130a80347..79a2a58425ee 100644 --- a/Documentation/power/pm_qos_interface.txt +++ b/Documentation/power/pm_qos_interface.txt @@ -99,7 +99,7 @@ reading the aggregated value does not require any locking mechanism. From kernel mode the use of this interface is the following: -int dev_pm_qos_add_request(device, handle, value): +int dev_pm_qos_add_request(device, handle, type, value): Will insert an element into the list for that identified device with the target value. Upon change to this list the new target is recomputed and any registered notifiers are called only if the target value is now different. -- cgit v1.2.3 From e39473d0b9448e770f49b0b15e514be884264438 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Wed, 24 Oct 2012 02:08:18 +0200 Subject: PM / QoS: Make it possible to expose PM QoS device flags to user space Define two device PM QoS flags, PM_QOS_FLAG_NO_POWER_OFF and PM_QOS_FLAG_REMOTE_WAKEUP, and introduce routines dev_pm_qos_expose_flags() and dev_pm_qos_hide_flags() allowing the caller to expose those two flags to user space or to hide them from it, respectively. After the flags have been exposed, user space will see two additional sysfs attributes, pm_qos_no_power_off and pm_qos_remote_wakeup, under the device's /sys/devices/.../power/ directory. Then, writing 1 to one of them will update the PM QoS flags request owned by user space so that the corresponding flag is requested to be set. In turn, writing 0 to one of them will cause the corresponding flag in the user space's request to be cleared (however, the owners of the other PM QoS flags requests for the same device may still request the flag to be set and it may be effectively set even if user space doesn't request that). Signed-off-by: Rafael J. Wysocki Reviewed-by: Jean Pihet Acked-by: mark gross --- Documentation/ABI/testing/sysfs-devices-power | 31 +++++ drivers/base/power/power.h | 6 +- drivers/base/power/qos.c | 168 ++++++++++++++++++++------ drivers/base/power/sysfs.c | 94 ++++++++++++-- include/linux/pm.h | 1 - include/linux/pm_qos.h | 26 ++++ 6 files changed, 278 insertions(+), 48 deletions(-) (limited to 'Documentation') diff --git a/Documentation/ABI/testing/sysfs-devices-power b/Documentation/ABI/testing/sysfs-devices-power index 45000f0db4d4..7fc2997b23a6 100644 --- a/Documentation/ABI/testing/sysfs-devices-power +++ b/Documentation/ABI/testing/sysfs-devices-power @@ -204,3 +204,34 @@ Description: This attribute has no effect on system-wide suspend/resume and hibernation. + +What: /sys/devices/.../power/pm_qos_no_power_off +Date: September 2012 +Contact: Rafael J. Wysocki +Description: + The /sys/devices/.../power/pm_qos_no_power_off attribute + is used for manipulating the PM QoS "no power off" flag. If + set, this flag indicates to the kernel that power should not + be removed entirely from the device. + + Not all drivers support this attribute. If it isn't supported, + it is not present. + + This attribute has no effect on system-wide suspend/resume and + hibernation. + +What: /sys/devices/.../power/pm_qos_remote_wakeup +Date: September 2012 +Contact: Rafael J. Wysocki +Description: + The /sys/devices/.../power/pm_qos_remote_wakeup attribute + is used for manipulating the PM QoS "remote wakeup required" + flag. If set, this flag indicates to the kernel that the + device is a source of user events that have to be signaled from + its low-power states. + + Not all drivers support this attribute. If it isn't supported, + it is not present. + + This attribute has no effect on system-wide suspend/resume and + hibernation. diff --git a/drivers/base/power/power.h b/drivers/base/power/power.h index 0dbfdf4419af..b16686a0a5a2 100644 --- a/drivers/base/power/power.h +++ b/drivers/base/power/power.h @@ -93,8 +93,10 @@ extern void dpm_sysfs_remove(struct device *dev); extern void rpm_sysfs_remove(struct device *dev); extern int wakeup_sysfs_add(struct device *dev); extern void wakeup_sysfs_remove(struct device *dev); -extern int pm_qos_sysfs_add(struct device *dev); -extern void pm_qos_sysfs_remove(struct device *dev); +extern int pm_qos_sysfs_add_latency(struct device *dev); +extern void pm_qos_sysfs_remove_latency(struct device *dev); +extern int pm_qos_sysfs_add_flags(struct device *dev); +extern void pm_qos_sysfs_remove_flags(struct device *dev); #else /* CONFIG_PM */ diff --git a/drivers/base/power/qos.c b/drivers/base/power/qos.c index 3c66f75d14b0..167834dcc82a 100644 --- a/drivers/base/power/qos.c +++ b/drivers/base/power/qos.c @@ -40,6 +40,7 @@ #include #include #include +#include #include "power.h" @@ -321,6 +322,37 @@ int dev_pm_qos_add_request(struct device *dev, struct dev_pm_qos_request *req, } EXPORT_SYMBOL_GPL(dev_pm_qos_add_request); +/** + * __dev_pm_qos_update_request - Modify an existing device PM QoS request. + * @req : PM QoS request to modify. + * @new_value: New value to request. + */ +static int __dev_pm_qos_update_request(struct dev_pm_qos_request *req, + s32 new_value) +{ + s32 curr_value; + int ret = 0; + + if (!req->dev->power.qos) + return -ENODEV; + + switch(req->type) { + case DEV_PM_QOS_LATENCY: + curr_value = req->data.pnode.prio; + break; + case DEV_PM_QOS_FLAGS: + curr_value = req->data.flr.flags; + break; + default: + return -EINVAL; + } + + if (curr_value != new_value) + ret = apply_constraint(req, PM_QOS_UPDATE_REQ, new_value); + + return ret; +} + /** * dev_pm_qos_update_request - modifies an existing qos request * @req : handle to list element holding a dev_pm_qos request to use @@ -336,11 +368,9 @@ EXPORT_SYMBOL_GPL(dev_pm_qos_add_request); * -EINVAL in case of wrong parameters, -ENODEV if the device has been * removed from the system */ -int dev_pm_qos_update_request(struct dev_pm_qos_request *req, - s32 new_value) +int dev_pm_qos_update_request(struct dev_pm_qos_request *req, s32 new_value) { - s32 curr_value; - int ret = 0; + int ret; if (!req) /*guard against callers passing in null */ return -EINVAL; @@ -350,29 +380,9 @@ int dev_pm_qos_update_request(struct dev_pm_qos_request *req, return -EINVAL; mutex_lock(&dev_pm_qos_mtx); - - if (!req->dev->power.qos) { - ret = -ENODEV; - goto out; - } - - switch(req->type) { - case DEV_PM_QOS_LATENCY: - curr_value = req->data.pnode.prio; - break; - case DEV_PM_QOS_FLAGS: - curr_value = req->data.flr.flags; - break; - default: - ret = -EINVAL; - goto out; - } - - if (curr_value != new_value) - ret = apply_constraint(req, PM_QOS_UPDATE_REQ, new_value); - - out: + __dev_pm_qos_update_request(req, new_value); mutex_unlock(&dev_pm_qos_mtx); + return ret; } EXPORT_SYMBOL_GPL(dev_pm_qos_update_request); @@ -533,10 +543,19 @@ int dev_pm_qos_add_ancestor_request(struct device *dev, EXPORT_SYMBOL_GPL(dev_pm_qos_add_ancestor_request); #ifdef CONFIG_PM_RUNTIME -static void __dev_pm_qos_drop_user_request(struct device *dev) +static void __dev_pm_qos_drop_user_request(struct device *dev, + enum dev_pm_qos_req_type type) { - dev_pm_qos_remove_request(dev->power.pq_req); - dev->power.pq_req = NULL; + switch(type) { + case DEV_PM_QOS_LATENCY: + dev_pm_qos_remove_request(dev->power.qos->latency_req); + dev->power.qos->latency_req = NULL; + break; + case DEV_PM_QOS_FLAGS: + dev_pm_qos_remove_request(dev->power.qos->flags_req); + dev->power.qos->flags_req = NULL; + break; + } } /** @@ -552,7 +571,7 @@ int dev_pm_qos_expose_latency_limit(struct device *dev, s32 value) if (!device_is_registered(dev) || value < 0) return -EINVAL; - if (dev->power.pq_req) + if (dev->power.qos && dev->power.qos->latency_req) return -EEXIST; req = kzalloc(sizeof(*req), GFP_KERNEL); @@ -563,10 +582,10 @@ int dev_pm_qos_expose_latency_limit(struct device *dev, s32 value) if (ret < 0) return ret; - dev->power.pq_req = req; - ret = pm_qos_sysfs_add(dev); + dev->power.qos->latency_req = req; + ret = pm_qos_sysfs_add_latency(dev); if (ret) - __dev_pm_qos_drop_user_request(dev); + __dev_pm_qos_drop_user_request(dev, DEV_PM_QOS_LATENCY); return ret; } @@ -578,10 +597,87 @@ EXPORT_SYMBOL_GPL(dev_pm_qos_expose_latency_limit); */ void dev_pm_qos_hide_latency_limit(struct device *dev) { - if (dev->power.pq_req) { - pm_qos_sysfs_remove(dev); - __dev_pm_qos_drop_user_request(dev); + if (dev->power.qos && dev->power.qos->latency_req) { + pm_qos_sysfs_remove_latency(dev); + __dev_pm_qos_drop_user_request(dev, DEV_PM_QOS_LATENCY); } } EXPORT_SYMBOL_GPL(dev_pm_qos_hide_latency_limit); + +/** + * dev_pm_qos_expose_flags - Expose PM QoS flags of a device to user space. + * @dev: Device whose PM QoS flags are to be exposed to user space. + * @val: Initial values of the flags. + */ +int dev_pm_qos_expose_flags(struct device *dev, s32 val) +{ + struct dev_pm_qos_request *req; + int ret; + + if (!device_is_registered(dev)) + return -EINVAL; + + if (dev->power.qos && dev->power.qos->flags_req) + return -EEXIST; + + req = kzalloc(sizeof(*req), GFP_KERNEL); + if (!req) + return -ENOMEM; + + ret = dev_pm_qos_add_request(dev, req, DEV_PM_QOS_FLAGS, val); + if (ret < 0) + return ret; + + dev->power.qos->flags_req = req; + ret = pm_qos_sysfs_add_flags(dev); + if (ret) + __dev_pm_qos_drop_user_request(dev, DEV_PM_QOS_FLAGS); + + return ret; +} +EXPORT_SYMBOL_GPL(dev_pm_qos_expose_flags); + +/** + * dev_pm_qos_hide_flags - Hide PM QoS flags of a device from user space. + * @dev: Device whose PM QoS flags are to be hidden from user space. + */ +void dev_pm_qos_hide_flags(struct device *dev) +{ + if (dev->power.qos && dev->power.qos->flags_req) { + pm_qos_sysfs_remove_flags(dev); + __dev_pm_qos_drop_user_request(dev, DEV_PM_QOS_FLAGS); + } +} +EXPORT_SYMBOL_GPL(dev_pm_qos_hide_flags); + +/** + * dev_pm_qos_update_flags - Update PM QoS flags request owned by user space. + * @dev: Device to update the PM QoS flags request for. + * @mask: Flags to set/clear. + * @set: Whether to set or clear the flags (true means set). + */ +int dev_pm_qos_update_flags(struct device *dev, s32 mask, bool set) +{ + s32 value; + int ret; + + if (!dev->power.qos || !dev->power.qos->flags_req) + return -EINVAL; + + pm_runtime_get_sync(dev); + mutex_lock(&dev_pm_qos_mtx); + + value = dev_pm_qos_requested_flags(dev); + if (set) + value |= mask; + else + value &= ~mask; + + ret = __dev_pm_qos_update_request(dev->power.qos->flags_req, value); + + mutex_unlock(&dev_pm_qos_mtx); + pm_runtime_put(dev); + + return ret; +} #endif /* CONFIG_PM_RUNTIME */ diff --git a/drivers/base/power/sysfs.c b/drivers/base/power/sysfs.c index 54c61ffa2044..50d16e3cb0a9 100644 --- a/drivers/base/power/sysfs.c +++ b/drivers/base/power/sysfs.c @@ -221,7 +221,7 @@ static DEVICE_ATTR(autosuspend_delay_ms, 0644, autosuspend_delay_ms_show, static ssize_t pm_qos_latency_show(struct device *dev, struct device_attribute *attr, char *buf) { - return sprintf(buf, "%d\n", dev->power.pq_req->data.pnode.prio); + return sprintf(buf, "%d\n", dev_pm_qos_requested_latency(dev)); } static ssize_t pm_qos_latency_store(struct device *dev, @@ -237,12 +237,66 @@ static ssize_t pm_qos_latency_store(struct device *dev, if (value < 0) return -EINVAL; - ret = dev_pm_qos_update_request(dev->power.pq_req, value); + ret = dev_pm_qos_update_request(dev->power.qos->latency_req, value); return ret < 0 ? ret : n; } static DEVICE_ATTR(pm_qos_resume_latency_us, 0644, pm_qos_latency_show, pm_qos_latency_store); + +static ssize_t pm_qos_no_power_off_show(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + return sprintf(buf, "%d\n", !!