提交 052c6f19 编写于 作者: V Viresh Kumar 提交者: Rafael J. Wysocki

PM / OPP: Move away from RCU locking

The RCU locking isn't well suited for the OPP core. The RCU locking fits
better for reader heavy stuff, while the OPP core have at max one or two
readers only at a time.

Over that, it was getting very confusing the way RCU locking was used
with the OPP core. The individual OPPs are mostly well handled, i.e. for
an update a new structure was created and then that replaced the older
one. But the OPP tables were updated directly all the time from various
parts of the core. Though they were mostly used from within RCU locked
region, they didn't had much to do with RCU and were governed by the
mutex instead.

And that mixed with the 'opp_table_lock' has made the core even more
confusing.

Now that we are already managing the OPPs and the OPP tables with kernel
reference infrastructure, we can get rid of RCU locking completely and
simplify the code a lot.

Remove all RCU references from code and comments.

Acquire opp_table->lock while parsing the list of OPPs though.
Signed-off-by: NViresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: NStephen Boyd <sboyd@codeaurora.org>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
上级 5b650b38
此差异已折叠。
......@@ -137,12 +137,6 @@ void _dev_pm_opp_cpumask_remove_table(const struct cpumask *cpumask, bool of)
* This removes the OPP tables for CPUs present in the @cpumask.
* This should be used to remove all the OPPs entries associated with
* the cpus in @cpumask.
*
* Locking: The internal opp_table and opp structures are RCU protected.
* Hence this function internally uses RCU updater strategy with mutex locks
* to keep the integrity of the internal data structures. Callers should ensure
* that this function is *NOT* called under RCU protection or in contexts where
* mutex cannot be locked.
*/
void dev_pm_opp_cpumask_remove_table(const struct cpumask *cpumask)
{
......@@ -159,12 +153,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_cpumask_remove_table);
* @cpumask.
*
* Returns -ENODEV if OPP table isn't already present.
*
* Locking: The internal opp_table and opp structures are RCU protected.
* Hence this function internally uses RCU updater strategy with mutex locks
* to keep the integrity of the internal data structures. Callers should ensure
* that this function is *NOT* called under RCU protection or in contexts where
* mutex cannot be locked.
*/
int dev_pm_opp_set_sharing_cpus(struct device *cpu_dev,
const struct cpumask *cpumask)
......@@ -215,12 +203,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_set_sharing_cpus);
*
* Returns -ENODEV if OPP table isn't already present and -EINVAL if the OPP
* table's status is access-unknown.
*
* Locking: The internal opp_table and opp structures are RCU protected.
* Hence this function internally uses RCU updater strategy with mutex locks
* to keep the integrity of the internal data structures. Callers should ensure
* that this function is *NOT* called under RCU protection or in contexts where
* mutex cannot be locked.
*/
int dev_pm_opp_get_sharing_cpus(struct device *cpu_dev, struct cpumask *cpumask)
{
......
......@@ -28,7 +28,7 @@ static struct opp_table *_managed_opp(const struct device_node *np)
mutex_lock(&opp_table_lock);
list_for_each_entry_rcu(opp_table, &opp_tables, node) {
list_for_each_entry(opp_table, &opp_tables, node) {
if (opp_table->np == np) {
/*
* Multiple devices can point to the same OPP table and
......@@ -235,12 +235,6 @@ static int opp_parse_supplies(struct dev_pm_opp *opp, struct device *dev,
* @dev: device pointer used to lookup OPP table.
*
* Free OPPs created using static entries present in DT.
*
* Locking: The internal opp_table and opp structures are RCU protected.
* Hence this function indirectly uses RCU updater strategy with mutex locks
* to keep the integrity of the internal data structures. Callers should ensure
* that this function is *NOT* called under RCU protection or in contexts where
* mutex cannot be locked.
*/
void dev_pm_opp_of_remove_table(struct device *dev)
{
......@@ -269,12 +263,6 @@ static struct device_node *_of_get_opp_desc_node(struct device *dev)
* opp can be controlled using dev_pm_opp_enable/disable functions and may be
* removed by dev_pm_opp_remove.
*
* Locking: The internal opp_table and opp structures are RCU protected.
* Hence this function internally uses RCU updater strategy with mutex locks
* to keep the integrity of the internal data structures. Callers should ensure
* that this function is *NOT* called under RCU protection or in contexts where
* mutex cannot be locked.
*
* Return:
* 0 On success OR
* Duplicate OPPs (both freq and volt are same) and opp->available
......@@ -358,7 +346,7 @@ static int _opp_add_static_v2(struct opp_table *opp_table, struct device *dev,
* Notify the changes in the availability of the operable
* frequency/voltage list.
*/
srcu_notifier_call_chain(&opp_table->srcu_head, OPP_EVENT_ADD, new_opp);
blocking_notifier_call_chain(&opp_table->head, OPP_EVENT_ADD, new_opp);
return 0;
free_opp:
......@@ -470,12 +458,6 @@ static int _of_add_opp_table_v1(struct device *dev)
*
* Register the initial OPP table with the OPP library for given device.
*
* Locking: The internal opp_table and opp structures are RCU protected.
* Hence this function indirectly uses RCU updater strategy with mutex locks
