[PATCH RFC 0/7] ARM: shmobile: Support L2 shutdown mode in Suspend-to-RAM

[PATCH RFC 0/7] ARM: shmobile: Support L2 shutdown mode in Suspend-to-RAM

[Khiem Nguyen]

These patches implement L2 shutdown mode in RCar Gen2 SoC and change
Suspend-to-RAM from Core Standby mode to L2 shutdown mode.
The impact to CPUIdle operation was also put into consideration during
implementation.

I have confirmed Suspend-to-RAM in Lager board and Koelsch board. 
So far, there is no regression detected during my testing. 
There might be further patches to support Alt board. But it would be
after upstream kernel can boot up in Alt board.

I hope to receive some comments for this patch series whether it's
right direction or not. Therefore, I marked it with RFC.
Any comments are appreciated.

This patch series is written on top of renesas-devel-20140924-v3.17-rc6
and has dependency on CPUIdle patches from Kobayashi-san 
(http://www.spinics.net/lists/linux-sh/msg34150.html).

Khiem Nguyen (7):
  ARM: shmobile: use common ARMv7 cache disabling code
  ARM: shmobile: use signals from SYSC to wake up from Suspend-to-RAM
  ARM: shmobile: Implement L2 shutdown mode for Suspend-to-RAM
  ARM: shmobile: Backup and restore L2CTLR in Suspend-to-RAM
  ARM: shmobile: Clear External Event Request in system resume
  ARM: shmobile: Do additional settings for L2shutdown mode
  ARM: shmobile: Use separate CPUCMCR addresses for CA15/CA7

 arch/arm/mach-shmobile/common.h       |    4 ++
 arch/arm/mach-shmobile/headsmp.S      |   28 ++++++++++++
 arch/arm/mach-shmobile/platsmp-apmu.c |   81 ++++++++++++++++++++++++---------
 arch/arm/mach-shmobile/pm-rcar.c      |   22 +++++++++
 arch/arm/mach-shmobile/pm-rcar.h      |    1 +
 5 files changed, 115 insertions(+), 21 deletions(-)

-- 
1.7.9.5

Re: [PATCH RFC 0/7] ARM: shmobile: Support L2 shutdown mode in Suspend-to-RAM

[Geert Uytterhoeven]

Hi Khiem-san,

On Fri, Sep 26, 2014 at 4:39 AM, Khiem Nguyen
<khiem.nguyen.xt@renesas.com> wrote:
> These patches implement L2 shutdown mode in RCar Gen2 SoC and change
> Suspend-to-RAM from Core Standby mode to L2 shutdown mode.
> The impact to CPUIdle operation was also put into consideration during
> implementation.
>
> I have confirmed Suspend-to-RAM in Lager board and Koelsch board.
> So far, there is no regression detected during my testing.
> There might be further patches to support Alt board. But it would be
> after upstream kernel can boot up in Alt board.
>
> I hope to receive some comments for this patch series whether it's
> right direction or not. Therefore, I marked it with RFC.
> Any comments are appreciated.

While testing with various combinations of CONFIG_PM options, I noticed
the build fails if CONFIG_ARM_CPU_SUSPEND is not set:

arch/arm/mach-shmobile/built-in.o: In function
`shmobile_smp_apmu_enter_cpuidle':
:(.text+0x27c): undefined reference to `cpu_suspend'
arch/arm/mach-shmobile/built-in.o: In function `shmobile_cpuidle_notifier_call':
:(.text+0x305c): undefined reference to `cpu_resume'
arch/arm/mach-shmobile/built-in.o: In function `shmobile_smp_apmu_prepare_cpus':
:(.init.text+0xddc): undefined reference to `cpu_resume'
arch/arm/mach-shmobile/built-in.o: In function `shmobile_cpuidle_init':
:(.init.text+0xe9c): undefined reference to `cpu_resume'
make[3]: *** [vmlinux] Error 1
make[2]: *** [sub-make] Error 2
make[1]: *** [__sub-make] Error 2
make: *** [all] Error 2

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@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

Re: [PATCH RFC 0/7] ARM: shmobile: Support L2 shutdown mode in Suspend-to-RAM

[Khiem Nguyen]