(dev_pm_qos_requested_flags(dev) + & PM_QOS_FLAG_NO_POWER_OFF)); +} + +static ssize_t pm_qos_no_power_off_store(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t n) +{ + int ret; + + if (kstrtoint(buf, 0, &ret)) + return -EINVAL; + + if (ret != 0 && ret != 1) + return -EINVAL; + + ret = dev_pm_qos_update_flags(dev, PM_QOS_FLAG_NO_POWER_OFF, ret); + return ret < 0 ? ret : n; +} + +static DEVICE_ATTR(pm_qos_no_power_off, 0644, + pm_qos_no_power_off_show, pm_qos_no_power_off_store); + +static ssize_t pm_qos_remote_wakeup_show(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + return sprintf(buf, "%d\n", !!(dev_pm_qos_requested_flags(dev) + & PM_QOS_FLAG_REMOTE_WAKEUP)); +} + +static ssize_t pm_qos_remote_wakeup_store(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t n) +{ + int ret; + + if (kstrtoint(buf, 0, &ret)) + return -EINVAL; + + if (ret != 0 && ret != 1) + return -EINVAL; + + ret = dev_pm_qos_update_flags(dev, PM_QOS_FLAG_REMOTE_WAKEUP, ret); + return ret < 0 ? ret : n; +} + +static DEVICE_ATTR(pm_qos_remote_wakeup, 0644, + pm_qos_remote_wakeup_show, pm_qos_remote_wakeup_store); #endif /* CONFIG_PM_RUNTIME */ #ifdef CONFIG_PM_SLEEP @@ -564,15 +618,27 @@ static struct attribute_group pm_runtime_attr_group = { .attrs = runtime_attrs, }; -static struct attribute *pm_qos_attrs[] = { +static struct attribute *pm_qos_latency_attrs[] = { #ifdef CONFIG_PM_RUNTIME &dev_attr_pm_qos_resume_latency_us.attr, #endif /* CONFIG_PM_RUNTIME */ NULL, }; -static struct attribute_group pm_qos_attr_group = { +static struct attribute_group pm_qos_latency_attr_group = { .name = power_group_name, - .attrs = pm_qos_attrs, + .attrs = pm_qos_latency_attrs, +}; + +static struct attribute *pm_qos_flags_attrs[] = { +#ifdef CONFIG_PM_RUNTIME + &dev_attr_pm_qos_no_power_off.attr, + &dev_attr_pm_qos_remote_wakeup.attr, +#endif /* CONFIG_PM_RUNTIME */ + NULL, +}; +static struct attribute_group pm_qos_flags_attr_group = { + .name = power_group_name, + .attrs = pm_qos_flags_attrs, }; int dpm_sysfs_add(struct device *dev) @@ -615,14 +681,24 @@ void wakeup_sysfs_remove(struct device *dev) sysfs_unmerge_group(&dev->kobj, &pm_wakeup_attr_group); } -int pm_qos_sysfs_add(struct device *dev) +int pm_qos_sysfs_add_latency(struct device *dev) +{ + return sysfs_merge_group(&dev->kobj, &pm_qos_latency_attr_group); +} + +void pm_qos_sysfs_remove_latency(struct device *dev) +{ + sysfs_unmerge_group(&dev->kobj, &pm_qos_latency_attr_group); +} + +int pm_qos_sysfs_add_flags(struct device *dev) { - return sysfs_merge_group(&dev->kobj, &pm_qos_attr_group); + return sysfs_merge_group(&dev->kobj, &pm_qos_flags_attr_group); } -void pm_qos_sysfs_remove(struct device *dev) +void pm_qos_sysfs_remove_flags(struct device *dev) { - sysfs_unmerge_group(&dev->kobj, &pm_qos_attr_group); + sysfs_unmerge_group(&dev->kobj, &pm_qos_flags_attr_group); } void rpm_sysfs_remove(struct device *dev) diff --git a/include/linux/pm.h b/include/linux/pm.h index 0ce6df94221a..03d7bb145311 100644 --- a/include/linux/pm.h +++ b/include/linux/pm.h @@ -546,7 +546,6 @@ struct dev_pm_info { unsigned long active_jiffies; unsigned long suspended_jiffies; unsigned long accounting_timestamp; - struct dev_pm_qos_request *pq_req; #endif struct pm_subsys_data *subsys_data; /* Owned by the subsystem. */ struct dev_pm_qos *qos; diff --git a/include/linux/pm_qos.h b/include/linux/pm_qos.h index 3af7d8573c23..5a95013905c8 100644 --- a/include/linux/pm_qos.h +++ b/include/linux/pm_qos.h @@ -34,6 +34,9 @@ enum pm_qos_flags_status { #define PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE 0 #define PM_QOS_DEV_LAT_DEFAULT_VALUE 0 +#define PM_QOS_FLAG_NO_POWER_OFF (1 << 0) +#define PM_QOS_FLAG_REMOTE_WAKEUP (1 << 1) + struct pm_qos_request { struct plist_node node; int pm_qos_class; @@ -86,6 +89,8 @@ struct pm_qos_flags { struct dev_pm_qos { struct pm_qos_constraints latency; struct pm_qos_flags flags; + struct dev_pm_qos_request *latency_req; + struct dev_pm_qos_request *flags_req; }; /* Action requested to pm_qos_update_target */ @@ -187,10 +192,31 @@ static inline int dev_pm_qos_add_ancestor_request(struct device *dev, #ifdef CONFIG_PM_RUNTIME int dev_pm_qos_expose_latency_limit(struct device *dev, s32 value); void dev_pm_qos_hide_latency_limit(struct device *dev); +int dev_pm_qos_expose_flags(struct device *dev, s32 value); +void dev_pm_qos_hide_flags(struct device *dev); +int dev_pm_qos_update_flags(struct device *dev, s32 mask, bool set); + +static inline s32 dev_pm_qos_requested_latency(struct device *dev) +{ + return dev->power.qos->latency_req->data.pnode.prio; +} + +static inline s32 dev_pm_qos_requested_flags(struct device *dev) +{ + return dev->power.qos->flags_req->data.flr.flags; +} #else static inline int dev_pm_qos_expose_latency_limit(struct device *dev, s32 value) { return 0; } static inline void dev_pm_qos_hide_latency_limit(struct device *dev) {} +static inline int dev_pm_qos_expose_flags(struct device *dev, s32 value) + { return 0; } +static inline void dev_pm_qos_hide_flags(struct device *dev) {} +static inline int dev_pm_qos_update_flags(struct device *dev, s32 m, bool set) + { return 0; } + +static inline s32 dev_pm_qos_requested_latency(struct device *dev) { return 0; } +static inline s32 dev_pm_qos_requested_flags(struct device *dev) { return 0; } #endif #endif -- cgit v1.2.3 From 7e6fdd4bad033fa2d73716377b184fa975b0d985 Mon Sep 17 00:00:00 2001 From: Rajagopal Venkat Date: Fri, 26 Oct 2012 01:50:09 +0200 Subject: PM / devfreq: Core updates to support devices which can idle Prepare devfreq core framework to support devices which can idle. When device idleness is detected perhaps through runtime-pm, need some mechanism to suspend devfreq load monitoring and resume back when device is online. Present code continues monitoring unless device is removed from devfreq core. This patch introduces following design changes, - use per device work instead of global work to monitor device load. This enables suspend/resume of device devfreq and reduces monitoring code complexity. - decouple delayed work based load monitoring logic from core by introducing helpers functions to be used by governors. This provides flexibility for governors either to use delayed work based monitoring functions or to implement their own mechanism. - devfreq core interacts with governors via events to perform specific actions. These events include start/stop devfreq. This sets ground for adding suspend/resume events. The devfreq apis are not modified and are kept intact. Signed-off-by: Rajagopal Venkat Acked-by: MyungJoo Ham Signed-off-by: Rafael J. Wysocki --- Documentation/ABI/testing/sysfs-class-devfreq | 8 - drivers/devfreq/devfreq.c | 442 +++++++++++--------------- drivers/devfreq/governor.h | 11 + drivers/devfreq/governor_performance.c | 16 +- drivers/devfreq/governor_powersave.c | 16 +- drivers/devfreq/governor_simpleondemand.c | 24 ++ drivers/devfreq/governor_userspace.c | 23 +- include/linux/devfreq.h | 34 +- 8 files changed, 278 insertions(+), 296 deletions(-) (limited to 'Documentation') diff --git a/Documentation/ABI/testing/sysfs-class-devfreq b/Documentation/ABI/testing/sysfs-class-devfreq index 23d78b5aab11..89283b1b0240 100644 --- a/Documentation/ABI/testing/sysfs-class-devfreq +++ b/Documentation/ABI/testing/sysfs-class-devfreq @@ -21,14 +21,6 @@ Description: The /sys/class/devfreq/.../cur_freq shows the current frequency of the corresponding devfreq object. -What: /sys/class/devfreq/.../central_polling -Date: September 2011 -Contact: MyungJoo Ham -Description: - The /sys/class/devfreq/.../central_polling shows whether - the devfreq ojbect is using devfreq-provided central - polling mechanism or not. - What: /sys/class/devfreq/.../polling_interval Date: September 2011 Contact: MyungJoo Ham diff --git a/drivers/devfreq/devfreq.c b/drivers/devfreq/devfreq.c index b146d76f04cf..1aaf1aeb1f1d 100644 --- a/drivers/devfreq/devfreq.c +++ b/drivers/devfreq/devfreq.c @@ -30,17 +30,11 @@ struct class *devfreq_class; /* - * devfreq_work periodically monitors every registered device. - * The minimum polling interval is one jiffy. The polling interval is - * determined by the minimum polling period among all polling devfreq - * devices. The resolution of polling interval is one jiffy. + * devfreq core provides delayed work based load monitoring helper + * functions. Governors can use these or can implement their own + * monitoring mechanism. */ -static bool polling; static struct workqueue_struct *devfreq_wq; -static struct delayed_work devfreq_work; - -/* wait removing if this is to be removed */ -static struct devfreq *wait_remove_device; /* The list of all device-devfreq */ static LIST_HEAD(devfreq_list); @@ -72,6 +66,8 @@ static struct devfreq *find_device_devfreq(struct device *dev) return ERR_PTR(-ENODEV); } +/* Load monitoring helper functions for governors use */ + /** * update_devfreq() - Reevaluate the device and configure frequency. * @devfreq: the devfreq instance. @@ -120,6 +116,152 @@ int update_devfreq(struct devfreq *devfreq) return err; } +/** + * devfreq_monitor() - Periodically poll devfreq objects. + * @work: the work struct used to run devfreq_monitor periodically. + * + */ +static void devfreq_monitor(struct work_struct *work) +{ + int err; + struct devfreq *devfreq = container_of(work, + struct devfreq, work.work); + + mutex_lock(&devfreq->lock); + err = update_devfreq(devfreq); + if (err) + dev_err(&devfreq->dev, "dvfs failed with (%d) error\n", err); + + queue_delayed_work(devfreq_wq, &devfreq->work, + msecs_to_jiffies(devfreq->profile->polling_ms)); + mutex_unlock(&devfreq->lock); +} + +/** + * devfreq_monitor_start() - Start load monitoring of devfreq instance + * @devfreq: the devfreq instance. + * + * Helper function for starting devfreq device load monitoing. By + * default delayed work based monitoring is supported. Function + * to be called from governor in response to DEVFREQ_GOV_START + * event when device is added to devfreq framework. + */ +void devfreq_monitor_start(struct devfreq *devfreq) +{ + INIT_DEFERRABLE_WORK(&devfreq->work, devfreq_monitor); + if (devfreq->profile->polling_ms) + queue_delayed_work(devfreq_wq, &devfreq->work, + msecs_to_jiffies(devfreq->profile->polling_ms)); +} + +/** + * devfreq_monitor_stop() - Stop