Re: [PATCH] ARM: mach-shmobile: Fix build when SMP is enabled and EMEV2 is not enabled

[Nobuhiro Iwamatsu <nobuhiro.iwamatsu.yj@renesas.com>]

Hi,

Simon Horman さんは書きました:
> Hi Iwamatsu-san,
> 
> On Thu, Jun 07, 2012 at 10:40:55AM +0900, Nobuhiro Iwamatsu wrote:
>> Build failed, when SMP is enabled and EMEV2 is not enabled.
>>
>> -----
>> arch/arm/mach-shmobile/built-in.o: In function `shmobile_platform_cpu_kill':
>> /home/iwamatsu/work/kernel/sh-2.6-devel/arch/arm/mach-shmobile/platsmp.c:62: undefined reference to `emev2_platform_cpu_kill'
>> arch/arm/mach-shmobile/built-in.o: In function `shmobile_smp_get_core_count':
>> /home/iwamatsu/work/kernel/sh-2.6-devel/arch/arm/mach-shmobile/platsmp.c:39: undefined reference to `emev2_get_core_count'
>> arch/arm/mach-shmobile/built-in.o: In function `shmobile_smp_prepare_cpus':
>> /home/iwamatsu/work/kernel/sh-2.6-devel/arch/arm/mach-shmobile/platsmp.c:53: undefined reference to `emev2_smp_prepare_cpus'
>> arch/arm/mach-shmobile/built-in.o: In function `platform_secondary_init':
>> /home/iwamatsu/work/kernel/sh-2.6-devel/arch/arm/mach-shmobile/platsmp.c:78: undefined reference to `emev2_secondary_init'
>> arch/arm/mach-shmobile/built-in.o: In function `boot_secondary':
>> /home/iwamatsu/work/kernel/sh-2.6-devel/arch/arm/mach-shmobile/platsmp.c:90: undefined reference to `emev2_boot_secondary
>> -----
>>
>> This is the cause by when EMEV2 is disabled, that the check by OF of EMEV2 is
>> performed in platsmp.c.
>> This patch revise what the function about EMEV2 may not be used in this file,
>> when EMEV2 is not enabled.
> 
> I am curious to know why, for example, sh73a0_smp_prepare_cpus() and
> r8a7779_smp_prepare_cpus() do not have similar problems. Is it because
> they are guarded by machine_if_* rather than of_machine_is_compatible() ?

Yes. for example, If machine_is_marzen() was set to (0), if line of is_r8a7779()
was deleted by cpp.

> 
>> Signed-off-by: Nobuhiro Iwamatsu <nobuhiro.iwamatsu.yj@renesas.com>
>> ---
>>  arch/arm/mach-shmobile/platsmp.c |    5 +++++
>>  1 file changed, 5 insertions(+)
>>
>> diff --git a/arch/arm/mach-shmobile/platsmp.c b/arch/arm/mach-shmobile/platsmp.c
>> index bacdd66..e859fcd 100644
>> --- a/arch/arm/mach-shmobile/platsmp.c
>> +++ b/arch/arm/mach-shmobile/platsmp.c
>> @@ -25,7 +25,12 @@
>>  #define is_sh73a0() (machine_is_ag5evm() || machine_is_kota2() || \
>>  			of_machine_is_compatible("renesas,sh73a0"))
>>  #define is_r8a7779() machine_is_marzen()
> 
> Which tree does this version of platsmp.c appear in?
> The use of of_machine_is_compatible() is new to me.

I confirmed by linus/HEAD and next branch/HEAD of rafael's.
> 
>> +
>> +#ifdef CONFIG_ARCH_EMEV2
>>  #define is_emev2() of_machine_is_compatible("renesas,emev2")
>> +#else
>> +#define is_emev2() (0)
>> +#endif
> 
> It would be nice to avoid a journey to ifdef hell :)
> 

I think too.
But currentry, mechanism in which we judge CPU seems not to spread on CONFIG.

Best regards,
   Nobuhiro