| Message ID | 1377686689-30828-6-git-send-email-Sudeep.KarkadaNagesha@arm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 28/08/13 11:44, Sudeep KarkadaNagesha wrote: > From: Sudeep KarkadaNagesha <sudeep.karkadanagesha@arm.com> > > Few control settings done in architected timer as part of initialisation > can be lost when CPU enters deeper power states. They need to be > re-initialised when the CPU is (warm)reset again. > > This patch adds CPU PM notifiers to do the timer initialisation on warm > resets. It also save the event stream divider value calculated during > cold reset and uses the same in warm reset path. > Ah, forgot to update the commit log. It's now updated: Few control settings done in architected timer as part of initialisation can be lost when CPU enters deeper power states. They need to be restored when the CPU is (warm)reset again. This patch adds CPU PM notifiers to save the counter control register when entering low power modes and restore it when CPU exits low power. > Cc: Catalin Marinas <catalin.marinas@arm.com> > Reviewed-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> > Reviewed-by: Will Deacon <will.deacon@arm.com> > Signed-off-by: Sudeep KarkadaNagesha <sudeep.karkadanagesha@arm.com> > --- > drivers/clocksource/arm_arch_timer.c | 34 ++++++++++++++++++++++++++++++++++ > 1 file changed, 34 insertions(+) > > diff --git a/drivers/clocksource/arm_arch_timer.c b/drivers/clocksource/arm_arch_timer.c > index f42bef4..ffb0bfb 100644 > --- a/drivers/clocksource/arm_arch_timer.c > +++ b/drivers/clocksource/arm_arch_timer.c > @@ -13,6 +13,7 @@ > #include <linux/device.h> > #include <linux/smp.h> > #include <linux/cpu.h> > +#include <linux/cpu_pm.h> > #include <linux/clockchips.h> > #include <linux/interrupt.h> > #include <linux/of_irq.h> > @@ -281,6 +282,33 @@ static struct notifier_block arch_timer_cpu_nb = { > .notifier_call = arch_timer_cpu_notify, > }; > > +#ifdef CONFIG_CPU_PM > +static unsigned int saved_cntkctl; > +static int arch_timer_cpu_pm_notify(struct notifier_block *self, > + unsigned long action, void *hcpu) > +{ > + if (action == CPU_PM_ENTER) > + saved_cntkctl = arch_timer_get_cntkctl(); > + else if (action == CPU_PM_ENTER_FAILED || action == CPU_PM_EXIT) > + arch_timer_set_cntkctl(saved_cntkctl); > + return NOTIFY_OK; > +} > + > +static struct notifier_block arch_timer_cpu_pm_notifier = { > + .notifier_call = arch_timer_cpu_pm_notify, > +}; > + > +static int __init arch_timer_cpu_pm_init(void) > +{ > + return cpu_pm_register_notifier(&arch_timer_cpu_pm_notifier); > +} > +#else > +static int __init arch_timer_cpu_pm_init(void) > +{ > + return 0; > +} > +#endif > + > static int __init arch_timer_register(void) > { > int err; > @@ -330,11 +358,17 @@ static int __init arch_timer_register(void) > if (err) > goto out_free_irq; > > + err = arch_timer_cpu_pm_init(); > + if (err) > + goto out_unreg_notify; > + > /* Immediately configure the timer on the boot CPU */ > arch_timer_setup(this_cpu_ptr(arch_timer_evt)); > > return 0; > > +out_unreg_notify: > + unregister_cpu_notifier(&arch_timer_cpu_nb); > out_free_irq: > if (arch_timer_use_virtual) > free_percpu_irq(arch_timer_ppi[VIRT_PPI], arch_timer_evt); >