load monitoring of a devfreq instance + * @devfreq: the devfreq instance. + * + * Helper function to stop devfreq device load monitoing. Function + * to be called from governor in response to DEVFREQ_GOV_STOP + * event when device is removed from devfreq framework. + */ +void devfreq_monitor_stop(struct devfreq *devfreq) +{ + cancel_delayed_work_sync(&devfreq->work); +} + +/** + * devfreq_monitor_suspend() - Suspend load monitoring of a devfreq instance + * @devfreq: the devfreq instance. + * + * Helper function to suspend devfreq device load monitoing. Function + * to be called from governor in response to DEVFREQ_GOV_SUSPEND + * event or when polling interval is set to zero. + * + * Note: Though this function is same as devfreq_monitor_stop(), + * intentionally kept separate to provide hooks for collecting + * transition statistics. + */ +void devfreq_monitor_suspend(struct devfreq *devfreq) +{ + mutex_lock(&devfreq->lock); + if (devfreq->stop_polling) { + mutex_unlock(&devfreq->lock); + return; + } + + devfreq->stop_polling = true; + mutex_unlock(&devfreq->lock); + cancel_delayed_work_sync(&devfreq->work); +} + +/** + * devfreq_monitor_resume() - Resume load monitoring of a devfreq instance + * @devfreq: the devfreq instance. + * + * Helper function to resume devfreq device load monitoing. Function + * to be called from governor in response to DEVFREQ_GOV_RESUME + * event or when polling interval is set to non-zero. + */ +void devfreq_monitor_resume(struct devfreq *devfreq) +{ + mutex_lock(&devfreq->lock); + if (!devfreq->stop_polling) + goto out; + + if (!delayed_work_pending(&devfreq->work) && + devfreq->profile->polling_ms) + queue_delayed_work(devfreq_wq, &devfreq->work, + msecs_to_jiffies(devfreq->profile->polling_ms)); + devfreq->stop_polling = false; + +out: + mutex_unlock(&devfreq->lock); +} + +/** + * devfreq_interval_update() - Update device devfreq monitoring interval + * @devfreq: the devfreq instance. + * @delay: new polling interval to be set. + * + * Helper function to set new load monitoring polling interval. Function + * to be called from governor in response to DEVFREQ_GOV_INTERVAL event. + */ +void devfreq_interval_update(struct devfreq *devfreq, unsigned int *delay) +{ + unsigned int cur_delay = devfreq->profile->polling_ms; + unsigned int new_delay = *delay; + + mutex_lock(&devfreq->lock); + devfreq->profile->polling_ms = new_delay; + + if (devfreq->stop_polling) + goto out; + + /* if new delay is zero, stop polling */ + if (!new_delay) { + mutex_unlock(&devfreq->lock); + cancel_delayed_work_sync(&devfreq->work); + return; + } + + /* if current delay is zero, start polling with new delay */ + if (!cur_delay) { + queue_delayed_work(devfreq_wq, &devfreq->work, + msecs_to_jiffies(devfreq->profile->polling_ms)); + goto out; + } + + /* if current delay is greater than new delay, restart polling */ + if (cur_delay > new_delay) { + mutex_unlock(&devfreq->lock); + cancel_delayed_work_sync(&devfreq->work); + mutex_lock(&devfreq->lock); + if (!devfreq->stop_polling) + queue_delayed_work(devfreq_wq, &devfreq->work, + msecs_to_jiffies(devfreq->profile->polling_ms)); + } +out: + mutex_unlock(&devfreq->lock); +} + /** * devfreq_notifier_call() - Notify that the device frequency requirements * has been changed out of devfreq framework. @@ -143,59 +285,32 @@ static int devfreq_notifier_call(struct notifier_block *nb, unsigned long type, } /** - * _remove_devfreq() - Remove devfreq from the device. + * _remove_devfreq() - Remove devfreq from the list and release its resources. * @devfreq: the devfreq struct * @skip: skip calling device_unregister(). - * - * Note that the caller should lock devfreq->lock before calling - * this. _remove_devfreq() will unlock it and free devfreq - * internally. devfreq_list_lock should be locked by the caller - * as well (not relased at return) - * - * Lock usage: - * devfreq->lock: locked before call. - * unlocked at return (and freed) - * devfreq_list_lock: locked before call. - * kept locked at return. - * if devfreq is centrally polled. - * - * Freed memory: - * devfreq */ static void _remove_devfreq(struct devfreq *devfreq, bool skip) { - if (!mutex_is_locked(&devfreq->lock)) { - WARN(true, "devfreq->lock must be locked by the caller.\n"); - return; - } - if (!devfreq->governor->no_central_polling && - !mutex_is_locked(&devfreq_list_lock)) { - WARN(true, "devfreq_list_lock must be locked by the caller.\n"); + mutex_lock(&devfreq_list_lock); + if (IS_ERR(find_device_devfreq(devfreq->dev.parent))) { + mutex_unlock(&devfreq_list_lock); + dev_warn(&devfreq->dev, "releasing devfreq which doesn't exist\n"); return; } + list_del(&devfreq->node); + mutex_unlock(&devfreq_list_lock); - if (devfreq->being_removed) - return; - - devfreq->being_removed = true; + devfreq->governor->event_handler(devfreq, DEVFREQ_GOV_STOP, NULL); if (devfreq->profile->exit) devfreq->profile->exit(devfreq->dev.parent); - if (devfreq->governor->exit) - devfreq->governor->exit(devfreq); - if (!skip && get_device(&devfreq->dev)) { device_unregister(&devfreq->dev); put_device(&devfreq->dev); } - if (!devfreq->governor->no_central_polling) - list_del(&devfreq->node); - - mutex_unlock(&devfreq->lock); mutex_destroy(&devfreq->lock); - kfree(devfreq); } @@ -210,130 +325,8 @@ static void _remove_devfreq(struct devfreq *devfreq, bool skip) static void devfreq_dev_release(struct device *dev) { struct devfreq *devfreq = to_devfreq(dev); - bool central_polling = !devfreq->governor->no_central_polling; - - /* - * If devfreq_dev_release() was called by device_unregister() of - * _remove_devfreq(), we cannot mutex_lock(&devfreq->lock) and - * being_removed is already set. This also partially checks the case - * where devfreq_dev_release() is called from a thread other than - * the one called _remove_devfreq(); however, this case is - * dealt completely with another following being_removed check. - * - * Because being_removed is never being - * unset, we do not need to worry about race conditions on - * being_removed. - */ - if (devfreq->being_removed) - return; - if (central_polling) - mutex_lock(&devfreq_list_lock); - - mutex_lock(&devfreq->lock); - - /* - * Check being_removed flag again for the case where - * devfreq_dev_release() was called in a thread other than the one - * possibly called _remove_devfreq(). - */ - if (devfreq->being_removed) { - mutex_unlock(&devfreq->lock); - goto out; - } - - /* devfreq->lock is unlocked and removed in _removed_devfreq() */ _remove_devfreq(devfreq, true); - -out: - if (central_polling) - mutex_unlock(&devfreq_list_lock); -} - -/** - * devfreq_monitor() - Periodically poll devfreq objects. - * @work: the work struct used to run devfreq_monitor periodically. - * - */ -static void devfreq_monitor(struct work_struct *work) -{ - static unsigned long last_polled_at; - struct devfreq *devfreq, *tmp; - int error; - unsigned long jiffies_passed; - unsigned long next_jiffies = ULONG_MAX, now = jiffies; - struct device *dev; - - /* Initially last_polled_at = 0, polling every device at bootup */ - jiffies_passed = now - last_polled_at; - last_polled_at = now; - if (jiffies_passed == 0) - jiffies_passed = 1; - - mutex_lock(&devfreq_list_lock); - list_for_each_entry_safe(devfreq, tmp, &devfreq_list, node) { - mutex_lock(&devfreq->lock); - dev = devfreq->dev.parent; - - /* Do not remove tmp for a while */ - wait_remove_device = tmp; - - if (devfreq->governor->no_central_polling || - devfreq->next_polling == 0) { - mutex_unlock(&devfreq->lock); - continue; - } - mutex_unlock(&devfreq_list_lock); - - /* - * Reduce more next_polling if devfreq_wq took an extra - * delay. (i.e., CPU has been idled.) - */ - if (devfreq->next_polling <= jiffies_passed) { - error = update_devfreq(devfreq); - - /* Remove a devfreq with an error. */ - if (error && error != -EAGAIN) { - - dev_err(dev, "Due to update_devfreq error(%d), devfreq(%s) is removed from the device\n", - error, devfreq->governor->name); - - /* - * Unlock devfreq before locking the list - * in order to avoid deadlock with - * find_device_devfreq or others - */ - mutex_unlock(&devfreq->lock); - mutex_lock(&devfreq_list_lock); - /* Check if devfreq is already removed */ - if (IS_ERR(find_device_devfreq(dev))) - continue; - mutex_lock(&devfreq->lock); - /* This unlocks devfreq->lock and free it */ - _remove_devfreq(devfreq, false); - continue; - } - devfreq->next_polling = devfreq->polling_jiffies; - } else { - devfreq->next_polling -= jiffies_passed; - } - - if (devfreq->next_polling) - next_jiffies = (next_jiffies > devfreq->next_polling) ? - devfreq->next_polling : next_jiffies; - - mutex_unlock(&devfreq->lock); - mutex_lock(&devfreq_list_lock); - } - wait_remove_device = NULL; - mutex_unlock(&devfreq_list_lock); - - if (next_jiffies > 0 && next_jiffies < ULONG_MAX) { - polling = true; - queue_delayed_work(devfreq_wq, &devfreq_work, next_jiffies); - } else { - polling = false; - } } /** @@ -357,16 +350,13 @@ struct devfreq *devfreq_add_device(struct device *dev, return ERR_PTR(-EINVAL); } - - if (!governor->no_central_polling) { - mutex_lock(&devfreq_list_lock); - devfreq = find_device_devfreq(dev); - mutex_unlock(&devfreq_list_lock); - if (!IS_ERR(devfreq)) { - dev_err(dev, "%s: Unable to create devfreq for the device. It already has one.\n", __func__); - err = -EINVAL; - goto err_out; - } + mutex_lock(&devfreq_list_lock); + devfreq = find_device_devfreq(dev); + mutex_unlock(&devfreq_list_lock); + if (!IS_ERR(devfreq)) { + dev_err(dev, "%s: Unable to create devfreq for the device. It already has one.