[PATCH/RFC] ARM: shmobile: Enable PM_RUNTIME in defconfig

[PATCH/RFC] ARM: shmobile: Enable PM_RUNTIME in defconfig

[Simon Horman]

Signed-off-by: Simon Horman <horms+renesas@verge.net.au>
---
 arch/arm/configs/shmobile_defconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm/configs/shmobile_defconfig b/arch/arm/configs/shmobile_defconfig
index 114e88e..4cbaf52 100644
--- a/arch/arm/configs/shmobile_defconfig
+++ b/arch/arm/configs/shmobile_defconfig
@@ -35,6 +35,7 @@ CONFIG_KEXEC=y
 CONFIG_VFP=y
 CONFIG_NEON=y
 # CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS is not set
+CONFIG_PM_RUNTIME=y
 CONFIG_NET=y
 CONFIG_PACKET=y
 CONFIG_UNIX=y
-- 
1.8.5.2

[PATCH/RFC] ARM: shmobile: Enable PM_RUNTIME in defconfig

[Simon Horman]

Hi Geert,

I realise this change was your idea but I'd like to confirm some details.

My assumption is that it requires

[PATCH v4] drivers: sh: compile drivers/sh/pm_runtime.c if ARCH_SHMOBILE_MULTI

Is that indeed required?
Are there other dependencies that are not in the next branch
of my renesas tree?

On Fri, May 16, 2014 at 09:04:42AM +0900, Simon Horman wrote:
> Signed-off-by: Simon Horman <horms+renesas@verge.net.au>
> ---
>  arch/arm/configs/shmobile_defconfig | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/arch/arm/configs/shmobile_defconfig b/arch/arm/configs/shmobile_defconfig
> index 114e88e..4cbaf52 100644
> --- a/arch/arm/configs/shmobile_defconfig
> +++ b/arch/arm/configs/shmobile_defconfig
> @@ -35,6 +35,7 @@ CONFIG_KEXEC=y
>  CONFIG_VFP=y
>  CONFIG_NEON=y
>  # CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS is not set
> +CONFIG_PM_RUNTIME=y
>  CONFIG_NET=y
>  CONFIG_PACKET=y
>  CONFIG_UNIX=y
> -- 
> 1.8.5.2
> 

[PATCH/RFC] ARM: shmobile: Enable PM_RUNTIME in defconfig

[Geert Uytterhoeven]

Hi Simon,

On Fri, May 16, 2014 at 4:05 AM, Simon Horman <horms@verge.net.au> wrote:
> I realise this change was your idea but I'd like to confirm some details.
>
> My assumption is that it requires
>
> [PATCH v4] drivers: sh: compile drivers/sh/pm_runtime.c if ARCH_SHMOBILE_MULTI
>
> Is that indeed required?
> Are there other dependencies that are not in the next branch
> of my renesas tree?

Your assumption is wrong. The aforementioned patch is always needed in
the multi-platform case.

Please let me clarify:
drivers/sh/pm_runtime.c registers a bus notifier for platform devices, which
registers the default clock for automatic clock management by the PM core.
If CONFIG_PM_RUNTIME=y, the clock will be managed dynamically, based
on runtime requirements.
If CONFIG_PM_RUNTIME=n, the clock will be enabled at registration time.

So even if CONFIG_PM_RUNTIME=n, you need this patch, else the affected
clocks won't be enabled (if not already enabled by reset state or the boot
loader).

Note that some drivers (e.g. sh-sci) will still work fine without
drivers/sh/pm_runtime.c, as they manage their clocks themselves.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

[PATCH/RFC] ARM: shmobile: Enable PM_RUNTIME in defconfig

[Simon Horman]

On Fri, May 16, 2014 at 09:17:03AM +0200, Geert Uytterhoeven wrote:
> Hi Simon,
> 
> On Fri, May 16, 2014 at 4:05 AM, Simon Horman <horms@verge.net.au> wrote:
> > I realise this change was your idea but I'd like to confirm some details.
> >
> > My assumption is that it requires
> >
> > [PATCH v4] drivers: sh: compile drivers/sh/pm_runtime.c if ARCH_SHMOBILE_MULTI
> >
> > Is that indeed required?
> > Are there other dependencies that are not in the next branch
> > of my renesas tree?
> 
> Your assumption is wrong. The aforementioned patch is always needed in
> the multi-platform case.
> 
> Please let me clarify:
> drivers/sh/pm_runtime.c registers a bus notifier for platform devices, which
> registers the default clock for automatic clock management by the PM core.
> If CONFIG_PM_RUNTIME=y, the clock will be managed dynamically, based
> on runtime requirements.
> If CONFIG_PM_RUNTIME=n, the clock will be enabled at registration time.
> 
> So even if CONFIG_PM_RUNTIME=n, you need this patch, else the affected
> clocks won't be enabled (if not already enabled by reset state or the boot
> loader).
> 
> Note that some drivers (e.g. sh-sci) will still work fine without
> drivers/sh/pm_runtime.c, as they manage their clocks themselves.

Thanks, I now understand that the above mentioned patch is required
regardless of the value of CONFIG_PM_RUNTIME.

With that in mind I'll rephrase my question: are you are ware
of any patches this one (the defconfig update) should depend on?

[PATCH/RFC] ARM: shmobile: Enable PM_RUNTIME in defconfig

[Geert Uytterhoeven]

Hi Simon,

On Fri, May 16, 2014 at 10:08 AM, Simon Horman <horms@verge.net.au> wrote:
> With that in mind I'll rephrase my question: are you are ware
> of any patches this one (the defconfig update) should depend on?

No, I don't think it will introduce any regressions.
We had CONFIG_PM_RUNTIME in most other defconfigs, and I've been
running quite some real multi-platform kernels on Koelsch before.

All known clock issues, for which patches have been send/accepted, happen
regardless of the value of CONFIG_PM_RUNTIME.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

[PATCH/RFC] ARM: shmobile: Enable PM_RUNTIME in defconfig

[Simon Horman]

On Fri, May 16, 2014 at 10:39:46AM +0200, Geert Uytterhoeven wrote:
> Hi Simon,
> 
> On Fri, May 16, 2014 at 10:08 AM, Simon Horman <horms@verge.net.au> wrote:
> > With that in mind I'll rephrase my question: are you are ware
> > of any patches this one (the defconfig update) should depend on?
> 
> No, I don't think it will introduce any regressions.
> We had CONFIG_PM_RUNTIME in most other defconfigs, and I've been
> running quite some real multi-platform kernels on Koelsch before.
> 
> All known clock issues, for which patches have been send/accepted, happen
> regardless of the value of CONFIG_PM_RUNTIME.

Thanks, I'll queue this up.

[PATCH/RFC] ARM: shmobile: Enable PM_RUNTIME in defconfig

[Geert Uytterhoeven]

On Fri, May 16, 2014 at 2:04 AM, Simon Horman
<horms+renesas@verge.net.au> wrote:
> Signed-off-by: Simon Horman <horms+renesas@verge.net.au>

Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds