[PATCH 1/2] ARM: local timers: Unmask interrupt before new TVAL is set

[PATCH 1/2] ARM: local timers: Unmask interrupt before new TVAL is set

[Rohit Vaswani]

Level triggered interrupt is deasserted when a new TVAL is written
only when the interrupt is unmasked. Make sure that the interrupt
is unmasked in CTL register before TVAL is written.
If this order is not followed, there are chances that on some
hardware you would not receive any timer interrupts.

Signed-off-by: Rohit Vaswani <rvaswani@codeaurora.org>
---
 arch/arm/kernel/arch_timer.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/arm/kernel/arch_timer.c b/arch/arm/kernel/arch_timer.c
index dd58035..1d0d9df 100644
--- a/arch/arm/kernel/arch_timer.c
+++ b/arch/arm/kernel/arch_timer.c
@@ -126,8 +126,8 @@ static int arch_timer_set_next_event(unsigned long evt,
 	ctrl |= ARCH_TIMER_CTRL_ENABLE;
 	ctrl &= ~ARCH_TIMER_CTRL_IT_MASK;
 
-	arch_timer_reg_write(ARCH_TIMER_REG_TVAL, evt);
 	arch_timer_reg_write(ARCH_TIMER_REG_CTRL, ctrl);
+	arch_timer_reg_write(ARCH_TIMER_REG_TVAL, evt);
 
 	return 0;
 }
-- 
Sent by an employee of the Qualcomm Innovation Center,Inc
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [PATCH 1/2] ARM: local timers: Unmask interrupt before new TVAL is set

[Marc Zyngier]

On Fri, 10 Aug 2012 14:57:34 -0700, Rohit Vaswani
<rvaswani@codeaurora.org>
wrote:
> Level triggered interrupt is deasserted when a new TVAL is written
> only when the interrupt is unmasked. Make sure that the interrupt
> is unmasked in CTL register before TVAL is written.
> If this order is not followed, there are chances that on some
> hardware you would not receive any timer interrupts.
> 
> Signed-off-by: Rohit Vaswani <rvaswani@codeaurora.org>
> ---
>  arch/arm/kernel/arch_timer.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/arch/arm/kernel/arch_timer.c b/arch/arm/kernel/arch_timer.c
> index dd58035..1d0d9df 100644
> --- a/arch/arm/kernel/arch_timer.c
> +++ b/arch/arm/kernel/arch_timer.c
> @@ -126,8 +126,8 @@ static int arch_timer_set_next_event(unsigned long
evt,
>  	ctrl |= ARCH_TIMER_CTRL_ENABLE;
>  	ctrl &= ~ARCH_TIMER_CTRL_IT_MASK;
>  
> -	arch_timer_reg_write(ARCH_TIMER_REG_TVAL, evt);
>  	arch_timer_reg_write(ARCH_TIMER_REG_CTRL, ctrl);
> +	arch_timer_reg_write(ARCH_TIMER_REG_TVAL, evt);

But by doing so, you're opening a window where TVAL can be negative (from
a previous timer trigger) and the interrupt unmasked, which would lead to
an immediate trigger, before TVAL is updated with the new value.

Does your hardware deassert the interrupt even when the enable bit is not
set? If so, would the following sequence work?

        ctrl &= ~(ARCH_TIMER_CTRL_ENABLE | ARCH_TIMER_CTRL_IT_MASK);
        arch_timer_reg_write(ARCH_TIMER_REG_CTRL, ctrl);

	arch_timer_reg_write(ARCH_TIMER_REG_TVAL, evt);

        ctrl |= ARCH_TIMER_CTRL_ENABLE;
        arch_timer_reg_write(ARCH_TIMER_REG_CTRL, ctrl);

Thanks,

        M.
-- 
Fast, cheap, reliable. Pick two.

Re: [PATCH 1/2] ARM: local timers: Unmask interrupt before new TVAL is set

[Rohit Vaswani]

On 8/11/2012 3:17 AM, Marc Zyngier wrote:
> On Fri, 10 Aug 2012 14:57:34 -0700, Rohit Vaswani
> <rvaswani@codeaurora.org>
> wrote:
>> Level triggered interrupt is deasserted when a new TVAL is written
>> only when the interrupt is unmasked. Make sure that the interrupt
>> is unmasked in CTL register before TVAL is written.
>> If this order is not followed, there are chances that on some
>> hardware you would not receive any timer interrupts.
>>
>> Signed-off-by: Rohit Vaswani <rvaswani@codeaurora.org>
>> ---
>>   arch/arm/kernel/arch_timer.c |    2 +-
>>   1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/arch/arm/kernel/arch_timer.c b/arch/arm/kernel/arch_timer.c
>> index dd58035..1d0d9df 100644
>> --- a/arch/arm/kernel/arch_timer.c
>> +++ b/arch/arm/kernel/arch_timer.c
>> @@ -126,8 +126,8 @@ static int arch_timer_set_next_event(unsigned long
> evt,
>>   	ctrl |= ARCH_TIMER_CTRL_ENABLE;
>>   	ctrl &= ~ARCH_TIMER_CTRL_IT_MASK;
>>   
>> -	arch_timer_reg_write(ARCH_TIMER_REG_TVAL, evt);
>>   	arch_timer_reg_write(ARCH_TIMER_REG_CTRL, ctrl);
>> +	arch_timer_reg_write(ARCH_TIMER_REG_TVAL, evt);
> But by doing so, you're opening a window where TVAL can be negative (from
> a previous timer trigger) and the interrupt unmasked, which would lead to
> an immediate trigger, before TVAL is updated with the new value.
>
> Does your hardware deassert the interrupt even when the enable bit is not
> set? If so, would the following sequence work?
>
>          ctrl &= ~(ARCH_TIMER_CTRL_ENABLE | ARCH_TIMER_CTRL_IT_MASK);
>          arch_timer_reg_write(ARCH_TIMER_REG_CTRL, ctrl);
>
> 	arch_timer_reg_write(ARCH_TIMER_REG_TVAL, evt);
>
>          ctrl |= ARCH_TIMER_CTRL_ENABLE;
>          arch_timer_reg_write(ARCH_TIMER_REG_CTRL, ctrl);
>
> Thanks,
>
>          M.
Thanks Marc, this works. I will re-send the patch with these changes.

Thanks,
Rohit Vaswani

-- 
Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.