\n", __func__); + err = -EINVAL; + goto err_out; } devfreq = kzalloc(sizeof(struct devfreq), GFP_KERNEL); @@ -386,48 +376,41 @@ struct devfreq *devfreq_add_device(struct device *dev, devfreq->governor = governor; devfreq->previous_freq = profile->initial_freq; devfreq->data = data; - devfreq->next_polling = devfreq->polling_jiffies - = msecs_to_jiffies(devfreq->profile->polling_ms); devfreq->nb.notifier_call = devfreq_notifier_call; dev_set_name(&devfreq->dev, dev_name(dev)); err = device_register(&devfreq->dev); if (err) { put_device(&devfreq->dev); + mutex_unlock(&devfreq->lock); goto err_dev; } - if (governor->init) - err = governor->init(devfreq); - if (err) - goto err_init; - mutex_unlock(&devfreq->lock); - if (governor->no_central_polling) - goto out; - mutex_lock(&devfreq_list_lock); - list_add(&devfreq->node, &devfreq_list); + mutex_unlock(&devfreq_list_lock); - if (devfreq_wq && devfreq->next_polling && !polling) { - polling = true; - queue_delayed_work(devfreq_wq, &devfreq_work, - devfreq->next_polling); + err = devfreq->governor->event_handler(devfreq, + DEVFREQ_GOV_START, NULL); + if (err) { + dev_err(dev, "%s: Unable to start governor for the device\n", + __func__); + goto err_init; } - mutex_unlock(&devfreq_list_lock); -out: + return devfreq; err_init: + list_del(&devfreq->node); device_unregister(&devfreq->dev); err_dev: - mutex_unlock(&devfreq->lock); kfree(devfreq); err_out: return ERR_PTR(err); } +EXPORT_SYMBOL(devfreq_add_device); /** * devfreq_remove_device() - Remove devfreq feature from a device. @@ -435,30 +418,14 @@ err_out: */ int devfreq_remove_device(struct devfreq *devfreq) { - bool central_polling; - if (!devfreq) return -EINVAL; - central_polling = !devfreq->governor->no_central_polling; - - if (central_polling) { - mutex_lock(&devfreq_list_lock); - while (wait_remove_device == devfreq) { - mutex_unlock(&devfreq_list_lock); - schedule(); - mutex_lock(&devfreq_list_lock); - } - } - - mutex_lock(&devfreq->lock); - _remove_devfreq(devfreq, false); /* it unlocks devfreq->lock */ - - if (central_polling) - mutex_unlock(&devfreq_list_lock); + _remove_devfreq(devfreq, false); return 0; } +EXPORT_SYMBOL(devfreq_remove_device); static ssize_t show_governor(struct device *dev, struct device_attribute *attr, char *buf) @@ -490,35 +457,13 @@ static ssize_t store_polling_interval(struct device *dev, if (ret != 1) goto out; - mutex_lock(&df->lock); - df->profile->polling_ms = value; - df->next_polling = df->polling_jiffies - = msecs_to_jiffies(value); - mutex_unlock(&df->lock); - + df->governor->event_handler(df, DEVFREQ_GOV_INTERVAL, &value); ret = count; - if (df->governor->no_central_polling) - goto out; - - mutex_lock(&devfreq_list_lock); - if (df->next_polling > 0 && !polling) { - polling = true; - queue_delayed_work(devfreq_wq, &devfreq_work, - df->next_polling); - } - mutex_unlock(&devfreq_list_lock); out: return ret; } -static ssize_t show_central_polling(struct device *dev, - struct device_attribute *attr, char *buf) -{ - return sprintf(buf, "%d\n", - !to_devfreq(dev)->governor->no_central_polling); -} - static ssize_t store_min_freq(struct device *dev, struct device_attribute *attr, const char *buf, size_t count) { @@ -590,7 +535,6 @@ static ssize_t show_max_freq(struct device *dev, struct device_attribute *attr, static struct device_attribute devfreq_attrs[] = { __ATTR(governor, S_IRUGO, show_governor, NULL), __ATTR(cur_freq, S_IRUGO, show_freq, NULL), - __ATTR(central_polling, S_IRUGO, show_central_polling, NULL), __ATTR(polling_interval, S_IRUGO | S_IWUSR, show_polling_interval, store_polling_interval), __ATTR(min_freq, S_IRUGO | S_IWUSR, show_min_freq, store_min_freq), @@ -598,23 +542,6 @@ static struct device_attribute devfreq_attrs[] = { { }, }; -/** - * devfreq_start_polling() - Initialize data structure for devfreq framework and - * start polling registered devfreq devices. - */ -static int __init devfreq_start_polling(void) -{ - mutex_lock(&devfreq_list_lock); - polling = false; - devfreq_wq = create_freezable_workqueue("devfreq_wq"); - INIT_DEFERRABLE_WORK(&devfreq_work, devfreq_monitor); - mutex_unlock(&devfreq_list_lock); - - devfreq_monitor(&devfreq_work.work); - return 0; -} -late_initcall(devfreq_start_polling); - static int __init devfreq_init(void) { devfreq_class = class_create(THIS_MODULE, "devfreq"); @@ -622,7 +549,15 @@ static int __init devfreq_init(void) pr_err("%s: couldn't create class\n", __FILE__); return PTR_ERR(devfreq_class); } + + devfreq_wq = create_freezable_workqueue("devfreq_wq"); + if (IS_ERR(devfreq_wq)) { + class_destroy(devfreq_class); + pr_err("%s: couldn't create workqueue\n", __FILE__); + return PTR_ERR(devfreq_wq); + } devfreq_class->dev_attrs = devfreq_attrs; + return 0; } subsys_initcall(devfreq_init); @@ -630,6 +565,7 @@ subsys_initcall(devfreq_init); static void __exit devfreq_exit(void) { class_destroy(devfreq_class); + destroy_workqueue(devfreq_wq); } module_exit(devfreq_exit); diff --git a/drivers/devfreq/governor.h b/drivers/devfreq/governor.h index ea7f13c58ded..bb3aff32d627 100644 --- a/drivers/devfreq/governor.h +++ b/drivers/devfreq/governor.h @@ -18,7 +18,18 @@ #define to_devfreq(DEV) container_of((DEV), struct devfreq, dev) +/* Devfreq events */ +#define DEVFREQ_GOV_START 0x1 +#define DEVFREQ_GOV_STOP 0x2 +#define DEVFREQ_GOV_INTERVAL 0x3 + /* Caution: devfreq->lock must be locked before calling update_devfreq */ extern int update_devfreq(struct devfreq *devfreq); +extern void devfreq_monitor_start(struct devfreq *devfreq); +extern void devfreq_monitor_stop(struct devfreq *devfreq); +extern void devfreq_monitor_suspend(struct devfreq *devfreq); +extern void devfreq_monitor_resume(struct devfreq *devfreq); +extern void devfreq_interval_update(struct devfreq *devfreq, + unsigned int *delay); #endif /* _GOVERNOR_H */ diff --git a/drivers/devfreq/governor_performance.c b/drivers/devfreq/governor_performance.c index af75ddd4f158..eea3f9bd7894 100644 --- a/drivers/devfreq/governor_performance.c +++ b/drivers/devfreq/governor_performance.c @@ -26,14 +26,22 @@ static int devfreq_performance_func(struct devfreq *df, return 0; } -static int performance_init(struct devfreq *devfreq) +static int devfreq_performance_handler(struct devfreq *devfreq, + unsigned int event, void *data) { - return update_devfreq(devfreq); + int ret = 0; + + if (event == DEVFREQ_GOV_START) { + mutex_lock(&devfreq->lock); + ret = update_devfreq(devfreq); + mutex_unlock(&devfreq->lock); + } + + return ret; } const struct devfreq_governor devfreq_performance = { .name = "performance", - .init = performance_init, .get_target_freq = devfreq_performance_func, - .no_central_polling = true, + .event_handler = devfreq_performance_handler, }; diff --git a/drivers/devfreq/governor_powersave.c b/drivers/devfreq/governor_powersave.c index fec0cdbd2477..2868d98ed3e2 100644 --- a/drivers/devfreq/governor_powersave.c +++ b/drivers/devfreq/governor_powersave.c @@ -23,14 +23,22 @@ static int devfreq_powersave_func(struct devfreq *df, return 0; } -static int powersave_init(struct devfreq *devfreq) +static int devfreq_powersave_handler(struct devfreq *devfreq, + unsigned int event, void *data) { - return update_devfreq(devfreq); + int ret = 0; + + if (event == DEVFREQ_GOV_START) { + mutex_lock(&devfreq->lock); + ret = update_devfreq(devfreq); + mutex_unlock(&devfreq->lock); + } + + return ret; } const struct devfreq_governor devfreq_powersave = { .name = "powersave", - .init = powersave_init, .get_target_freq = devfreq_powersave_func, - .no_central_polling = true, + .event_handler = devfreq_powersave_handler, }; diff --git a/drivers/devfreq/governor_simpleondemand.c b/drivers/devfreq/governor_simpleondemand.c index a2e3eae79011..3716a659122b 100644 --- a/drivers/devfreq/governor_simpleondemand.c +++ b/drivers/devfreq/governor_simpleondemand.c @@ -12,6 +12,7 @@ #include #include #include +#include "governor.h" /* Default constants for DevFreq-Simple-Ondemand (DFSO) */ #define DFSO_UPTHRESHOLD (90) @@ -88,7 +89,30 @@ static int devfreq_simple_ondemand_func(struct devfreq *df, return 0; } +static int devfreq_simple_ondemand_handler(struct devfreq *devfreq, + unsigned int event, void *data) +{ + switch (event) { + case DEVFREQ_GOV_START: + devfreq_monitor_start(devfreq); + break; + + case DEVFREQ_GOV_STOP: + devfreq_monitor_stop(devfreq); + break; + + case DEVFREQ_GOV_INTERVAL: + devfreq_interval_update(devfreq, (unsigned int *)data); + break; + default: + break; + } + + return 0; +} + const struct devfreq_governor devfreq_simple_ondemand = { .name = "simple_ondemand", .get_target_freq = devfreq_simple_ondemand_func, + .event_handler = devfreq_simple_ondemand_handler, }; diff --git a/drivers/devfreq/governor_userspace.c b/drivers/devfreq/governor_userspace.c index 0681246fc89d..7067555bd444 100644 --- a/drivers/devfreq/governor_userspace.c +++ b/drivers/devfreq/governor_userspace.c @@ -116,10 +116,27 @@ static void userspace_exit(struct devfreq *devfreq) devfreq->data = NULL; } +static int devfreq_userspace_handler(struct devfreq *devfreq, + unsigned int event, void *data) +{ + int ret = 0; + + switch (event) { + case DEVFREQ_GOV_START: + ret = userspace_init(devfreq); + break; + case DEVFREQ_GOV_STOP: + userspace_exit(devfreq); + break; + default: + break; + } + + return ret; +} + const struct devfreq_governor devfreq_userspace = { .name = "userspace", .get_target_freq = devfreq_userspace_func, - .init = userspace_init, - .exit = userspace_exit, - .no_central_polling = true, + .event_handler = devfreq_userspace_handler, }; diff --git a/include/linux/devfreq.h b/include/linux/devfreq.h index 281c72a3b9d5..9cdffde74bb5 100644 --- a/include/linux/devfreq.h +++ b/include/linux/devfreq.h @@ -91,25 +91,18 @@ struct devfreq_dev_profile { * status of the device (load = busy_time / total_time). * If no_central_polling is set, this callback is called * only with update_devfreq() notified by OPP. - * @init Called when the devfreq is being attached to a device - * @exit Called when the devfreq is being removed from a - * device. Governor should stop any internal routines - * before return because related data may be - * freed after exit(). - * @no_central_polling Do not use devfreq's central polling mechanism. - * When this is set, devfreq will not call - * get_target_freq with devfreq_monitor(). However, - * devfreq will call get_target_freq with - * devfreq_update() notified by OPP framework. + * @event_handler Callback for devfreq core framework to notify events + * to governors. Events include per device governor + * init and exit, opp changes out of devfreq, suspend + * and resume of per device devfreq during device idle. * * Note that the callbacks are called with devfreq->lock locked by devfreq. */ struct devfreq_governor { const char name[DEVFREQ_NAME_LEN]; int (*get_target_freq)(struct devfreq *this, unsigned long *freq); - int (*init)(struct devfreq *this); - void (*exit)(struct devfreq *this); - const bool no_central_polling; + int (*event_handler)(struct devfreq *devfreq, + unsigned int event, void *data); }; /** @@ -124,18 +117,13 @@ struct devfreq_governor { * @nb notifier block used to notify devfreq object that it should * reevaluate operable frequencies. Devfreq users may use * devfreq.nb to the corresponding register notifier call chain. - * @polling_jiffies interval in jiffies. + * @work delayed work for load monitoring. * @previous_freq previously configured frequency value. - * @next_polling the number of remaining jiffies to poll with - * "devfreq_monitor" executions to reevaluate - * frequency/voltage of the device. Set by - * profile's polling_ms interval. * @data Private data of the governor. The devfreq framework does not * touch this. - * @being_removed a flag to mark that this object is being removed in - * order to prevent trying to remove the object multiple times. * @min_freq Limit minimum frequency requested by user (0: none) * @max_freq Limit maximum frequency requested by user (0: none) + * @stop_polling devfreq polling status of a device. * * This structure stores the devfreq information for a