Hi Venkappa, First of all, please make sure the relevant mailing list(ALKML in this case) and maintainers (MarcZ and Catalin as you are addressing him) are cc-ed, else there's every chance that the mail gets lost. On Thu, Feb 19, 2015 at 1:03 PM, Venkappa Mala <venkappa.m@samsung.com> wrote: > Sudeep KarkadaNagesha <Sudeep.KarkadaNagesha <at> arm.com> writes: > >> >> On 28/08/13 11:44, Sudeep KarkadaNagesha wrote: >> > From: Sudeep KarkadaNagesha <sudeep.karkadanagesha <at> arm.com> >> > >> > Few control settings done in architected timer as part of > initialisation >> > can be lost when CPU enters deeper power states. They need to be >> > re-initialised when the CPU is (warm)reset again. >> > >> > This patch adds CPU PM notifiers to do the timer initialisation on > warm >> > resets. It also save the event stream divider value calculated > during >> > cold reset and uses the same in warm reset path. >> > >> Ah, forgot to update the commit log. It's now updated: >> >> Few control settings done in architected timer as part of > initialisation >> can be lost when CPU enters deeper power states. They need to be >> restored when the CPU is (warm)reset again. >> >> This patch adds CPU PM notifiers to save the counter control register >> when entering low power modes and restore it when CPU exits low power. >> >> > Cc: Catalin Marinas <catalin.marinas <at> arm.com> >> > Reviewed-by: Lorenzo Pieralisi <lorenzo.pieralisi <at> arm.com> >> > Reviewed-by: Will Deacon <will.deacon <at> arm.com> >> > Signed-off-by: Sudeep KarkadaNagesha <sudeep.karkadanagesha <at> > arm.com> [...] > Hello Catalin, > > I have noticed this patch merge in > https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git > at >= linux-3.13.y branches. > > As part of ARMv7/ARMv8 vDSO experiments, i have tried to enable arch > timer in 3.10 branch but it is failed to boot like > "[1246696886.455361] swapper/0: 0] [c0] [c0] Calibrating delay > loop (skipped), value calculated using timer frequency.. 52.00 BogoMIPS > (lpj=260000)" > due to cntkctl context lost on both armv7/armv8 cases.However, it works > fine when I have pulled the patch to 3.10 branch. > > Perhaps, it will be good to have this patch in 3.10.y (LTS kernel) > branch as well. > is there any potential issue if we merge this patch or is it > intentional? > Are you targeting this on ARM64 platform ? If so, I don't understand how is the arch timer context is lost as the cpuidle driver is not merge to 3.10 stable kernel ? Regards, Sudeep
diff --git a/drivers/clocksource/arm_arch_timer.c b/drivers/clocksource/arm_arch_timer.c index f42bef4..ffb0bfb 100644 --- a/drivers/clocksource/arm_arch_timer.c +++ b/drivers/clocksource/arm_arch_timer.c @@ -13,6 +13,7 @@ #include <linux/device.h> #include <linux/smp.h> #include <linux/cpu.h> +#include <linux/cpu_pm.h> #include <linux/clockchips.h> #include <linux/interrupt.h> #include <linux/of_irq.h> @@ -281,6 +282,33 @@ static struct notifier_block arch_timer_cpu_nb = { .notifier_call = arch_timer_cpu_notify, }; +#ifdef CONFIG_CPU_PM +static unsigned int saved_cntkctl; +static int arch_timer_cpu_pm_notify(struct notifier_block *self, + unsigned long action, void *hcpu) +{ + if (action == CPU_PM_ENTER) + saved_cntkctl = arch_timer_get_cntkctl(); + else if (action == CPU_PM_ENTER_FAILED || action == CPU_PM_EXIT) + arch_timer_set_cntkctl(saved_cntkctl); + return NOTIFY_OK; +} + +static struct notifier_block arch_timer_cpu_pm_notifier = { + .notifier_call = arch_timer_cpu_pm_notify, +}; + +static int __init arch_timer_cpu_pm_init(void) +{ + return cpu_pm_register_notifier(&arch_timer_cpu_pm_notifier); +} +#else +static int __init arch_timer_cpu_pm_init(void) +{ + return 0; +} +#endif + static int __init arch_timer_register(void) { int err; @@ -330,11 +358,17 @@ static int __init arch_timer_register(void) if (err) goto out_free_irq; + err = arch_timer_cpu_pm_init(); + if (err) + goto out_unreg_notify; + /* Immediately configure the timer on the boot CPU */ arch_timer_setup(this_cpu_ptr(arch_timer_evt)); return 0; +out_unreg_notify: + unregister_cpu_notifier(&arch_timer_cpu_nb); out_free_irq: if (arch_timer_use_virtual) free_percpu_irq(arch_timer_ppi[VIRT_PPI], arch_timer_evt);