Re: [PATCH v7 4/6] cpuidle: introduce HAVE_GENERIC_CPUIDLE_ENTER for ARM{32,64} platforms

[Sudeep Holla]

On 07/07/16 14:21, Rafael J. Wysocki wrote:

On Tuesday, June 28, 2016 02:55:50 PM Sudeep Holla wrote:

The function arm_enter_idle_state is exactly the same in both generic
ARM{32,64} CPUIdle driver and will be the same even on ARM64 backend
for ACPI processor idle driver. So we can unify it and move it as
generic_cpuidle_enter by introducing HAVE_GENERIC_CPUIDLE_ENTER and
enabling the same on both ARM{32,64}.

This is in preparation of reuse of the generic cpuidle entry function
for ACPI LPI support on ARM64.

Cc: "Rafael J. Wysocki" <r...@rjwysocki.net>
Cc: Daniel Lezcano <daniel.lezc...@linaro.org>
Cc: Lorenzo Pieralisi <lorenzo.pieral...@arm.com>
Signed-off-by: Sudeep Holla <sudeep.ho...@arm.com>
---
  arch/arm/Kconfig              |  1 +
  arch/arm/kernel/cpuidle.c     |  4 ++--
  arch/arm64/Kconfig            |  1 +
  arch/arm64/kernel/cpuidle.c   |  6 +++---
  drivers/cpuidle/Kconfig       |  3 +++
  drivers/cpuidle/cpuidle-arm.c | 21 +--------------------
  drivers/cpuidle/cpuidle.c     | 35 +++++++++++++++++++++++++++++++++++
  include/linux/cpuidle.h       |  8 ++++++++
  8 files changed, 54 insertions(+), 25 deletions(-)

diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index 90542db1220d..52b3dca0381c 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
@@ -54,6 +54,7 @@ config ARM
        select HAVE_FTRACE_MCOUNT_RECORD if (!XIP_KERNEL)
        select HAVE_FUNCTION_GRAPH_TRACER if (!THUMB2_KERNEL)
        select HAVE_FUNCTION_TRACER if (!XIP_KERNEL)
+       select HAVE_GENERIC_CPUIDLE_ENTER

That "generic" part in the name concerns me a bit, because the thing is not
really generic.  It is "common on ARM" rather.

I agree and that's exactly what I told Daniel. It's rather just
*ARM Generic*. Any preference on the name ? I had it header file under
include/linu/cpuidle-arm.h in the previous version. Do you prefer that ?

[..]

@@ -255,6 +256,40 @@ int cpuidle_select(struct cpuidle_driver *drv, struct 
cpuidle_device *dev)
  }

  /**
+ * cpuidle_generic_enter_state - enter into the specified idle state using the
+ *                              generic callback if implemented
+ *
+ * @index: the index in the idle state table
+ *
+ * Returns the index in the idle state, -1 in case of error.
+ */
+int cpuidle_generic_enter_state(int idx)
+{
+       int ret;
+
+       if (!idx) {
+               cpu_do_idle();
+               return idx;
+       }
+
+       ret = cpu_pm_enter();

If you look at the users of cpu_pm_enter(), they are all ARM.  Nobody else uses
it, so at least please put cpuidle_generic_enter_state() under a #ifdef to avoid
building it in vain on non-ARM.

Agreed, that should be taken care once I do the changes you are
suggesting below.

[...]

@@ -246,6 +248,12 @@ static inline int cpuidle_register_governor(struct 
cpuidle_governor *gov)
  {return 0;}
  #endif

+#ifdef CONFIG_HAVE_GENERIC_CPUIDLE_ENTER
+extern int cpuidle_generic_enter(int idx);

And if you put it under the #ifdef here, its definition also should go under
the same #ifdef.

Yes I will do that.

--
Regards,
Sudeep