give device. * @@ -153,17 +141,15 @@ struct devfreq { struct devfreq_dev_profile *profile; const struct devfreq_governor *governor; struct notifier_block nb; + struct delayed_work work; - unsigned long polling_jiffies; unsigned long previous_freq; - unsigned int next_polling; void *data; /* private data for governors */ - bool being_removed; - unsigned long min_freq; unsigned long max_freq; + bool stop_polling; }; #if defined(CONFIG_PM_DEVFREQ) -- cgit v1.2.3 From 7f98a905dca6e4f144cdd4462edeac00c2bdc379 Mon Sep 17 00:00:00 2001 From: Rajagopal Venkat Date: Fri, 26 Oct 2012 01:50:26 +0200 Subject: PM / devfreq: Add current freq callback in device profile Devfreq returns governor predicted frequency as current frequency via sysfs interface. But device may not support all frequencies that governor predicts. So add a callback in device profile to get current freq from driver. Also add a new sysfs node to expose governor predicted next target frequency. Signed-off-by: Rajagopal Venkat Acked-by: MyungJoo Ham Signed-off-by: Rafael J. Wysocki --- Documentation/ABI/testing/sysfs-class-devfreq | 11 ++++++++++- drivers/devfreq/devfreq.c | 14 ++++++++++++++ include/linux/devfreq.h | 3 +++ 3 files changed, 27 insertions(+), 1 deletion(-) (limited to 'Documentation') diff --git a/Documentation/ABI/testing/sysfs-class-devfreq b/Documentation/ABI/testing/sysfs-class-devfreq index 89283b1b0240..e6cf08e6734d 100644 --- a/Documentation/ABI/testing/sysfs-class-devfreq +++ b/Documentation/ABI/testing/sysfs-class-devfreq @@ -19,7 +19,16 @@ Date: September 2011 Contact: MyungJoo Ham Description: The /sys/class/devfreq/.../cur_freq shows the current - frequency of the corresponding devfreq object. + frequency of the corresponding devfreq object. Same as + target_freq when get_cur_freq() is not implemented by + devfreq driver. + +What: /sys/class/devfreq/.../target_freq +Date: September 2012 +Contact: Rajagopal Venkat +Description: + The /sys/class/devfreq/.../target_freq shows the next governor + predicted target frequency of the corresponding devfreq object. What: /sys/class/devfreq/.../polling_interval Date: September 2011 diff --git a/drivers/devfreq/devfreq.c b/drivers/devfreq/devfreq.c index 999600da21c7..f2f8a976c465 100644 --- a/drivers/devfreq/devfreq.c +++ b/drivers/devfreq/devfreq.c @@ -463,6 +463,19 @@ static ssize_t show_governor(struct device *dev, static ssize_t show_freq(struct device *dev, struct device_attribute *attr, char *buf) +{ + unsigned long freq; + struct devfreq *devfreq = to_devfreq(dev); + + if (devfreq->profile->get_cur_freq && + !devfreq->profile->get_cur_freq(devfreq->dev.parent, &freq)) + return sprintf(buf, "%lu\n", freq); + + return sprintf(buf, "%lu\n", devfreq->previous_freq); +} + +static ssize_t show_target_freq(struct device *dev, + struct device_attribute *attr, char *buf) { return sprintf(buf, "%lu\n", to_devfreq(dev)->previous_freq); } @@ -563,6 +576,7 @@ static ssize_t show_max_freq(struct device *dev, struct device_attribute *attr, static struct device_attribute devfreq_attrs[] = { __ATTR(governor, S_IRUGO, show_governor, NULL), __ATTR(cur_freq, S_IRUGO, show_freq, NULL), + __ATTR(target_freq, S_IRUGO, show_target_freq, NULL), __ATTR(polling_interval, S_IRUGO | S_IWUSR, show_polling_interval, store_polling_interval), __ATTR(min_freq, S_IRUGO | S_IWUSR, show_min_freq, store_min_freq), diff --git a/include/linux/devfreq.h b/include/linux/devfreq.h index ee243a3229b8..7e2e2ea4a70f 100644 --- a/include/linux/devfreq.h +++ b/include/linux/devfreq.h @@ -66,6 +66,8 @@ struct devfreq_dev_status { * explained above with "DEVFREQ_FLAG_*" macros. * @get_dev_status The device should provide the current performance * status to devfreq, which is used by governors. + * @get_cur_freq The device should provide the current frequency + * at which it is operating. * @exit An optional callback that is called when devfreq * is removing the devfreq object due to error or * from devfreq_remove_device() call. If the user @@ -79,6 +81,7 @@ struct devfreq_dev_profile { int (*target)(struct device *dev, unsigned long *freq, u32 flags); int (*get_dev_status)(struct device *dev, struct devfreq_dev_status *stat); + int (*get_cur_freq)(struct device *dev, unsigned long *freq); void (*exit)(struct device *dev); }; -- cgit v1.2.3 From bb74ac23b10820d8722c3e1f4add9ef59e703f63 Mon Sep 17 00:00:00 2001 From: Yasuaki Ishimatsu Date: Fri, 16 Nov 2012 02:56:59 +0100 Subject: ACPI: create _SUN sysfs file _SUN method provides the slot unique-ID in the ACPI namespace. And The value is written in Advanced Configuration and Power Interface Specification as follows: "The _SUN value is required to be unique among the slots ofthe same type. It is also recommended that this number match the slot number printed on the physical slot whenever possible." So if we can know the value, we can identify the physical position of the slot in the system. The patch creates "sun" file in sysfs for identifying physical position of the slot. Signed-off-by: Yasuaki Ishimatsu Reviewed-by: Toshi Kani Signed-off-by: Rafael J. Wysocki --- Documentation/ABI/testing/sysfs-devices-sun | 14 ++++++++++++++ drivers/acpi/scan.c | 24 ++++++++++++++++++++++++ include/acpi/acpi_bus.h | 1 + 3 files changed, 39 insertions(+) create mode 100644 Documentation/ABI/testing/sysfs-devices-sun (limited to 'Documentation') diff --git a/Documentation/ABI/testing/sysfs-devices-sun b/Documentation/ABI/testing/sysfs-devices-sun new file mode 100644 index 000000000000..86be9848a77e --- /dev/null +++ b/Documentation/ABI/testing/sysfs-devices-sun @@ -0,0 +1,14 @@ +Whatt: /sys/devices/.../sun +Date: October 2012 +Contact: Yasuaki Ishimatsu +Description: + The file contains a Slot-unique ID which provided by the _SUN + method in the ACPI namespace. The value is written in Advanced + Configuration and Power Interface Specification as follows: + + "The _SUN value is required to be unique among the slots of + the same type. It is also recommended that this number match + the slot number printed on the physical slot whenever possible." + + So reading the sysfs file, we can identify a physical position + of the slot in the system. diff --git a/drivers/acpi/scan.c b/drivers/acpi/scan.c index 9d43532d69b1..d0b38ab47ab5 100644 --- a/drivers/acpi/scan.c +++ b/drivers/acpi/scan.c @@ -283,11 +283,21 @@ static ssize_t description_show(struct device *dev, } static DEVICE_ATTR(description, 0444, description_show, NULL); +static ssize_t +acpi_device_sun_show(struct device *dev, struct device_attribute *attr, + char *buf) { + struct acpi_device *acpi_dev = to_acpi_device(dev); + + return sprintf(buf, "%lu\n", acpi_dev->pnp.sun); +} +static DEVICE_ATTR(sun, 0444, acpi_device_sun_show, NULL); + static int acpi_device_setup_files(struct acpi_device *dev) { struct acpi_buffer buffer = {ACPI_ALLOCATE_BUFFER, NULL}; acpi_status status; acpi_handle temp; + unsigned long long sun; int result = 0; /* @@ -329,6 +339,16 @@ static int acpi_device_setup_files(struct acpi_device *dev) if (dev->pnp.unique_id) result = device_create_file(&dev->dev, &dev_attr_uid); + status = acpi_evaluate_integer(dev->handle, "_SUN", NULL, &sun); + if (ACPI_SUCCESS(status)) { + dev->pnp.sun = (unsigned long)sun; + result = device_create_file(&dev->dev, &dev_attr_sun); + if (result) + goto end; + } else { + dev->pnp.sun = (unsigned long)-1; + } + /* * If device has _EJ0, 'eject' file is created that is used to trigger * hot-removal function from userland. @@ -360,6 +380,10 @@ static void acpi_device_remove_files(struct acpi_device *dev) if (ACPI_SUCCESS(status)) device_remove_file(&dev->dev, &dev_attr_eject); + status = acpi_get_handle(dev->handle, "_SUN", &temp); + if (ACPI_SUCCESS(status)) + device_remove_file(&dev->dev, &dev_attr_sun); + if (dev->pnp.unique_id) device_remove_file(&dev->dev, &dev_attr_uid); if (dev->flags.bus_address) diff --git a/include/acpi/acpi_bus.h b/include/acpi/acpi_bus.h index e8b2877aea33..6f385e5909db 100644 --- a/include/acpi/acpi_bus.h +++ b/include/acpi/acpi_bus.h @@ -179,6 +179,7 @@ struct acpi_device_pnp { acpi_device_name device_name; /* Driver-determined */ acpi_device_class device_class; /* " */ union acpi_object *str_obj; /* unicode string for _STR method */ + unsigned long sun; /* _SUN */ }; #define acpi_device_bid(d) ((d)->pnp.bus_id) -- cgit v1.2.3 From d287de855f97c56ca7146ff627e652bd7cd64f3f Mon Sep 17 00:00:00 2001 From: Nishanth Menon Date: Thu, 25 Oct 2012 19:48:59 -0500 Subject: PM / devfreq: Add sysfs node to expose available frequencies devfreq governors such as ondemand are controlled by a min and max frequency, while governors like userspace governor allow us to set a specific frequency. However, for the same specific device, depending on the SoC, the available frequencies can vary. So expose the available frequencies as a snapshot over sysfs to allow informed decisions. This was inspired by cpufreq framework's equivalent for similar usage sysfs node: scaling_available_frequencies. Cc: Rajagopal Venkat Cc: MyungJoo Ham Cc: Kyungmin Park Cc: "Rafael J. Wysocki" Cc: Kevin Hilman Cc: linux-pm@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Nishanth Menon Signed-off-by: MyungJoo Ham --- Documentation/ABI/testing/sysfs-class-devfreq | 9 ++++++++ drivers/devfreq/devfreq.c | 32 +++++++++++++++++++++++++++ 2 files changed, 41 insertions(+) (limited to 'Documentation') diff --git a/Documentation/ABI/testing/sysfs-class-devfreq b/Documentation/ABI/testing/sysfs-class-devfreq index e6cf08e6734d..e672ccb02e7f 100644 --- a/Documentation/ABI/testing/sysfs-class-devfreq +++ b/Documentation/ABI/testing/sysfs-class-devfreq @@ -51,3 +51,12 @@ Description: The /sys/class/devfreq/.../userspace/set_freq shows and sets the requested frequency for the devfreq object if userspace governor is in effect. + +What: /sys/class/devfreq/.../available_frequencies +Date: October 2012 +Contact: Nishanth Menon +Description: + The /sys/class/devfreq/.../available_frequencies shows + the available frequencies of the corresponding devfreq object. + This is a snapshot of available frequencies and not limited + by the min/max frequency restrictions. diff --git a/drivers/devfreq/devfreq.c b/drivers/devfreq/devfreq.c index 789af4ff5c9c..c44e562bdfe0 100644 --- a/drivers/devfreq/devfreq.c +++ b/drivers/devfreq/devfreq.c @@ -570,9 +570,41 @@ static ssize_t show_max_freq(struct device *dev, struct device_attribute *attr, return sprintf(buf, "%lu\n", to_devfreq(dev)->max_freq); } +static ssize_t show_available_freqs(struct device *d, + struct device_attribute *attr, + char *buf) +{ + struct devfreq *df = to_devfreq(d); + struct device *dev = df->dev.parent; + struct opp *opp; + ssize_t count = 0; + unsigned long freq = 0; + + rcu_read_lock(); + do { + opp = opp_find_freq_ceil(dev, &freq); + if (IS_ERR(opp)) + break; + + count += scnprintf(&buf[count], (PAGE_SIZE - count - 2), + "%lu ", freq); + freq++; + } while (1); + rcu_read_unlock(); + + /* Truncate the trailing space */ + if (count) + count--; + + count += sprintf(&buf[count], "\n"); + + return count; +} + static struct device_attribute devfreq_attrs[] = { __ATTR(governor, S_IRUGO, show_governor, NULL), __ATTR(cur_freq, S_IRUGO, show_freq, NULL), + __ATTR(available_frequencies, S_IRUGO, show_available_freqs, NULL), __ATTR(target_freq, S_IRUGO, show_target_freq, NULL), __ATTR(polling_interval, S_IRUGO | S_IWUSR, show_polling_interval, store_polling_interval), -- cgit v1.2.3 From e552bbaf5b987f57c43e6981a452b8a3c700b1ae Mon Sep 17 00:00:00 2001 From: Jonghwa Lee Date: Thu, 23 Aug 2012 20:00:46 +0900 Subject: PM / devfreq: Add sysfs node for representing frequency transition information. This patch adds sysfs node which can be used to get information of frequency transition. It represents transition table which contains total number of transition of each freqeuncy state and time spent. It is inspired CPUFREQ's status driver. Signed-off-by: Jonghwa Lee [Added Documentation/ABI entry, updated kernel-doc, and resolved merge conflict] Signed-off-by: MyungJoo Ham --- Documentation/ABI/testing/sysfs-class-devfreq | 11 +++ drivers/devfreq/devfreq.c | 101 ++++++++++++++++++++++++++ include/linux/devfreq.h | 15 ++++ 3 files changed, 127 insertions(+) (limited to 'Documentation') diff --git a/Documentation/ABI/testing/sysfs-class-devfreq b/Documentation/ABI/testing/sysfs-class-devfreq index e672ccb02e7f..40f98a9428dc 100644 --- a/Documentation/ABI/testing/sysfs-class-devfreq +++ b/Documentation/ABI/testing/sysfs-class-devfreq @@ -44,6 +44,17 @@ Description: (/sys/class/devfreq/.../central_polling is 0), this value may be useless. +What: /sys/class/devfreq/.../trans_stat +Date: October 2012 +Contact: MyungJoo Ham +Descrtiption: + This ABI shows the statistics of devfreq behavior on a + specific device. It shows the time spent in each state and + the number of transitions between states. + In order to activate this ABI, the devfreq target device + driver should provide the list of available frequencies + with its profile. + What: /sys/class/devfreq/.../userspace/set_freq Date: September 2011 Contact: MyungJoo Ham diff --git a/drivers/devfreq/devfreq.c b/drivers/devfreq/devfreq.c index c44e562bdfe0..bf6de38190cf 100644 --- a/drivers/devfreq/devfreq.c +++ b/drivers/devfreq/devfreq.c @@ -66,6 +66,51 @@ static struct devfreq *find_device_devfreq(struct device *dev) return ERR_PTR(-ENODEV); } +/** + * devfreq_get_freq_level() - Lookup freq_table for the frequency + * @devfreq: the devfreq instance + * @freq: the target frequency + */ +static int devfreq_get_freq_level(struct devfreq *devfreq, unsigned long freq) +{ + int lev; + + for (lev = 0; lev < devfreq->profile->max_state; lev++) + if (freq == devfreq->profile->freq_table[lev]) + return lev; + + return -EINVAL; +} + +/** + * devfreq_update_status() - Update statistics of devfreq behavior + * @devfreq: the devfreq instance + * @freq: the update target frequency + */ +static int devfreq_update_status(struct devfreq *devfreq, unsigned long freq) +{ + int lev, prev_lev; + unsigned long cur_time; + + lev = devfreq_get_freq_level(devfreq, freq); + if (lev < 0) + return lev; + + cur_time = jiffies; + devfreq->time_in_state[lev] += + cur_time - devfreq->last_stat_updated; + if (freq != devfreq->previous_freq) { + prev_lev = devfreq_get_freq_level(devfreq, + devfreq->previous_freq); + devfreq->trans_table[(prev_lev * + devfreq->profile->max_state) + lev]++; + devfreq->total_trans++; + } + devfreq->last_stat_updated = cur_time; + + return 0; +} + /* Load monitoring helper functions for governors use */ /** @@ -112,6 +157,11 @@ int update_devfreq(struct devfreq *devfreq) if (err) return err; + if (devfreq->profile->freq_table) + if (devfreq_update_status(devfreq, freq)) + dev_err(&devfreq->dev, + "Couldn't update frequency transition information.\n"); + devfreq->previous_freq = freq; return err; } @@ -378,6 +428,15 @@ struct devfreq *devfreq_add_device(struct device *dev, devfreq->data = data; devfreq->nb.notifier_call = devfreq_notifier_call; + devfreq->trans_table = devm_kzalloc(dev, sizeof(unsigned int) * + devfreq->profile->max_state * + devfreq->profile->max_state, + GFP_KERNEL); + devfreq->time_in_state = devm_kzalloc(dev, sizeof(unsigned int) * + devfreq->profile->max_state, + GFP_KERNEL); + devfreq->last_stat_updated = jiffies; + dev_set_name(&devfreq->dev, dev_name(dev)); err = device_register(&devfreq->dev); if (err) { @@ -601,6 +660,47 @@ static ssize_t show_available_freqs(struct device *d, return count; } +static ssize_t show_trans_table(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct devfreq *devfreq = to_devfreq(dev); + ssize_t len; + int i, j, err; + unsigned int max_state = devfreq->profile->max_state; + + err = devfreq_update_status(devfreq, devfreq->previous_freq); + if (err) + return 0; + + len = sprintf(buf, " From : To\n"); + len += sprintf(buf + len, " :"); + for (i = 0; i < max_state; i++) + len += sprintf(buf + len, "%8u", + devfreq->profile->freq_table[i]); + + len += sprintf(buf + len, " time(ms)\n"); + + for (i = 0; i < max_state; i++) { + if (devfreq->profile->freq_table[i] + == devfreq->previous_freq) { + len += sprintf(buf + len, "*"); + } else { + len += sprintf(buf + len, " "); + } + len += sprintf(buf + len, "%8u:", + devfreq->profile->freq_table[i]); + for (j = 0; j < max_state; j++) + len += sprintf(buf + len, "%8u", + devfreq->trans_table[(i * max_state) + j]); + len += sprintf(buf + len, "%10u\n", + jiffies_to_msecs(devfreq->time_in_state[i])); + } + + len += sprintf(buf + len, "Total transition : %u\n", + devfreq->total_trans); + return len; +} + static struct device_attribute devfreq_attrs[] = { __ATTR(governor, S_IRUGO, show_governor, NULL), __ATTR(cur_freq, S_IRUGO, show_freq, NULL), @@ -610,6 +710,7 @@ static struct device_attribute devfreq_attrs[] = { store_polling_interval), __ATTR(min_freq, S_IRUGO | S_IWUSR, show_min_freq, store_min_freq), __ATTR(max_freq, S_IRUGO | S_IWUSR, show_max_freq, store_max_freq), + __ATTR(trans_stat, S_IRUGO, show_trans_table, NULL), { }, }; diff --git a/include/linux/devfreq.h b/include/linux/devfreq.h index 1461fb2355ad..bc35c4aee6a3 100644 --- a/include/linux/devfreq.h +++ b/include/linux/devfreq.h @@ -73,6 +73,8 @@ struct devfreq_dev_status { * from devfreq_remove_device() call. If the user * has registered devfreq->nb at a notifier-head, * this is the time to unregister it. + * @freq_table: Optional list of frequencies to support statistics. + * @max_state: The size of freq_table. */ struct devfreq_dev_profile { unsigned long initial_freq; @@ -83,6 +85,9 @@ struct devfreq_dev_profile { struct devfreq_dev_status *stat); int (*get_cur_freq)(struct device *dev, unsigned long *freq); void (*exit)(struct device *dev); + + unsigned int *freq_table; + unsigned int max_state; }; /** @@ -127,6 +132,10 @@ struct devfreq_governor { * @min_freq: Limit minimum frequency requested by user (0: none) * @max_freq: Limit maximum frequency requested by user (0: none) * @stop_polling: devfreq polling status of a device. + * @total_trans: Number of devfreq transitions + * @trans_table: Statistics of devfreq transitions + * @time_in_state: Statistics of devfreq states + * @last_stat_updated: The last time stat updated * * This structure stores the devfreq information for a give device. * @@ -153,6 +162,12 @@ struct devfreq { unsigned long min_freq; unsigned long max_freq; bool stop_polling; + + /* information for device freqeuncy transition */ + unsigned int total_trans; + unsigned int *trans_table; + unsigned long *time_in_state; + unsigned long last_stat_updated; }; #if defined(CONFIG_PM_DEVFREQ) -- cgit v1.2.3 From 0359d1afe4013d1a216908b6be4c6695a1db6fd6 Mon Sep 17 00:00:00 2001 From: Nishanth Menon Date: Mon, 29 Oct 2012 15:01:47 -0500 Subject: PM / devfreq: allow sysfs governor node to switch governor This allows us to select governor runtime from the default configuration without having to rebuild kernel or the devfreq driver using the sysfs node: /sys/class/devfreq/.../governor cat of the governor will return valid governor and an echo 'governor_name'>governor will switch governor Cc: Rajagopal Venkat Cc: MyungJoo Ham Cc: Kyungmin Park Cc: "Rafael J. Wysocki" Cc: Kevin Hilman Cc: linux-pm@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Nishanth Menon Acked-by: MyungJoo Ham Signed-off-by: MyungJoo Ham --- Documentation/ABI/testing/sysfs-class-devfreq | 2 +- drivers/devfreq/devfreq.c | 45 ++++++++++++++++++++++++++- 2 files changed, 45 insertions(+), 2 deletions(-) (limited to 'Documentation') diff --git a/Documentation/ABI/testing/sysfs-class-devfreq b/Documentation/ABI/testing/sysfs-class-devfreq index 40f98a9428dc..66876debca18 100644 --- a/Documentation/ABI/testing/sysfs-class-devfreq +++ b/Documentation/ABI/testing/sysfs-class-devfreq @@ -11,7 +11,7 @@ What: /sys/class/devfreq/.../governor Date: September 2011 Contact: MyungJoo Ham Description: - The /sys/class/devfreq/.../governor shows the name of the + The /sys/class/devfreq/.../governor show or set the name of the governor used by the corresponding devfreq object. What: /sys/class/devfreq/.../cur_freq diff --git a/drivers/devfreq/devfreq.c b/drivers/devfreq/devfreq.c index 0d7be03d561f..ff960f084c11 100644 --- a/drivers/devfreq/devfreq.c +++ b/drivers/devfreq/devfreq.c @@ -688,6 +688,49 @@ static ssize_t show_governor(struct device *dev, return sprintf(buf, "%s\n", to_devfreq(dev)->governor->name); } +static ssize_t store_governor(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct devfreq *df = to_devfreq(dev); + int ret; + char str_governor[DEVFREQ_NAME_LEN + 1]; + struct devfreq_governor *governor; + + ret = sscanf(buf, "%" __stringify(DEVFREQ_NAME_LEN) "s", str_governor); + if (ret != 1) + return -EINVAL; + + mutex_lock(&devfreq_list_lock); + governor = find_devfreq_governor(str_governor); + if (IS_ERR(governor)) { + ret = PTR_ERR(governor); + goto out; + } + if (df->governor == governor) + goto out; + + if (df->governor) { + ret = df->governor->event_handler(df, DEVFREQ_GOV_STOP, NULL); + if (ret) { + dev_warn(dev, "%s: Governor %s not stopped(%d)\n", + __func__, df->governor->name, ret); + goto out; + } + } + df->governor = governor; + strncpy(df->governor_name, governor->name, DEVFREQ_NAME_LEN); + ret = df->governor->event_handler(df, DEVFREQ_GOV_START, NULL); + if (ret) + dev_warn(dev, "%s: Governor %s not started(%d)\n", + __func__, df->governor->name, ret); +out: + mutex_unlock(&devfreq_list_lock); + + if (!ret) + ret = count; + return ret; +} + static ssize_t show_freq(struct device *dev, struct device_attribute *attr, char *buf) { @@ -873,7 +916,7 @@ static ssize_t show_trans_table(struct device *dev, struct device_attribute *att } static struct device_attribute devfreq_attrs[] = { - __ATTR(governor, S_IRUGO, show_governor, NULL), + __ATTR(governor, S_IRUGO | S_IWUSR, show_governor, store_governor), __ATTR(cur_freq, S_IRUGO, show_freq, NULL), __ATTR(available_frequencies, S_IRUGO, show_available_freqs, NULL), __ATTR(target_freq, S_IRUGO, show_target_freq, NULL), -- cgit v1.2.3 From 50a5b33e0159f8783ef617cdb9d5fbb6a3955b6f Mon Sep 17 00:00:00 2001 From: Nishanth Menon Date: Mon, 29 Oct 2012 15:01:48 -0500 Subject: PM / devfreq: Add sysfs node to expose available governors Now that governor list can be variable, knowing the available governors is useful to be able to select a governor using relevant sysfs node. Cc: Rajagopal Venkat Cc: MyungJoo Ham Cc: Kyungmin Park Cc: "Rafael J. Wysocki" Cc: Kevin Hilman Cc: linux-pm@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Nishanth Menon Signed-off-by: MyungJoo Ham --- Documentation/ABI/testing/sysfs-class-devfreq | 7 +++++++ drivers/devfreq/devfreq.c | 22 ++++++++++++++++++++++ 2 files changed, 29 insertions(+) (limited to 'Documentation') diff --git a/Documentation/ABI/testing/sysfs-class-devfreq b/Documentation/ABI/testing/sysfs-class-devfreq index 66876debca18..0ba6ea2f89d9 100644 --- a/Documentation/ABI/testing/sysfs-class-devfreq +++ b/Documentation/ABI/testing/sysfs-class-devfreq @@ -71,3 +71,10 @@ Description: the available frequencies of the corresponding devfreq object. This is a snapshot of available frequencies and not limited by the min/max frequency restrictions. + +What: /sys/class/devfreq/.../available_governors +Date: October 2012 +Contact: Nishanth Menon +Description: + The /sys/class/devfreq/.../available_governors shows + currently available governors in the system. diff --git a/drivers/devfreq/devfreq.c b/drivers/devfreq/devfreq.c index ff960f084c11..45e053e5b139 100644 --- a/drivers/devfreq/devfreq.c +++ b/drivers/devfreq/devfreq.c @@ -730,6 +730,27 @@ out: ret = count; return ret; } +static ssize_t show_available_governors(struct device *d, + struct device_attribute *attr, + char *buf) +{ + struct devfreq_governor *tmp_governor; + ssize_t count = 0; + + mutex_lock(&devfreq_list_lock); + list_for_each_entry(tmp_governor, &devfreq_governor_list, node) + count += scnprintf(&buf[count], (PAGE_SIZE - count - 2), + "%s ", tmp_governor->name); + mutex_unlock(&devfreq_list_lock); + + /* Truncate the trailing space */ + if (count) + count--; + + count += sprintf(&buf[count], "\n"); + + return count; +} static ssize_t show_freq(struct device *dev, struct device_attribute *attr, char *buf) @@ -917,6 +938,7 @@ static ssize_t show_trans_table(struct device *dev, struct device_attribute *att static struct device_attribute devfreq_attrs[] = { __ATTR(governor, S_IRUGO | S_IWUSR, show_governor, store_governor), + __ATTR(available_governors, S_IRUGO, show_available_governors, NULL), __ATTR(cur_freq, S_IRUGO, show_freq, NULL), __ATTR(available_frequencies, S_IRUGO, show_available_freqs, NULL), __ATTR(target_freq, S_IRUGO, show_target_freq, NULL), -- cgit v1.2.3 From 420993221175a45db5af012c53fd2fef4d9533dc Mon Sep 17 00:00:00 2001 From: Deepak Sikri Date: Tue, 27 Nov 2012 14:05:26 +0100 Subject: cpufreq: SPEAr: Add CPUFreq driver SPEAr is an ARM based family of SoCs. This patch adds in support of cpufreq driver for SPEAr SoCs. It is supported via DT only and so bindings are present in binding document. Signed-off-by: Deepak Sikri Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- .../devicetree/bindings/cpufreq/cpufreq-spear.txt | 42 +++ arch/arm/Kconfig | 1 + drivers/cpufreq/Kconfig.arm | 7 + drivers/cpufreq/Makefile | 1 + drivers/cpufreq/spear-cpufreq.c | 291 +++++++++++++++++++++ 5 files changed, 342 insertions(+) create mode 100644 Documentation/devicetree/bindings/cpufreq/cpufreq-spear.txt create mode 100644 drivers/cpufreq/spear-cpufreq.c (limited to 'Documentation') diff --git a/Documentation/devicetree/bindings/cpufreq/cpufreq-spear.txt b/Documentation/devicetree/bindings/cpufreq/cpufreq-spear.txt new file mode 100644 index 000000000000..f3d44984d91c --- /dev/null +++ b/Documentation/devicetree/bindings/cpufreq/cpufreq-spear.txt @@ -0,0 +1,42 @@ +SPEAr cpufreq driver +------------------- + +SPEAr SoC cpufreq driver for CPU frequency scaling. +It supports both uniprocessor (UP) and symmetric multiprocessor (SMP) systems +which share clock across all CPUs. + +Required properties: +- cpufreq_tbl: Table of frequencies CPU could be transitioned into, in the + increasing order. + +Optional properties: +- clock-latency: Specify the possible maximum transition latency for clock, in + unit of nanoseconds. + +Both required and optional properties listed above must be defined under node +/cpus/cpu@0. + +Examples: +-------- +cpus { + + <...> + + cpu@0 { + compatible = "arm,cortex-a9"; + reg = <0>; + + <...> + + cpufreq_tbl = < 166000 + 200000 + 250000 + 300000 + 400000 + 500000 + 600000 >; + }; + + <...> + +}; diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig index ade7e924bef5..159e99737e31 100644 --- a/arch/arm/Kconfig +++ b/arch/arm/Kconfig @@ -904,6 +904,7 @@ config ARCH_NOMADIK config PLAT_SPEAR bool "ST SPEAr" + select ARCH_HAS_CPUFREQ select ARCH_REQUIRE_GPIOLIB select ARM_AMBA select CLKDEV_LOOKUP diff --git a/drivers/cpufreq/Kconfig.arm b/drivers/cpufreq/Kconfig.arm index 5961e6415f08..a0b3661d90b0 100644 --- a/drivers/cpufreq/Kconfig.arm +++ b/drivers/cpufreq/Kconfig.arm @@ -76,3 +76,10 @@ config ARM_EXYNOS5250_CPUFREQ help This adds the CPUFreq driver for Samsung EXYNOS5250 SoC. + +config ARM_SPEAR_CPUFREQ + bool "SPEAr CPUFreq support" + depends on PLAT_SPEAR + default y + help + This adds the CPUFreq driver support for SPEAr SOCs. diff --git a/drivers/cpufreq/Makefile b/drivers/cpufreq/Makefile index 5b1413e47216..1f254ec087c1 100644 --- a/drivers/cpufreq/Makefile +++ b/drivers/cpufreq/Makefile @@ -50,6 +50,7 @@ obj-$(CONFIG_ARM_EXYNOS4210_CPUFREQ) += exynos4210-cpufreq.o obj-$(CONFIG_ARM_EXYNOS4X12_CPUFREQ) += exynos4x12-cpufreq.o obj-$(CONFIG_ARM_EXYNOS5250_CPUFREQ) += exynos5250-cpufreq.o obj-$(CONFIG_ARM_OMAP2PLUS_CPUFREQ) += omap-cpufreq.o +obj-$(CONFIG_ARM_SPEAR_CPUFREQ) += spear-cpufreq.o ################################################################################## # PowerPC platform drivers diff --git a/drivers/cpufreq/spear-cpufreq.c b/drivers/cpufreq/spear-cpufreq.c new file mode 100644 index 000000000000..4575cfe41755 --- /dev/null +++ b/drivers/cpufreq/spear-cpufreq.c @@ -0,0 +1,291 @@ +/* + * drivers/cpufreq/spear-cpufreq.c + * + * CPU Frequency Scaling for SPEAr platform + * + * Copyright (C) 2012 ST Microelectronics + * Deepak Sikri + * + * This file is licensed under the terms of the GNU General Public + * License version 2. This program is licensed "as is" without any + * warranty of any kind, whether express or implied. + */ + +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + +#include +#include +#include +#include +#include +#include +#include +#include + +/* SPEAr CPUFreq driver data structure */ +static struct { + struct clk *clk; + unsigned int transition_latency; + struct cpufreq_frequency_table *freq_tbl; + u32 cnt; +} spear_cpufreq; + +int spear_cpufreq_verify(struct cpufreq_policy *policy) +{ + return cpufreq_frequency_table_verify(policy, spear_cpufreq.freq_tbl); +} + +static unsigned int spear_cpufreq_get(unsigned int cpu) +{ + return clk_get_rate(spear_cpufreq.clk) / 1000; +} + +static struct clk *spear1340_cpu_get_possible_parent(unsigned long newfreq) +{ + struct clk *sys_pclk; + int pclk; + /* + * In SPEAr1340, cpu clk's parent sys clk can take input from + * following sources + */ + const char *sys_clk_src[] = { + "sys_syn_clk", + "pll1_clk", + "pll2_clk", + "pll3_clk", + }; + + /* + * As sys clk can have multiple source with their own range + * limitation so we choose possible sources accordingly + */ + if (newfreq <= 300000000) + pclk = 0; /* src is sys_syn_clk */ + else if (newfreq > 300000000 && newfreq <= 500000000) + pclk = 3; /* src is pll3_clk */ + else if (newfreq == 600000000) + pclk = 1; /* src is pll1_clk */ + else + return ERR_PTR(-EINVAL); + + /* Get parent to sys clock */ + sys_pclk = clk_get(NULL, sys_clk_src[pclk]); + if (IS_ERR(sys_pclk)) + pr_err("Failed to get %s clock\n", sys_clk_src[pclk]); + + return sys_pclk; +} + +/* + * In SPEAr1340, we cannot use newfreq directly because we need to actually + * access a source clock (clk) which might not be ancestor of cpu at present. + * Hence in SPEAr1340 we would operate on source clock directly before switching + * cpu clock to it. + */ +static int spear1340_set_cpu_rate(struct clk *sys_pclk, unsigned long newfreq) +{ + struct clk *sys_clk; + int ret = 0; + + sys_clk = clk_get_parent(spear_cpufreq.clk); + if (IS_ERR(sys_clk)) { + pr_err("failed to get cpu's parent (sys) clock\n"); + return PTR_ERR(sys_clk); + } + + /* Set the rate of the source clock before changing the parent */ + ret = clk_set_rate(sys_pclk, newfreq); + if (ret) { + pr_err("Failed to set sys clk rate to %lu\n", newfreq); + return ret; + } + + ret = clk_set_parent(sys_clk, sys_pclk); + if (ret) { + pr_err("Failed to set sys clk parent\n"); + return ret; + } + + return 0; +} + +static int spear_cpufreq_target(struct cpufreq_policy *policy, + unsigned int target_freq, unsigned int relation) +{ + struct cpufreq_freqs freqs; + unsigned long newfreq; + struct clk *srcclk; + int index, ret, mult = 1; + + if (cpufreq_frequency_table_target(policy, spear_cpufreq.freq_tbl, + target_freq, relation, &index)) + return -EINVAL; + + freqs.cpu = policy->cpu; + freqs.old = spear_cpufreq_get(0); + + newfreq = spear_cpufreq.freq_tbl[index].frequency * 1000; + if (of_machine_is_compatible("st,spear1340")) { + /* + * SPEAr1340 is special in the sense that due to the possibility + * of multiple clock sources for cpu clk's parent we can have + * different clock source for different frequency of cpu clk. + * Hence we need to choose one from amongst these possible clock + * sources. + */ + srcclk = spear1340_cpu_get_possible_parent(newfreq); + if (IS_ERR(srcclk)) { + pr_err("Failed to get src clk\n"); + return PTR_ERR(srcclk); + } + + /* SPEAr1340: src clk is always 2 * intended cpu clk */ + mult = 2; + } else { + /* + * src clock to be altered is ancestor of cpu clock. Hence we + * can directly work on cpu clk + */ + srcclk = spear_cpufreq.clk; + } + + newfreq = clk_round_rate(srcclk, newfreq * mult); + if (newfreq < 0) { + pr_err("clk_round_rate failed for cpu src clock\n"); + return newfreq; + } + + freqs.new = newfreq / 1000; + freqs.new /= mult; + cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE); + + if (mult == 2) + ret = spear1340_set_cpu_rate(srcclk, newfreq); + else + ret = clk_set_rate(spear_cpufreq.clk, newfreq); + + /* Get current rate after clk_set_rate, in