Re: next-20160210 build: 2 failures 4 warnings (next-20160210)

Re: next-20160210 build: 2 failures 4 warnings (next-20160210)

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 625 bytes --]

On Wed, Feb 10, 2016 at 09:12:25AM +0000, Build bot for Mark Brown wrote:

Today's -next fails to build on non-SMP configurations due to:

> 	arm-multi_v5_defconfig
> ../drivers/cpufreq/cpufreq_governor.c:251:3: error: implicit declaration of function 'irq_work_queue_on' [-Werror=implicit-function-declaration]

caused by 0144fa03ef4606ae (cpufreq: governor: Replace timers with
utilization update callbacks).  IS_ENABLED() in an if statement doesn't
do a good job of making things conditional here since the code still has
to compile cleanly in both cases and there is no definition at all of
irq_work_queue_on() for !SMP.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

Re: next-20160210 build: 2 failures 4 warnings (next-20160210)

[Rafael J. Wysocki]

On Wed, Feb 10, 2016 at 10:52 AM, Mark Brown <broonie@kernel.org> wrote:
> On Wed, Feb 10, 2016 at 09:12:25AM +0000, Build bot for Mark Brown wrote:
>
> Today's -next fails to build on non-SMP configurations due to:
>
>>       arm-multi_v5_defconfig
>> ../drivers/cpufreq/cpufreq_governor.c:251:3: error: implicit declaration of function 'irq_work_queue_on' [-Werror=implicit-function-declaration]
>
> caused by 0144fa03ef4606ae (cpufreq: governor: Replace timers with
> utilization update callbacks).  IS_ENABLED() in an if statement doesn't
> do a good job of making things conditional here since the code still has
> to compile cleanly in both cases and there is no definition at all of
> irq_work_queue_on() for !SMP.

Thanks for the report!

OK, I'll change it to #ifdef/#else then.

Thanks,
Rafael

Re: next-20160210 build: 2 failures 4 warnings (next-20160210)

[Arnd Bergmann]

On Wednesday 10 February 2016 15:27:48 Rafael J. Wysocki wrote:
> On Wed, Feb 10, 2016 at 10:52 AM, Mark Brown <broonie@kernel.org> wrote:
> > On Wed, Feb 10, 2016 at 09:12:25AM +0000, Build bot for Mark Brown wrote:
> >
> > Today's -next fails to build on non-SMP configurations due to:
> >
> >>       arm-multi_v5_defconfig
> >> ../drivers/cpufreq/cpufreq_governor.c:251:3: error: implicit declaration of function 'irq_work_queue_on' [-Werror=implicit-function-declaration]
> >
> > caused by 0144fa03ef4606ae (cpufreq: governor: Replace timers with
> > utilization update callbacks).  IS_ENABLED() in an if statement doesn't
> > do a good job of making things conditional here since the code still has
> > to compile cleanly in both cases and there is no definition at all of
> > irq_work_queue_on() for !SMP.
> 
> Thanks for the report!
> 
> OK, I'll change it to #ifdef/#else then.
> 

I have a better idea, patch follows.

	Arnd

[PATCH] irq_work: unhide irq_work_queue_on declaration on non-SMP

[Arnd Bergmann]

The cpufreq code uses 'if (IS_ENABLED(CONFIG_SMP))' to check
whether it should queue a task on the local CPU or a remote
one, however the irq_work_queue_on() function is not declared
when CONFIG_SMP is not set:

drivers/cpufreq/cpufreq_governor.c: In function 'gov_queue_irq_work':
drivers/cpufreq/cpufreq_governor.c:251:3: error: implicit declaration of function 'irq_work_queue_on' [-Werror=implicit-function-declaration]
   irq_work_queue_on(&policy_dbs->irq_work, smp_processor_id());