Hi Geert-san,

On 9/26/2014 7:04 PM, Geert Uytterhoeven wrote:
> Hi Khiem-san,
> 
> On Fri, Sep 26, 2014 at 4:39 AM, Khiem Nguyen
> <khiem.nguyen.xt@renesas.com> wrote:
>> These patches implement L2 shutdown mode in RCar Gen2 SoC and change
>> Suspend-to-RAM from Core Standby mode to L2 shutdown mode.
>> The impact to CPUIdle operation was also put into consideration during
>> implementation.
>>
>> I have confirmed Suspend-to-RAM in Lager board and Koelsch board.
>> So far, there is no regression detected during my testing.
>> There might be further patches to support Alt board. But it would be
>> after upstream kernel can boot up in Alt board.
>>
>> I hope to receive some comments for this patch series whether it's
>> right direction or not. Therefore, I marked it with RFC.
>> Any comments are appreciated.
> 
> While testing with various combinations of CONFIG_PM options, I noticed
> the build fails if CONFIG_ARM_CPU_SUSPEND is not set:
> 
> arch/arm/mach-shmobile/built-in.o: In function
> `shmobile_smp_apmu_enter_cpuidle':
> :(.text+0x27c): undefined reference to `cpu_suspend'
> arch/arm/mach-shmobile/built-in.o: In function `shmobile_cpuidle_notifier_call':
> :(.text+0x305c): undefined reference to `cpu_resume'
> arch/arm/mach-shmobile/built-in.o: In function `shmobile_smp_apmu_prepare_cpus':
> :(.init.text+0xddc): undefined reference to `cpu_resume'
> arch/arm/mach-shmobile/built-in.o: In function `shmobile_cpuidle_init':
> :(.init.text+0xe9c): undefined reference to `cpu_resume'
> make[3]: *** [vmlinux] Error 1
> make[2]: *** [sub-make] Error 2
> make[1]: *** [__sub-make] Error 2
> make: *** [all] Error 2
> 
That is interesting build error.
To make it clearer and able to fix it, please kindly confirm below points:
1. This only occurred after merging my patch ?
(The part related to shmobile_cpuidle_init seems out of my scope.)
2. What kind of 'various combinations of CONFIG_PM options' were you select ?

Thanks.

> Gr{oetje,eeting}s,
> 
>                         Geert
> 
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@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
> 

-- 
Best regards,
KHIEM Nguyen

Re: [PATCH RFC 0/7] ARM: shmobile: Support L2 shutdown mode in Suspend-to-RAM

[Khiem Nguyen]

On 9/26/2014 7:20 PM, Khiem Nguyen wrote:
> Hi Geert-san,
> 
> On 9/26/2014 7:04 PM, Geert Uytterhoeven wrote:
>> Hi Khiem-san,
>>
>> On Fri, Sep 26, 2014 at 4:39 AM, Khiem Nguyen
>> <khiem.nguyen.xt@renesas.com> wrote:
>>> These patches implement L2 shutdown mode in RCar Gen2 SoC and change
>>> Suspend-to-RAM from Core Standby mode to L2 shutdown mode.
>>> The impact to CPUIdle operation was also put into consideration during
>>> implementation.
>>>
>>> I have confirmed Suspend-to-RAM in Lager board and Koelsch board.
>>> So far, there is no regression detected during my testing.
>>> There might be further patches to support Alt board. But it would be
>>> after upstream kernel can boot up in Alt board.
>>>
>>> I hope to receive some comments for this patch series whether it's
>>> right direction or not. Therefore, I marked it with RFC.
>>> Any comments are appreciated.
>>
>> While testing with various combinations of CONFIG_PM options, I noticed
>> the build fails if CONFIG_ARM_CPU_SUSPEND is not set:
>>
>> arch/arm/mach-shmobile/built-in.o: In function
>> `shmobile_smp_apmu_enter_cpuidle':
>> :(.text+0x27c): undefined reference to `cpu_suspend'
>> arch/arm/mach-shmobile/built-in.o: In function `shmobile_cpuidle_notifier_call':
>> :(.text+0x305c): undefined reference to `cpu_resume'
>> arch/arm/mach-shmobile/built-in.o: In function `shmobile_smp_apmu_prepare_cpus':
>> :(.init.text+0xddc): undefined reference to `cpu_resume'
>> arch/arm/mach-shmobile/built-in.o: In function `shmobile_cpuidle_init':
>> :(.init.text+0xe9c): undefined reference to `cpu_resume'
>> make[3]: *** [vmlinux] Error 1
>> make[2]: *** [sub-make] Error 2
>> make[1]: *** [__sub-make] Error 2
>> make: *** [all] Error 2
>>
> That is interesting build error.
> To make it clearer and able to fix it, please kindly confirm below points:
> 1. This only occurred after merging my patch ?
> (The part related to shmobile_cpuidle_init seems out of my scope.)
> 2. What kind of 'various combinations of CONFIG_PM options' were you select ?

To answer my own questions,
1. The build error occurred /after/ merging CPUIdle patches from Kobayashi-san. 
2. If CONFIG_SUSPEND is disabled (and CPUIdle patches was merged),
this build error occurred. 
(CONFIG_ARM_CPU_SUSPEND has dependency on CONFIG_SUSPEND)

@Kobayashi-san,
Please take care of this build error in your V2 patches.

Thanks.

> 
> Thanks.
> 
>> Gr{oetje,eeting}s,
>>
>>                         Geert
>>
>> --
>> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@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
>>
> 

-- 
Best regards,
KHIEM Nguyen