Subject: Re: [PATCH 1/2] PM / clk: Add stubs for pm_clk_suspend/resume
To: "Rafael J. Wysocki" <rjw@rjwysocki.net>
References: <1457105585-25157-1-git-send-email-jonathanh@nvidia.com>
 <56DF094E.4030809@nvidia.com> <4293335.ArkqpzdFRe@vostro.rjw.lan>
CC: <linux-pm@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
        <linux-tegra@vger.kernel.org>
From: Jon Hunter <jonathanh@nvidia.com>
Message-ID: <56E001C1.2050905@nvidia.com>
Date: Wed, 9 Mar 2016 10:58:09 +0000
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101
 Thunderbird/38.6.0
MIME-Version: 1.0
In-Reply-To: <4293335.ArkqpzdFRe@vostro.rjw.lan>
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2648
Lines: 73


On 08/03/16 22:58, Rafael J. Wysocki wrote:
> On Tuesday, March 08, 2016 05:18:06 PM Jon Hunter wrote:
>>
>> On 04/03/16 15:33, Jon Hunter wrote:
>>> The functions pm_clk_suspend() and pm_clk_resume() cannot be used
>>> directly within the main body of a function, because when CONFIG_PM_CLK
>>> is not defined the functions are defined as NULL and this will lead to
>>> a compilation error.
>>>
>>> Add stubs functions for pm_clk_suspend() and pm_clk_resume() so that
>>> these functions may be called directly.
>>>
>>> Please note that these functions are currently assigned to function
>>> pointers in the PM domain core and so to avoid have a valid function
>>> pointer defined when CONFIG_PM_CLK is not defined, define
>>> GENPD_FLAG_PM_CLK as 0 when CONFIG_PM_CLK is not defined to ensure
>>> the stubs are not populated as valid function pointers.
>>>
>>> Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
>>> ---
>>>  include/linux/pm_clock.h  | 10 ++++++++--
>>>  include/linux/pm_domain.h |  4 ++++
>>>  2 files changed, 12 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/include/linux/pm_clock.h b/include/linux/pm_clock.h
>>> index 25266c600021..ffefbf2df1aa 100644
>>> --- a/include/linux/pm_clock.h
>>> +++ b/include/linux/pm_clock.h
>>> @@ -72,8 +72,14 @@ static inline int pm_clk_add_clk(struct device *dev, struct clk *clk)
>>>  static inline void pm_clk_remove(struct device *dev, const char *con_id)
>>>  {
>>>  }
>>> -#define pm_clk_suspend	NULL
>>> -#define pm_clk_resume	NULL
>>> +static inline int pm_clk_suspend(struct device *dev)
>>> +{
>>> +	return -EINVAL;
>>> +}
>>> +static inline int pm_clk_resume(struct device *dev)
>>> +{
>>> +	return -EINVAL;
>>> +}
>>>  #endif
>>>  
>>>  #ifdef CONFIG_HAVE_CLK
>>> diff --git a/include/linux/pm_domain.h b/include/linux/pm_domain.h
>>> index 49cd8890b873..1074c2073df8 100644
>>> --- a/include/linux/pm_domain.h
>>> +++ b/include/linux/pm_domain.h
>>> @@ -17,7 +17,11 @@
>>>  #include <linux/notifier.h>
>>>  
>>>  /* Defines used for the flags field in the struct generic_pm_domain */
>>> +#ifdef CONFIG_PM_CLK
>>>  #define GENPD_FLAG_PM_CLK	(1U << 0) /* PM domain uses PM clk */
>>> +#else
>>> +#define GENPD_FLAG_PM_CLK	0
>>> +#endif
>>>  
>>>  #define GENPD_MAX_NUM_STATES	8 /* Number of possible low power states */
>>
>> Thinking about this some more, the alternative is for drivers using
>> PM_CLK to select it. Would that be appropriate? If so, we would not need
>> these stubs.
> 
> Well, I don't see why it might not be appropriate.

Ok, thanks. I will drop this patch for now and update the 2nd patch of
this series to address the comments from Geert.

Cheers
Jon