* to keep the integrity of the internal data structures. Callers should ensure
* that this function is *NOT* called under RCU protection or in contexts where
* mutex cannot be locked.
*
* Return:
* 0 On success OR
* Duplicate OPPs (both freq and volt are same) and opp->available
......@@ -520,12 +502,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_of_add_table);
*
* This removes the OPP tables for CPUs present in the @cpumask.
* This should be used only to remove static entries created from DT.
*
* Locking: The internal opp_table and opp structures are RCU protected.
* Hence this function internally uses RCU updater strategy with mutex locks
* to keep the integrity of the internal data structures. Callers should ensure
* that this function is *NOT* called under RCU protection or in contexts where
* mutex cannot be locked.
*/
void dev_pm_opp_of_cpumask_remove_table(const struct cpumask *cpumask)
{
......@@ -538,12 +514,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_of_cpumask_remove_table);
* @cpumask: cpumask for which OPP table needs to be added.
*
* This adds the OPP tables for CPUs present in the @cpumask.
*
* Locking: The internal opp_table and opp structures are RCU protected.
* Hence this function internally uses RCU updater strategy with mutex locks
* to keep the integrity of the internal data structures. Callers should ensure
* that this function is *NOT* called under RCU protection or in contexts where
* mutex cannot be locked.
*/
int dev_pm_opp_of_cpumask_add_table(const struct cpumask *cpumask)
{
......@@ -591,12 +561,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_of_cpumask_add_table);
* This updates the @cpumask with CPUs that are sharing OPPs with @cpu_dev.
*
* Returns -ENOENT if operating-points-v2 isn't present for @cpu_dev.
*
* Locking: The internal opp_table and opp structures are RCU protected.
* Hence this function internally uses RCU updater strategy with mutex locks
* to keep the integrity of the internal data structures. Callers should ensure
* that this function is *NOT* called under RCU protection or in contexts where
* mutex cannot be locked.
*/
int dev_pm_opp_of_get_sharing_cpus(struct device *cpu_dev,
struct cpumask *cpumask)
......
......@@ -20,8 +20,7 @@
#include <linux/list.h>
#include <linux/limits.h>
#include <linux/pm_opp.h>
#include <linux/rculist.h>
#include <linux/rcupdate.h>
#include <linux/notifier.h>
struct clk;
struct regulator;
......@@ -52,9 +51,6 @@ extern struct list_head opp_tables;
* @node: opp table node. The nodes are maintained throughout the lifetime
* of boot. It is expected only an optimal set of OPPs are
* added to the library by the SoC framework.
* RCU usage: opp table is traversed with RCU locks. node
* modification is possible realtime, hence the modifications
* are protected by the opp_table_lock for integrity.
* IMPORTANT: the opp nodes should be maintained in increasing
* order.
* @kref: for reference count of the OPP.
......@@ -67,7 +63,6 @@ extern struct list_head opp_tables;
* @clock_latency_ns: Latency (in nanoseconds) of switching to this OPP's
* frequency from any other OPP's frequency.
* @opp_table: points back to the opp_table struct this opp belongs to
* @rcu_head: RCU callback head used for deferred freeing
* @np: OPP's device node.
* @dentry: debugfs dentry pointer (per opp)
*
......@@ -88,7 +83,6 @@ struct dev_pm_opp {
unsigned long clock_latency_ns;
struct opp_table *opp_table;
struct rcu_head rcu_head;
struct device_node *np;
......@@ -101,7 +95,6 @@ struct dev_pm_opp {
* struct opp_device - devices managed by 'struct opp_table'
* @node: list node
* @dev: device to which the struct object belongs
* @rcu_head: RCU callback head used for deferred freeing
* @dentry: debugfs dentry pointer (per device)
*
* This is an internal data structure maintaining the devices that are managed
......@@ -110,7 +103,6 @@ struct dev_pm_opp {
struct opp_device {
struct list_head node;
const struct device *dev;
struct rcu_head rcu_head;
#ifdef CONFIG_DEBUG_FS
struct dentry *dentry;
......@@ -128,10 +120,7 @@ enum opp_table_access {
* @node: table node - contains the devices with OPPs that
* have been registered. Nodes once added are not modified in this
* table.
* RCU usage: nodes are not modified in the table of opp_table,
* however addition is possible and is secured by opp_table_lock
* @srcu_head: notifier head to notify the OPP availability changes.
* @rcu_head: RCU callback head used for deferred freeing
* @head: notifier head to notify the OPP availability changes.
* @dev_list: list of devices that share these OPPs
* @opp_list: table of opps
* @kref: for reference count of the table.
......@@ -156,16 +145,11 @@ enum opp_table_access {
* This is an internal data structure maintaining the link to opps attached to
* a device. This structure is not meant to be shared to users as it is
* meant for book keeping and private to OPP library.
*
* Because the opp structures can be used from both rcu and srcu readers, we
* need to wait for the grace period of both of them before freeing any
* resources. And so we have used kfree_rcu() from within call_srcu() handlers.
*/
struct opp_table {
struct list_head node;
struct srcu_notifier_head srcu_head;
struct rcu_head rcu_head;
struct blocking_notifier_head head;
struct list_head dev_list;
struct list_head opp_list;
struct kref kref;
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册