brintos

brintos / linux-shallow public Read only

0
0
Text · 12.3 KiB · 93bab53 Raw
313 lines · plain
1========================2The Common Clk Framework3========================4 5:Author: Mike Turquette <mturquette@ti.com>6 7This document endeavours to explain the common clk framework details,8and how to port a platform over to this framework.  It is not yet a9detailed explanation of the clock api in include/linux/clk.h, but10perhaps someday it will include that information.11 12Introduction and interface split13================================14 15The common clk framework is an interface to control the clock nodes16available on various devices today.  This may come in the form of clock17gating, rate adjustment, muxing or other operations.  This framework is18enabled with the CONFIG_COMMON_CLK option.19 20The interface itself is divided into two halves, each shielded from the21details of its counterpart.  First is the common definition of struct22clk which unifies the framework-level accounting and infrastructure that23has traditionally been duplicated across a variety of platforms.  Second24is a common implementation of the clk.h api, defined in25drivers/clk/clk.c.  Finally there is struct clk_ops, whose operations26are invoked by the clk api implementation.27 28The second half of the interface is comprised of the hardware-specific29callbacks registered with struct clk_ops and the corresponding30hardware-specific structures needed to model a particular clock.  For31the remainder of this document any reference to a callback in struct32clk_ops, such as .enable or .set_rate, implies the hardware-specific33implementation of that code.  Likewise, references to struct clk_foo34serve as a convenient shorthand for the implementation of the35hardware-specific bits for the hypothetical "foo" hardware.36 37Tying the two halves of this interface together is struct clk_hw, which38is defined in struct clk_foo and pointed to within struct clk_core.  This39allows for easy navigation between the two discrete halves of the common40clock interface.41 42Common data structures and api43==============================44 45Below is the common struct clk_core definition from46drivers/clk/clk.c, modified for brevity::47 48	struct clk_core {49		const char		*name;50		const struct clk_ops	*ops;51		struct clk_hw		*hw;52		struct module		*owner;53		struct clk_core		*parent;54		const char		**parent_names;55		struct clk_core		**parents;56		u8			num_parents;57		u8			new_parent_index;58		...59	};60 61The members above make up the core of the clk tree topology.  The clk62api itself defines several driver-facing functions which operate on63struct clk.  That api is documented in include/linux/clk.h.64 65Platforms and devices utilizing the common struct clk_core use the struct66clk_ops pointer in struct clk_core to perform the hardware-specific parts of67the operations defined in clk-provider.h::68 69	struct clk_ops {70		int		(*prepare)(struct clk_hw *hw);71		void		(*unprepare)(struct clk_hw *hw);72		int		(*is_prepared)(struct clk_hw *hw);73		void		(*unprepare_unused)(struct clk_hw *hw);74		int		(*enable)(struct clk_hw *hw);75		void		(*disable)(struct clk_hw *hw);76		int		(*is_enabled)(struct clk_hw *hw);77		void		(*disable_unused)(struct clk_hw *hw);78		unsigned long	(*recalc_rate)(struct clk_hw *hw,79						unsigned long parent_rate);80		long		(*round_rate)(struct clk_hw *hw,81						unsigned long rate,82						unsigned long *parent_rate);83		int		(*determine_rate)(struct clk_hw *hw,84						  struct clk_rate_request *req);85		int		(*set_parent)(struct clk_hw *hw, u8 index);86		u8		(*get_parent)(struct clk_hw *hw);87		int		(*set_rate)(struct clk_hw *hw,88					    unsigned long rate,89					    unsigned long parent_rate);90		int		(*set_rate_and_parent)(struct clk_hw *hw,91					    unsigned long rate,92					    unsigned long parent_rate,93					    u8 index);94		unsigned long	(*recalc_accuracy)(struct clk_hw *hw,95						unsigned long parent_accuracy);96		int		(*get_phase)(struct clk_hw *hw);97		int		(*set_phase)(struct clk_hw *hw, int degrees);98		void		(*init)(struct clk_hw *hw);99		void		(*debug_init)(struct clk_hw *hw,100					      struct dentry *dentry);101	};102 103Hardware clk implementations104============================105 106The strength of the common struct clk_core comes from its .ops and .hw pointers107which abstract the details of struct clk from the hardware-specific bits, and108vice versa.  To illustrate consider the simple gateable clk implementation in109drivers/clk/clk-gate.c::110 111	struct clk_gate {112		struct clk_hw	hw;113		void __iomem    *reg;114		u8              bit_idx;115		...116	};117 118struct clk_gate contains struct clk_hw hw as well as hardware-specific119knowledge about which register and bit controls this clk's gating.120Nothing about clock topology or accounting, such as enable_count or121notifier_count, is needed here.  That is all handled by the common122framework code and struct clk_core.123 124Let's walk through enabling this clk from driver code::125 126	struct clk *clk;127	clk = clk_get(NULL, "my_gateable_clk");128 129	clk_prepare(clk);130	clk_enable(clk);131 132The call graph for clk_enable is very simple::133 134	clk_enable(clk);135		clk->ops->enable(clk->hw);136		[resolves to...]137			clk_gate_enable(hw);138			[resolves struct clk gate with to_clk_gate(hw)]139				clk_gate_set_bit(gate);140 141And the definition of clk_gate_set_bit::142 143	static void clk_gate_set_bit(struct clk_gate *gate)144	{145		u32 reg;146 147		reg = __raw_readl(gate->reg);148		reg |= BIT(gate->bit_idx);149		writel(reg, gate->reg);150	}151 152Note that to_clk_gate is defined as::153 154	#define to_clk_gate(_hw) container_of(_hw, struct clk_gate, hw)155 156This pattern of abstraction is used for every clock hardware157representation.158 159Supporting your own clk hardware160================================161 162When implementing support for a new type of clock it is only necessary to163include the following header::164 165	#include <linux/clk-provider.h>166 167To construct a clk hardware structure for your platform you must define168the following::169 170	struct clk_foo {171		struct clk_hw hw;172		... hardware specific data goes here ...173	};174 175To take advantage of your data you'll need to support valid operations176for your clk::177 178	struct clk_ops clk_foo_ops = {179		.enable		= &clk_foo_enable,180		.disable	= &clk_foo_disable,181	};182 183Implement the above functions using container_of::184 185	#define to_clk_foo(_hw) container_of(_hw, struct clk_foo, hw)186 187	int clk_foo_enable(struct clk_hw *hw)188	{189		struct clk_foo *foo;190 191		foo = to_clk_foo(hw);192 193		... perform magic on foo ...194 195		return 0;196	};197 198Below is a matrix detailing which clk_ops are mandatory based upon the199hardware capabilities of that clock.  A cell marked as "y" means200mandatory, a cell marked as "n" implies that either including that201callback is invalid or otherwise unnecessary.  Empty cells are either202optional or must be evaluated on a case-by-case basis.203 204.. table:: clock hardware characteristics205 206   +----------------+------+-------------+---------------+-------------+------+207   |                | gate | change rate | single parent | multiplexer | root |208   +================+======+=============+===============+=============+======+209   |.prepare        |      |             |               |             |      |210   +----------------+------+-------------+---------------+-------------+------+211   |.unprepare      |      |             |               |             |      |212   +----------------+------+-------------+---------------+-------------+------+213   +----------------+------+-------------+---------------+-------------+------+214   |.enable         | y    |             |               |             |      |215   +----------------+------+-------------+---------------+-------------+------+216   |.disable        | y    |             |               |             |      |217   +----------------+------+-------------+---------------+-------------+------+218   |.is_enabled     | y    |             |               |             |      |219   +----------------+------+-------------+---------------+-------------+------+220   +----------------+------+-------------+---------------+-------------+------+221   |.recalc_rate    |      | y           |               |             |      |222   +----------------+------+-------------+---------------+-------------+------+223   |.round_rate     |      | y [1]_      |               |             |      |224   +----------------+------+-------------+---------------+-------------+------+225   |.determine_rate |      | y [1]_      |               |             |      |226   +----------------+------+-------------+---------------+-------------+------+227   |.set_rate       |      | y           |               |             |      |228   +----------------+------+-------------+---------------+-------------+------+229   +----------------+------+-------------+---------------+-------------+------+230   |.set_parent     |      |             | n             | y           | n    |231   +----------------+------+-------------+---------------+-------------+------+232   |.get_parent     |      |             | n             | y           | n    |233   +----------------+------+-------------+---------------+-------------+------+234   +----------------+------+-------------+---------------+-------------+------+235   |.recalc_accuracy|      |             |               |             |      |236   +----------------+------+-------------+---------------+-------------+------+237   +----------------+------+-------------+---------------+-------------+------+238   |.init           |      |             |               |             |      |239   +----------------+------+-------------+---------------+-------------+------+240 241.. [1] either one of round_rate or determine_rate is required.242 243Finally, register your clock at run-time with a hardware-specific244registration function.  This function simply populates struct clk_foo's245data and then passes the common struct clk parameters to the framework246with a call to::247 248	clk_register(...)249 250See the basic clock types in ``drivers/clk/clk-*.c`` for examples.251 252Disabling clock gating of unused clocks253=======================================254 255Sometimes during development it can be useful to be able to bypass the256default disabling of unused clocks. For example, if drivers aren't enabling257clocks properly but rely on them being on from the bootloader, bypassing258the disabling means that the driver will remain functional while the issues259are sorted out.260 261You can see which clocks have been disabled by booting your kernel with these262parameters::263 264 tp_printk trace_event=clk:clk_disable265 266To bypass this disabling, include "clk_ignore_unused" in the bootargs to the267kernel.268 269Locking270=======271 272The common clock framework uses two global locks, the prepare lock and the273enable lock.274 275The enable lock is a spinlock and is held across calls to the .enable,276.disable operations. Those operations are thus not allowed to sleep,277and calls to the clk_enable(), clk_disable() API functions are allowed in278atomic context.279 280For clk_is_enabled() API, it is also designed to be allowed to be used in281atomic context. However, it doesn't really make any sense to hold the enable282lock in core, unless you want to do something else with the information of283the enable state with that lock held. Otherwise, seeing if a clk is enabled is284a one-shot read of the enabled state, which could just as easily change after285the function returns because the lock is released. Thus the user of this API286needs to handle synchronizing the read of the state with whatever they're287using it for to make sure that the enable state doesn't change during that288time.289 290The prepare lock is a mutex and is held across calls to all other operations.291All those operations are allowed to sleep, and calls to the corresponding API292functions are not allowed in atomic context.293 294This effectively divides operations in two groups from a locking perspective.295 296Drivers don't need to manually protect resources shared between the operations297of one group, regardless of whether those resources are shared by multiple298clocks or not. However, access to resources that are shared between operations299of the two groups needs to be protected by the drivers. An example of such a300resource would be a register that controls both the clock rate and the clock301enable/disable state.302 303The clock framework is reentrant, in that a driver is allowed to call clock304framework functions from within its implementation of clock operations. This305can for instance cause a .set_rate operation of one clock being called from306within the .set_rate operation of another clock. This case must be considered307in the driver implementations, but the code flow is usually controlled by the308driver in that case.309 310Note that locking must also be considered when code outside of the common311clock framework needs to access resources used by the clock operations. This312is considered out of scope of this document.313