This changes the conditional declaration so that irq_work_queue_on
just queues the irq work on the only available CPU when CONFIG_SMP
is not set, which is presumably what most people need anyway.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: 0144fa03ef46 ("cpufreq: governor: Replace timers with utilization update callbacks")

diff --git a/include/linux/irq_work.h b/include/linux/irq_work.h
index 47b9ebd4a74f..c9bde50ef317 100644
--- a/include/linux/irq_work.h
+++ b/include/linux/irq_work.h
@@ -33,9 +33,13 @@ void init_irq_work(struct irq_work *work, void (*func)(struct irq_work *))
 #define DEFINE_IRQ_WORK(name, _f) struct irq_work name = { .func = (_f), }
 
 bool irq_work_queue(struct irq_work *work);
-
 #ifdef CONFIG_SMP
 bool irq_work_queue_on(struct irq_work *work, int cpu);
+#else
+static inline bool irq_work_queue_on(struct irq_work *work, int cpu)
+{
+	return irq_work_queue(work);
+}
 #endif
 
 void irq_work_tick(void);

Re: [PATCH] irq_work: unhide irq_work_queue_on declaration on non-SMP

[Rafael J. Wysocki]

On Wed, Feb 10, 2016 at 4:07 PM, Arnd Bergmann <arnd@arndb.de> wrote:
> The cpufreq code uses 'if (IS_ENABLED(CONFIG_SMP))' to check
> whether it should queue a task on the local CPU or a remote
> one, however the irq_work_queue_on() function is not declared
> when CONFIG_SMP is not set:
>
> drivers/cpufreq/cpufreq_governor.c: In function 'gov_queue_irq_work':
> drivers/cpufreq/cpufreq_governor.c:251:3: error: implicit declaration of function 'irq_work_queue_on' [-Werror=implicit-function-declaration]
>    irq_work_queue_on(&policy_dbs->irq_work, smp_processor_id());
>
> This changes the conditional declaration so that irq_work_queue_on
> just queues the irq work on the only available CPU when CONFIG_SMP
> is not set, which is presumably what most people need anyway.
>
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> Fixes: 0144fa03ef46 ("cpufreq: governor: Replace timers with utilization update callbacks")
>
> diff --git a/include/linux/irq_work.h b/include/linux/irq_work.h
> index 47b9ebd4a74f..c9bde50ef317 100644
> --- a/include/linux/irq_work.h
> +++ b/include/linux/irq_work.h
> @@ -33,9 +33,13 @@ void init_irq_work(struct irq_work *work, void (*func)(struct irq_work *))
>  #define DEFINE_IRQ_WORK(name, _f) struct irq_work name = { .func = (_f), }
>
>  bool irq_work_queue(struct irq_work *work);
> -
>  #ifdef CONFIG_SMP
>  bool irq_work_queue_on(struct irq_work *work, int cpu);
> +#else
> +static inline bool irq_work_queue_on(struct irq_work *work, int cpu)
> +{
> +       return irq_work_queue(work);
> +}
>  #endif
>
>  void irq_work_tick(void);

I was thinking about this too, but then cpufreq will be the only user of it.

In any case can do it at any time later. :-)

Thanks,
Rafael

Re: [PATCH] irq_work: unhide irq_work_queue_on declaration on non-SMP

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 627 bytes --]

On Wed, Feb 10, 2016 at 04:27:42PM +0100, Rafael J. Wysocki wrote:

> > +#else
> > +static inline bool irq_work_queue_on(struct irq_work *work, int cpu)
> > +{
> > +       return irq_work_queue(work);
> > +}
> >  #endif

> I was thinking about this too, but then cpufreq will be the only user of it.

> In any case can do it at any time later. :-)

Well, there's currently only two other users of irq_work_queue_on()
anyway so that's a third of the userbase and it does seem the obvious
way to support any other future users that want to scale down to !SMP
cases painlessly.

Reviwed-by: Mark Brown <broonie@kernel.org>

FWIW.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 473 bytes --]