case of failure */ + if (ret) { + pr_err("CPU Freq: cpu clk_set_rate failed: %d\n", ret); + freqs.new = clk_get_rate(spear_cpufreq.clk) / 1000; + } + + cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE); + return ret; +} + +static int spear_cpufreq_init(struct cpufreq_policy *policy) +{ + int ret; + + ret = cpufreq_frequency_table_cpuinfo(policy, spear_cpufreq.freq_tbl); + if (ret) { + pr_err("cpufreq_frequency_table_cpuinfo() failed"); + return ret; + } + + cpufreq_frequency_table_get_attr(spear_cpufreq.freq_tbl, policy->cpu); + policy->cpuinfo.transition_latency = spear_cpufreq.transition_latency; + policy->cur = spear_cpufreq_get(0); + + cpumask_copy(policy->cpus, topology_core_cpumask(policy->cpu)); + cpumask_copy(policy->related_cpus, policy->cpus); + + return 0; +} + +static int spear_cpufreq_exit(struct cpufreq_policy *policy) +{ + cpufreq_frequency_table_put_attr(policy->cpu); + return 0; +} + +static struct freq_attr *spear_cpufreq_attr[] = { + &cpufreq_freq_attr_scaling_available_freqs, + NULL, +}; + +static struct cpufreq_driver spear_cpufreq_driver = { + .name = "cpufreq-spear", + .flags = CPUFREQ_STICKY, + .verify = spear_cpufreq_verify, + .target = spear_cpufreq_target, + .get = spear_cpufreq_get, + .init = spear_cpufreq_init, + .exit = spear_cpufreq_exit, + .attr = spear_cpufreq_attr, +}; + +static int spear_cpufreq_driver_init(void) +{ + struct device_node *np; + const struct property *prop; + struct cpufreq_frequency_table *freq_tbl; + const __be32 *val; + int cnt, i, ret; + + np = of_find_node_by_path("/cpus/cpu@0"); + if (!np) { + pr_err("No cpu node found"); + return -ENODEV; + } + + if (of_property_read_u32(np, "clock-latency", + &spear_cpufreq.transition_latency)) + spear_cpufreq.transition_latency = CPUFREQ_ETERNAL; + + prop = of_find_property(np, "cpufreq_tbl", NULL); + if (!prop || !prop->value) { + pr_err("Invalid cpufreq_tbl"); + ret = -ENODEV; + goto out_put_node; + } + + cnt = prop->length / sizeof(u32); + val = prop->value; + + freq_tbl = kmalloc(sizeof(*freq_tbl) * (cnt + 1), GFP_KERNEL); + if (!freq_tbl) { + ret = -ENOMEM; + goto out_put_node; + } + + for (i = 0; i < cnt; i++) { + freq_tbl[i].index = i; + freq_tbl[i].frequency = be32_to_cpup(val++); + } + + freq_tbl[i].index = i; + freq_tbl[i].frequency = CPUFREQ_TABLE_END; + + spear_cpufreq.freq_tbl = freq_tbl; + + of_node_put(np); + + spear_cpufreq.clk = clk_get(NULL, "cpu_clk"); + if (IS_ERR(spear_cpufreq.clk)) { + pr_err("Unable to get CPU clock\n"); + ret = PTR_ERR(spear_cpufreq.clk); + goto out_put_mem; + } + + ret = cpufreq_register_driver(&spear_cpufreq_driver); + if (!ret) + return 0; + + pr_err("failed register driver: %d\n", ret); + clk_put(spear_cpufreq.clk); + +out_put_mem: + kfree(freq_tbl); + return ret; + +out_put_node: + of_node_put(np); + return ret; +} +late_initcall(spear_cpufreq_driver_init); + +MODULE_AUTHOR("Deepak Sikri "); +MODULE_DESCRIPTION("SPEAr CPUFreq driver"); +MODULE_LICENSE("GPL"); -- cgit v1.2.3 From 59c3987805a92f50f4c37392a36ab951327a6e29 Mon Sep 17 00:00:00 2001 From: Mika Westerberg Date: Fri, 7 Dec 2012 23:11:51 +0100 Subject: ACPI: add documentation about ACPI 5 enumeration Add a document that describes how to take advantage of ACPI enumeration for buses like platform, I2C and SPI. In addition to that we document how to translate ACPI GpioIo and GpioInt resources to be useful in Linux device drivers. Signed-off-by: Mika Westerberg Signed-off-by: Rafael J. Wysocki --- Documentation/acpi/enumeration.txt | 227 +++++++++++++++++++++++++++++++++++++ 1 file changed, 227 insertions(+) create mode 100644 Documentation/acpi/enumeration.txt (limited to 'Documentation') diff --git a/Documentation/acpi/enumeration.txt b/Documentation/acpi/enumeration.txt new file mode 100644 index 000000000000..4f27785ca0c8 --- /dev/null +++ b/Documentation/acpi/enumeration.txt @@ -0,0 +1,227 @@ +ACPI based device enumeration +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +ACPI 5 introduced a set of new resources (UartTSerialBus, I2cSerialBus, +SpiSerialBus, GpioIo and GpioInt) which can be used in enumerating slave +devices behind serial bus controllers. + +In addition we are starting to see peripherals integrated in the +SoC/Chipset to appear only in ACPI namespace. These are typically devices +that are accessed through memory-mapped registers. + +In order to support this and re-use the existing drivers as much as +possible we decided to do following: + + o Devices that have no bus connector resource are represented as + platform devices. + + o Devices behind real busses where there is a connector resource + are represented as struct spi_device or struct i2c_device + (standard UARTs are not busses so there is no struct uart_device). + +As both ACPI and Device Tree represent a tree of devices (and their +resources) this implementation follows the Device Tree way as much as +possible. + +The ACPI implementation enumerates devices behind busses (platform, SPI and +I2C), creates the physical devices and binds them to their ACPI handle in +the ACPI namespace. + +This means that when ACPI_HANDLE(dev) returns non-NULL the device was +enumerated from ACPI namespace. This handle can be used to extract other +device-specific configuration. There is an example of this below. + +Platform bus support +~~~~~~~~~~~~~~~~~~~~ +Since we are using platform devices to represent devices that are not +connected to any physical bus we only need to implement a platform driver +for the device and add supported ACPI IDs. If this same IP-block is used on +some other non-ACPI platform, the driver might work out of the box or needs +some minor changes. + +Adding ACPI support for an existing driver should be pretty +straightforward. Here is the simplest example: + + #ifdef CONFIG_ACPI + static struct acpi_device_id mydrv_acpi_match[] = { + /* ACPI IDs here */ + { } + }; + MODULE_DEVICE_TABLE(acpi, mydrv_acpi_match); + #endif + + static struct platform_driver my_driver = { + ... + .driver = { + .acpi_match_table = ACPI_PTR(mydrv_acpi_match), + }, + }; + +If the driver needs to perform more complex initialization like getting and +configuring GPIOs it can get its ACPI handle and extract this information +from ACPI tables. + +Currently the kernel is not able to automatically determine from which ACPI +device it should make the corresponding platform device so we need to add +the ACPI device explicitly to acpi_platform_device_ids list defined in +drivers/acpi/scan.c. This limitation is only for the platform devices, SPI +and I2C devices are created automatically as described below. + +SPI serial bus support +~~~~~~~~~~~~~~~~~~~~~~ +Slave devices behind SPI bus have SpiSerialBus resource attached to them. +This is extracted automatically by the SPI core and the slave devices are +enumerated once spi_register_master() is called by the bus driver. + +Here is what the ACPI namespace for a SPI slave might look like: + + Device (EEP0) + { + Name (_ADR, 1) + Name (_CID, Package() { + "ATML0025", + "AT25", + }) + ... + Method (_CRS, 0, NotSerialized) + { + SPISerialBus(1, PolarityLow, FourWireMode, 8, + ControllerInitiated, 1000000, ClockPolarityLow, + ClockPhaseFirst, "\\_SB.PCI0.SPI1",) + } + ... + +The SPI device drivers only need to add ACPI IDs in a similar way than with +the platform device drivers. Below is an example where we add ACPI support +to at25 SPI eeprom driver (this is meant for the above ACPI snippet): + + #ifdef CONFIG_ACPI + static struct acpi_device_id at25_acpi_match[] = { + { "AT25", 0 }, + { }, + }; + MODULE_DEVICE_TABLE(acpi, at25_acpi_match); + #endif + + static struct spi_driver at25_driver = { + .driver = { + ... + .acpi_match_table = ACPI_PTR(at25_acpi_match), + }, + }; + +Note that this driver actually needs more information like page size of the +eeprom etc. but at the time writing this there is no standard way of +passing those. One idea is to return this in _DSM method like: + + Device (EEP0) + { + ... + Method (_DSM, 4, NotSerialized) + { + Store (Package (6) + { + "byte-len", 1024, + "addr-mode", 2, + "page-size, 32 + }, Local0) + + // Check UUIDs etc. + + Return (Local0) + } + +Then the at25 SPI driver can get this configation by calling _DSM on its +ACPI handle like: + + struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER, NULL }; + struct acpi_object_list input; + acpi_status status; + + /* Fill in the input buffer */ + + status = acpi_evaluate_object(ACPI_HANDLE(&spi->dev), "_DSM", + &input, &output); + if (ACPI_FAILURE(status)) + /* Handle the error */ + + /* Extract the data here */ + + kfree(output.pointer); + +I2C serial bus support +~~~~~~~~~~~~~~~~~~~~~~ +The slaves behind I2C bus controller only need to add the ACPI IDs like +with the platform and SPI drivers. However the I2C bus controller driver +needs to call acpi_i2c_register_devices() after it has added the adapter. + +An I2C bus (controller) driver does: + + ... + ret = i2c_add_numbered_adapter(adapter); + if (ret) + /* handle error */ + + of_i2c_register_devices(adapter); + /* Enumerate the slave devices behind this bus via ACPI */ + acpi_i2c_register_devices(adapter); + +Below is an example of how to add ACPI support to the existing mpu3050 +input driver: + + #ifdef CONFIG_ACPI + static struct acpi_device_id mpu3050_acpi_match[] = { + { "MPU3050", 0 }, + { }, + }; + MODULE_DEVICE_TABLE(acpi, mpu3050_acpi_match); + #endif + + static struct i2c_driver mpu3050_i2c_driver = { + .driver = { + .name = "mpu3050", + .owner = THIS_MODULE, + .pm = &mpu3050_pm, + .of_match_table = mpu3050_of_match, + .acpi_match_table ACPI_PTR(mpu3050_acpi_match), + }, + .probe = mpu3050_probe, + .remove = __devexit_p(mpu3050_remove), + .id_table = mpu3050_ids, + }; + +GPIO support +~~~~~~~~~~~~ +ACPI 5 introduced two new resources to describe GPIO connections: GpioIo +and GpioInt. These resources are used be used to pass GPIO numbers used by +the device to the driver. For example: + + Method (_CRS, 0, NotSerialized) + { + Name (SBUF, ResourceTemplate() + { + GpioIo (Exclusive, PullDefault, 0x0000, 0x0000, + IoRestrictionOutputOnly, "\\_SB.PCI0.GPI0", + 0x00, ResourceConsumer,,) + { + // Pin List + 0x0055 + } + ... + + Return (SBUF) + } + } + +These GPIO numbers are controller relative and path "\\_SB.PCI0.GPI0" +specifies the path to the controller. In order to use these GPIOs in Linux +we need to translate them to the Linux GPIO numbers. + +The driver can do this by including and then calling +acpi_get_gpio(path, gpio). This will return the Linux GPIO number or +negative errno if there was no translation found. + +Other GpioIo parameters must be converted first by the driver to be +suitable to the gpiolib before passing them. + +In case of GpioInt resource an additional call to gpio_to_irq() must be +done before calling request_irq(). -- cgit v1.2.3