[PATCH v5 4/5] drivers: clocksource: add support for ARM architected timer event stream

[Catalin Marinas]

On Wed, Aug 28, 2013 at 11:44:48AM +0100, Sudeep KarkadaNagesha wrote:
> From: Will Deacon <will.deacon at arm.com>
> 
> The ARM architected timer can generate events (used for waking up
> CPUs executing the wfe instruction) at a frequency represented as a
> power-of-2 divisor of the clock rate.
> 
> An event stream might be used:
> - To impose a time-out on a wfe polling loop.
> - To safeguard against any programming error in case an expected event
>   is not generated.
> - To implement wfe-based timeouts for userspace locking implementations.
> 
> This patch computes the event stream frequency aiming for a period
> of 100us between events. It uses ARM/ARM64 specific backends to configure
> and enable the event stream.
> 
> Cc: Lorenzo Pieralisi <lorenzo.pieralisi at arm.com>
> Cc: Catalin Marinas <catalin.marinas at arm.com>
> Signed-off-by: Will Deacon <will.deacon at arm.com>
> [sudeep: moving ARM/ARM64 changes into separate patches
>          and adding Kconfig option]
> Signed-off-by: Sudeep KarkadaNagesha <sudeep.karkadanagesha at arm.com>
> ---
>  drivers/clocksource/Kconfig          | 14 ++++++++++++++
>  drivers/clocksource/arm_arch_timer.c | 21 +++++++++++++++++++++
>  include/clocksource/arm_arch_timer.h |  2 ++
>  3 files changed, 37 insertions(+)
> 
> diff --git a/drivers/clocksource/Kconfig b/drivers/clocksource/Kconfig
> index b7b9b04..74946ca 100644
> --- a/drivers/clocksource/Kconfig
> +++ b/drivers/clocksource/Kconfig
> @@ -74,6 +74,20 @@ config ARM_ARCH_TIMER
>  	bool
>  	select CLKSRC_OF if OF
>  
> +config ARM_ARCH_TIMER_EVTSTREAM
> +	bool "Support for ARM architected timer event stream generation"
> +	default y if ARM_ARCH_TIMER
> +	help
> +	  This option enables support for event stream generation based on
> +	  the ARM architected timer. It is used for waking up CPUs executing
> +	  the wfe instruction at a frequency represented as a power-of-2
> +	  divisor of the clock rate. An event stream might be useful for
> +	  imposing timeout on a wfe, safeguarding against any programming
> +	  errors in case an expected event is not generated or even to
> +	  implement wfe-based timeouts for userspace locking implementations.
> +	  This can be disabled for hardware validation purposes to detect any
> +	  hardware anomalies of missing events.

Maybe we should reorder this a bit and make the wfe-based timeouts as
the main reason.

> +
>  config ARM_GLOBAL_TIMER
>  	bool
>  	select CLKSRC_OF if OF
> diff --git a/drivers/clocksource/arm_arch_timer.c b/drivers/clocksource/arm_arch_timer.c
> index ffadd83..f42bef4 100644
> --- a/drivers/clocksource/arm_arch_timer.c
> +++ b/drivers/clocksource/arm_arch_timer.c
> @@ -123,6 +123,26 @@ static int arch_timer_set_next_event_phys(unsigned long evt,
>  	return 0;
>  }
>  
> +#ifdef CONFIG_ARM_ARCH_TIMER_EVTSTREAM
> +static void arch_timer_setup_evtstream(void)
> +{
> +	int evt_stream_div, pos;
> +
> +	/* Find the closest power of two to the divisor */
> +	evt_stream_div = arch_timer_rate / ARCH_TIMER_EVT_STREAM_FREQ;
> +	pos = fls(evt_stream_div);
> +	if (pos > 1 && !(evt_stream_div & (1 << (pos - 2))))
> +		pos--;
> +	/* enable event stream */
> +	arch_timer_evtstrm_config(true, min(pos, 15));
> +}
> +#else
> +static void arch_timer_setup_evtstream(void)
> +{
> +	arch_timer_evtstrm_config(false, 0);
> +}
> +#endif

I think we only need an arch_timer_evtstrm_enable() function without the
bool argument. We don't need to disable it at run-time.

> +
>  static int arch_timer_setup(struct clock_event_device *clk)
>  {
>  	clk->features = CLOCK_EVT_FEAT_ONESHOT | CLOCK_EVT_FEAT_C3STOP;
> @@ -154,6 +174,7 @@ static int arch_timer_setup(struct clock_event_device *clk)
>  	}
>  
>  	arch_counter_set_user_access();
> +	arch_timer_setup_evtstream();

and here just:

	if (IS_ENABLED(CONFIG_ARM_ARCH_TIMER_EVTSTREAM))
		arch_timer_evtstrm_enable();

-- 
Catalin