diff options
author | Nicolas Pitre <nicolas.pitre@linaro.org> | 2015-03-11 23:16:13 +0100 |
---|---|---|
committer | Olof Johansson <olof@lixom.net> | 2015-04-03 21:52:46 +0200 |
commit | d3a875444ad8d5e64c5a932361ca579312e49801 (patch) | |
tree | 88910c6b2e5b8f35541fbc57a975197981ea8113 /arch/arm/include/asm/mcpm.h | |
parent | Merge tag 'v4.1-rockchip-soc1' of git://git.kernel.org/pub/scm/linux/kernel/g... (diff) | |
download | linux-d3a875444ad8d5e64c5a932361ca579312e49801.tar.xz linux-d3a875444ad8d5e64c5a932361ca579312e49801.zip |
ARM: MCPM: move the algorithmic complexity to the core code
All backends are reimplementing a variation of the same CPU reference
count handling. They are also responsible for driving the MCPM special
low-level locking. This is needless duplication, involving algorithmic
requirements that are not necessarily obvious to the uninitiated.
And from past code review experience, those were all initially
implemented badly.
After 3 years, it is time to refactor as much common code to the core
MCPM facility to make the backends as simple as possible. To avoid a
flag day, the new scheme is introduced in parallel to the existing
backend interface. When all backends are converted over, the
compatibility interface could be removed.
The new MCPM backend interface implements simpler methods addressing
very platform specific tasks performed under lock protection while
keeping the algorithmic complexity and race avoidance local to the
core code.
Signed-off-by: Nicolas Pitre <nico@linaro.org>
Tested-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Signed-off-by: Olof Johansson <olof@lixom.net>
Diffstat (limited to 'arch/arm/include/asm/mcpm.h')
-rw-r--r-- | arch/arm/include/asm/mcpm.h | 65 |
1 files changed, 63 insertions, 2 deletions
diff --git a/arch/arm/include/asm/mcpm.h b/arch/arm/include/asm/mcpm.h index 3446f6a1d9fa..50b378f59e08 100644 --- a/arch/arm/include/asm/mcpm.h +++ b/arch/arm/include/asm/mcpm.h @@ -171,12 +171,73 @@ void mcpm_cpu_suspend(u64 expected_residency); int mcpm_cpu_powered_up(void); /* - * Platform specific methods used in the implementation of the above API. + * Platform specific callbacks used in the implementation of the above API. + * + * cpu_powerup: + * Make given CPU runable. Called with MCPM lock held and IRQs disabled. + * The given cluster is assumed to be set up (cluster_powerup would have + * been called beforehand). Must return 0 for success or negative error code. + * + * cluster_powerup: + * Set up power for given cluster. Called with MCPM lock held and IRQs + * disabled. Called before first cpu_powerup when cluster is down. Must + * return 0 for success or negative error code. + * + * cpu_suspend_prepare: + * Special suspend configuration. Called on target CPU with MCPM lock held + * and IRQs disabled. This callback is optional. If provided, it is called + * before cpu_powerdown_prepare. + * + * cpu_powerdown_prepare: + * Configure given CPU for power down. Called on target CPU with MCPM lock + * held and IRQs disabled. Power down must be effective only at the next WFI instruction. + * + * cluster_powerdown_prepare: + * Configure given cluster for power down. Called on one CPU from target + * cluster with MCPM lock held and IRQs disabled. A cpu_powerdown_prepare + * for each CPU in the cluster has happened when this occurs. + * + * cpu_cache_disable: + * Clean and disable CPU level cache for the calling CPU. Called on with IRQs + * disabled only. The CPU is no longer cache coherent with the rest of the + * system when this returns. + * + * cluster_cache_disable: + * Clean and disable the cluster wide cache as well as the CPU level cache + * for the calling CPU. No call to cpu_cache_disable will happen for this + * CPU. Called with IRQs disabled and only when all the other CPUs are done + * with their own cpu_cache_disable. The cluster is no longer cache coherent + * with the rest of the system when this returns. + * + * cpu_is_up: + * Called on given CPU after it has been powered up or resumed. The MCPM lock + * is held and IRQs disabled. This callback is optional. + * + * cluster_is_up: + * Called by the first CPU to be powered up or resumed in given cluster. + * The MCPM lock is held and IRQs disabled. This callback is optional. If + * provided, it is called before cpu_is_up for that CPU. + * + * wait_for_powerdown: + * Wait until given CPU is powered down. This is called in sleeping context. + * Some reasonable timeout must be considered. Must return 0 for success or + * negative error code. */ struct mcpm_platform_ops { + int (*cpu_powerup)(unsigned int cpu, unsigned int cluster); + int (*cluster_powerup)(unsigned int cluster); + void (*cpu_suspend_prepare)(unsigned int cpu, unsigned int cluster); + void (*cpu_powerdown_prepare)(unsigned int cpu, unsigned int cluster); + void (*cluster_powerdown_prepare)(unsigned int cluster); + void (*cpu_cache_disable)(void); + void (*cluster_cache_disable)(void); + void (*cpu_is_up)(unsigned int cpu, unsigned int cluster); + void (*cluster_is_up)(unsigned int cluster); + int (*wait_for_powerdown)(unsigned int cpu, unsigned int cluster); + + /* deprecated callbacks */ int (*power_up)(unsigned int cpu, unsigned int cluster); void (*power_down)(void); - int (*wait_for_powerdown)(unsigned int cpu, unsigned int cluster); void (*suspend)(u64); void (*powered_up)(void); }; |