| Message ID | CAMQu2gw0x6saWrK-M3teu49tJDGvS8RhGrVSh6JTs5DEy+22pA@mail.gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Wednesday 22 August 2012, Shilimkar, Santosh wrote: > Was just thinking whether we should just take care of it at > core cpuidle level itself. Will below be enough to kill the build > error what you mentioned in the change log ? > > diff --git a/drivers/cpuidle/coupled.c b/drivers/cpuidle/coupled.c > index 2c9bf26..df34534 100644 > --- a/drivers/cpuidle/coupled.c > +++ b/drivers/cpuidle/coupled.c > @@ -314,7 +314,9 @@ static void cpuidle_coupled_poke(int cpu) > struct call_single_data *csd = &per_cpu(cpuidle_coupled_poke_cb, cpu); > > if (!cpumask_test_and_set_cpu(cpu, &cpuidle_coupled_poked_mask)) > +#ifdef CONFIG_SMP > __smp_call_function_single(cpu, csd, 0); > +#endif > } > That would work, but isn't the entire concept of the cpuidle-coupled driver dependent on SMP? If this driver makes no sense on UP, I think we should not attempt to build it. Arnd
On Wed, Aug 22, 2012 at 10:52 PM, Arnd Bergmann <arnd@arndb.de> wrote: > > On Wednesday 22 August 2012, Shilimkar, Santosh wrote: > > > Was just thinking whether we should just take care of it at > > core cpuidle level itself. Will below be enough to kill the build > > error what you mentioned in the change log ? > > > > diff --git a/drivers/cpuidle/coupled.c b/drivers/cpuidle/coupled.c > > index 2c9bf26..df34534 100644 > > --- a/drivers/cpuidle/coupled.c > > +++ b/drivers/cpuidle/coupled.c > > @@ -314,7 +314,9 @@ static void cpuidle_coupled_poke(int cpu) > > struct call_single_data *csd = &per_cpu(cpuidle_coupled_poke_cb, > > cpu); > > > > if (!cpumask_test_and_set_cpu(cpu, &cpuidle_coupled_poked_mask)) > > +#ifdef CONFIG_SMP > > __smp_call_function_single(cpu, csd, 0); > > +#endif > > } > > > > That would work, but isn't the entire concept of the cpuidle-coupled > driver > dependent on SMP? If this driver makes no sense on UP, I think we should > not attempt to build it. > I see your point but alternate patch is pushing down the fix to the low level driver and that means you end up patching more drivers when they use COUPLE idle infrastructure. That was the only reason I was thinking of suppressing the error at the source. Since it is just for the random builds and actually doesn't impact the real functionality as such, I am fine with your proposed patch too. Regards santosh
On Thursday 23 August 2012, Shilimkar, Santosh wrote: > I see your point but alternate patch is pushing down the fix to the low > level driver and that means you end up patching more drivers when they > use COUPLE idle infrastructure. That was the only reason I was thinking > of suppressing the error at the source. > > Since it is just for the random builds and actually doesn't impact the real > functionality as such, I am fine with your proposed patch too. Ok. It would be nice of course to test if this actually works on uniprocessor configurations. Arnd
On Thu, Aug 23, 2012 at 5:42 PM, Arnd Bergmann <arnd@arndb.de> wrote: > > On Thursday 23 August 2012, Shilimkar, Santosh wrote: > > I see your point but alternate patch is pushing down the fix to the low > > level driver and that means you end up patching more drivers when they > > use COUPLE idle infrastructure. That was the only reason I was thinking > > of suppressing the error at the source. > > > > Since it is just for the random builds and actually doesn't impact the > > real > > functionality as such, I am fine with your proposed patch too. > > Ok. It would be nice of course to test if this actually works on > uniprocessor > configurations. > Have tested the patch and it does boot with UP build on OMAP. Acked-tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
On Thursday 23 August 2012, Shilimkar, Santosh wrote: > > On Thu, Aug 23, 2012 at 5:42 PM, Arnd Bergmann <arnd@arndb.de> wrote: > > > > On Thursday 23 August 2012, Shilimkar, Santosh wrote: > > > I see your point but alternate patch is pushing down the fix to the low > > > level driver and that means you end up patching more drivers when they > > > use COUPLE idle infrastructure. That was the only reason I was thinking > > > of suppressing the error at the source. > > > > > > Since it is just for the random builds and actually doesn't impact the > > > real > > > functionality as such, I am fine with your proposed patch too. > > > > Ok. It would be nice of course to test if this actually works on > > uniprocessor > > configurations. > > > Have tested the patch and it does boot with UP build on OMAP. > > Acked-tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com> Ok, thanks a lot! Arnd
diff --git a/drivers/cpuidle/coupled.c b/drivers/cpuidle/coupled.c index 2c9bf26..df34534 100644 --- a/drivers/cpuidle/coupled.c +++ b/drivers/cpuidle/coupled.c @@ -314,7 +314,9 @@ static void cpuidle_coupled_poke(int cpu) struct call_single_data *csd = &per_cpu(cpuidle_coupled_poke_cb, cpu); if (!cpumask_test_and_set_cpu(cpu, &cpuidle_coupled_poked_mask)) +#ifdef CONFIG_SMP __smp_call_function_single(cpu, csd, 0); +#endif } /**