top command shows 100% irq

[naveen yadav]

put cc to arm mailing list.


On Sun, Apr 1, 2012 at 9:52 AM, naveen yadav <yad.naveen at gmail.com> wrote:
> Dear Russell,
>
> I am using linux 3.0.22 kernel, I backport patch "ARM: fix rcu stalls
> on SMP platforms"(7deabca0acfe02b8e18f59a4c95676012f49a304)  and found
> that
>
> when run top command
>
> CPU:   0% usr   0% sys   0% nic 0% idle   0% io   100% irq   0% sirq
>
> I found the issue:
> -----------------------------------------------------------------------------------------------------------------------------
> working condition: I need to modify your original patch for
> ipi_timer() like below.
> --------------------------------------------------------------------------------------------------------------------------------
> diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c
> index 57db122..26cdc49 100644
> --- a/arch/arm/kernel/smp.c
> +++ b/arch/arm/kernel/smp.c
> @@ -443,9 +443,7 @@ static DEFINE_PER_CPU(struct clock_event_device,
> percpu_clockevent);
>  static void ipi_timer(void)
>  {
>        struct clock_event_device *evt = &__get_cpu_var(percpu_clockevent);
>       irq_enter();
>        evt->event_handler(evt);
>       irq_exit();
>  }
>
>  #ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST
> @@ -548,7 +546,9 @@ void handle_IPI(int ipinr, struct pt_regs *regs)
>
>        switch (ipinr) {
>        case IPI_TIMER:
> //+               irq_enter();
>                ipi_timer();
> //+               irq_exit();
>                break;
>
> ___________________________________________________________________________________________
> Fail: condition:
> ___________________________________________________________________________________________
> Signed-off-by: Russell King <rmk+kernel at arm.linux.org.uk>
> ---
>  arch/arm/kernel/smp.c |   10 ++++++++--
>  1 files changed, 8 insertions(+), 2 deletions(-)
>
> diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c
> index 57db122..26cdc49 100644
> --- a/arch/arm/kernel/smp.c
> +++ b/arch/arm/kernel/smp.c
> @@ -443,9 +443,7 @@ static DEFINE_PER_CPU(struct clock_event_device,
> percpu_clockevent);
>  static void ipi_timer(void)
>  {
>        struct clock_event_device *evt = &__get_cpu_var(percpu_clockevent);
> -       irq_enter();
>        evt->event_handler(evt);
> -       irq_exit();
>  }
>
>  #ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST
> @@ -548,7 +546,9 @@ void handle_IPI(int ipinr, struct pt_regs *regs)
>
>        switch (ipinr) {
>        case IPI_TIMER:
> +               irq_enter();
>                ipi_timer();
> +               irq_exit();
>                break;
>
>
>
> So it seams to me when struct clock_event_device *evt =
> &__get_cpu_var(percpu_clockevent); is in irq_enter() function, this
> leads  100% irq.
> what may be issue.
>
> Thanks a lot