[PATCH 06/10] ARM: clps711x: Add CLPS711X clocksource driver

[Alexander Shiyan]

On Fri, 19 Jul 2013 16:38:54 +0200
Daniel Lezcano <daniel.lezcano at linaro.org> wrote:
> (Added Thomas and John in Cc in case they want to review the patch).
> 
> On 07/18/2013 08:34 PM, Alexander Shiyan wrote:
> > This adds the clocksource driver for Cirrus Logic CLPS711X series SoCs.
> > Designed primarily for migration CLPS711X subarch for multiplatform & DT,
> > for this as the "OF" and "not-OF" calls implemented.
> 
> When a new driver is submitted, I ask for a more detailed changelog
> about the driver itself: how it works, any specificities, etc ...
> 
> That will help people to understand the code better at least for the review.

Basically, this series is a port of the CLPS711X subsystems in places specially
designed for this purpose. Additionally added to the standard description of the
subsystem for use with DT.
As for specs, I'm a little confused by this question. In this case, I'm doing
is not anything new for this CPU. The CPU has two timers and I use both to
provide the correct time sources.

> > Signed-off-by: Alexander Shiyan <shc_work at mail.ru>
> > ---
> >  arch/arm/Kconfig                      |   2 -
> >  drivers/clocksource/Kconfig           |   6 ++
> >  drivers/clocksource/Makefile          |   1 +
> >  drivers/clocksource/clps711x-clksrc.c | 151 ++++++++++++++++++++++++++++++++++
> >  4 files changed, 158 insertions(+), 2 deletions(-)
> >  create mode 100644 drivers/clocksource/clps711x-clksrc.c
[...]
> > +#define CLPS711X_SYSCON1	(0x0100)
> > +#define CLPS711X_TC1D		(0x0300)
> > +#define CLPS711X_TC2D		(0x0340)
> 
> Alignment.

Alignment? Do not see any problems.
Wrong alignment in this case is the result of quote.

> > +static struct {
> > +	void __iomem	*tc1d;
> > +	int		irq;
> > +} *clps711x_clksrc;
> 
> You don't need to define this structure, the struct clock_event_device
> already contains a field with irq.

Thank you, I will know.

[...]
> > +static void clps711x_clockevent_set_mode(enum clock_event_mode mode,
> > +					 struct clock_event_device *evt)
> > +{
> > +	disable_irq(clps711x_clksrc->irq);
> 
> Do you really need to disable the interrupt to re-enable it right after ?

That was the original implementation. Perhaps the best solution is to disable
interrupt by CLOCK_EVT_MODE_SHUTDOWN and enable it back by CLOCK_EVT_MODE_RESUME.

> > +> +	switch (mode) {
> > +	case CLOCK_EVT_MODE_PERIODIC:
> > +		enable_irq(clps711x_clksrc->irq);
> > +		break;
> > +	case CLOCK_EVT_MODE_ONESHOT:
> > +		/* Not supported */
> > +	case CLOCK_EVT_MODE_SHUTDOWN:
> > +	case CLOCK_EVT_MODE_UNUSED:
> > +	case CLOCK_EVT_MODE_RESUME:
> > +		/* Left event sources disabled, no more interrupts appear */
> > +		break;
> > +	}
> > +}
> 
> I am not expert in interrupts, but it is possible this interrupt could
> be shared ?

Timer interrupt? No.

> There isn't a register to enable/disable the timer ?

Unfortunately, no. The timer is running always, except STDBY CPU state,
so we can control it only by enable/disable interrupt.

[...]
> > +static struct irqaction clps711x_timer_irq = {
> > +	.name		= "clps711x-timer",
> > +	.flags		= IRQF_TIMER | IRQF_IRQPOLL,
> 
> Why do you need IRQF_IRQPOOL ?

linux/interrupt.h:
 * IRQF_IRQPOLL - Interrupt is used for polling (only the interrupt that is
 *                registered first in an shared interrupt is considered for
 *                performance reasons)

I do not see need this flag, but in fact, this flag is traditionally used
in almost all subsystems of this type.

[...]
> > +	tmp = readl(syscon1);
> > +	/* TC1 in free running mode */
> > +	tmp &= ~SYSCON1_TC1M;
> > +	/* TC2 in prescale mode */
> > +	tmp |= SYSCON1_TC2M;
> > +	writel(tmp, syscon1);
> 
> Could you encapsulate these calls inside a static inline function with a
> more detailed comment please ?

Why static inline?
There is a comment line for each bit, You think this is not enough?

[...]
> > +void __init clps711x_clksrc_init(phys_addr_t phys_base, int irq)
> 
> Is this function called somewhere ? I don't see the function definition ?

It is defined in the patch [9/10] for non-DT case.

[...]

Thanks.

-- 
Alexander Shiyan <shc_work at mail.ru>