2009-10-20 22:33:51

by Linus Walleij

[permalink] [raw]
Subject: [PATCH] Make MIPS dynamic clocksource/clockevent clock code generic v2

This moves the clocksource_set_clock() and clockevent_set_clock()
from the MIPS timer code into clockchips and clocksource where
it belongs. The patch was triggered by code posted by Mikael
Pettersson duplicating this code for the IOP ARM system. The
function signatures where altered slightly to fit into their
destination header files, unsigned int changed to u32 and inlined.

Signed-off-by: Linus Walleij <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Tested-by: Mikael Pettersson <[email protected]>
Reviewed-by: Ralf Baechle <[email protected]>
---
Changes v1->v2:
- Fixed Mikaels comments: spelling, terminology.
- Kept the functions inline: all uses and foreseen uses
are once per kernel and all are in __init or __cpuinit sections.
- Unable to break out common code - the code is not common and
implementing two execution paths will be more awkward.
- Hoping the tglx likes it anyway.
---
arch/mips/include/asm/time.h | 4 ----
arch/mips/kernel/time.c | 33 ---------------------------------
include/linux/clockchips.h | 22 ++++++++++++++++++++++
include/linux/clocksource.h | 23 ++++++++++++++++++++++-
4 files changed, 44 insertions(+), 38 deletions(-)

diff --git a/arch/mips/include/asm/time.h b/arch/mips/include/asm/time.h
index df6a430..e9b3f92 100644
--- a/arch/mips/include/asm/time.h
+++ b/arch/mips/include/asm/time.h
@@ -84,8 +84,4 @@ static inline int init_mips_clocksource(void)
#endif
}

-extern void clocksource_set_clock(struct clocksource *cs, unsigned int clock);
-extern void clockevent_set_clock(struct clock_event_device *cd,
- unsigned int clock);
-
#endif /* _ASM_TIME_H */
diff --git a/arch/mips/kernel/time.c b/arch/mips/kernel/time.c
index 1f467d5..fb74974 100644
--- a/arch/mips/kernel/time.c
+++ b/arch/mips/kernel/time.c
@@ -71,39 +71,6 @@ EXPORT_SYMBOL(perf_irq);

unsigned int mips_hpt_frequency;

-void __init clocksource_set_clock(struct clocksource *cs, unsigned int clock)
-{
- u64 temp;
- u32 shift;
-
- /* Find a shift value */
- for (shift = 32; shift > 0; shift--) {
- temp = (u64) NSEC_PER_SEC << shift;
- do_div(temp, clock);
- if ((temp >> 32) == 0)
- break;
- }
- cs->shift = shift;
- cs->mult = (u32) temp;
-}
-
-void __cpuinit clockevent_set_clock(struct clock_event_device *cd,
- unsigned int clock)
-{
- u64 temp;
- u32 shift;
-
- /* Find a shift value */
- for (shift = 32; shift > 0; shift--) {
- temp = (u64) clock << shift;
- do_div(temp, NSEC_PER_SEC);
- if ((temp >> 32) == 0)
- break;
- }
- cd->shift = shift;
- cd->mult = (u32) temp;
-}
-
/*
* This function exists in order to cause an error due to a duplicate
* definition if platform code should have its own implementation. The hook
diff --git a/include/linux/clockchips.h b/include/linux/clockchips.h
index 3a1dbba..9f9ec66 100644
--- a/include/linux/clockchips.h
+++ b/include/linux/clockchips.h
@@ -115,6 +115,28 @@ static inline unsigned long div_sc(unsigned long ticks, unsigned long nsec,
return (unsigned long) tmp;
}

+/**
+ * clockevent_set_clock - calculates an appropriate shift
+ * and mult values for a clockevent given a
+ * known clock frequency
+ * @dev: Clockevent device to initialize
+ * @hz: Clockevent clock frequency in Hz
+ */
+static inline void clockevent_set_clock(struct clock_event_device *dev, u32 hz)
+{
+ u64 temp;
+ u32 shift;
+
+ for (shift = 32; shift > 0; shift--) {
+ temp = (u64) hz << shift;
+ do_div(temp, NSEC_PER_SEC);
+ if ((temp >> 32) == 0)
+ break;
+ }
+ dev->shift = shift;
+ dev->mult = (u32) temp;
+}
+
/* Clock event layer functions */
extern unsigned long clockevent_delta2ns(unsigned long latch,
struct clock_event_device *evt);
diff --git a/include/linux/clocksource.h b/include/linux/clocksource.h
index 9ea40ff..fddd10e 100644
--- a/include/linux/clocksource.h
+++ b/include/linux/clocksource.h
@@ -257,6 +257,28 @@ static inline u32 clocksource_hz2mult(u32 hz, u32 shift_constant)
}

/**
+ * clocksource_set_clock - calculates an appropriate shift
+ * and mult values for a clocksource given a
+ * known clock frequency
+ * @cs: Clocksource to initialize
+ * @hz: Clocksource frequency in Hz
+ */
+static inline void clocksource_set_clock(struct clocksource *cs, u32 hz)
+{
+ u64 temp;
+ u32 shift;
+
+ for (shift = 32; shift > 0; shift--) {
+ temp = (u64) NSEC_PER_SEC << shift;
+ do_div(temp, hz);
+ if ((temp >> 32) == 0)
+ break;
+ }
+ cs->shift = shift;
+ cs->mult = (u32) temp;
+}
+
+/**
* clocksource_cyc2ns - converts clocksource cycles to nanoseconds
*
* Converts cycles to nanoseconds, using the given mult and shift.
@@ -268,7 +290,6 @@ static inline s64 clocksource_cyc2ns(cycle_t cycles, u32 mult, u32 shift)
return ((u64) cycles * mult) >> shift;
}

-
/* used to install a new clocksource */
extern int clocksource_register(struct clocksource*);
extern void clocksource_unregister(struct clocksource*);
--
1.6.2.5


2009-10-20 23:02:39

by Mikael Pettersson

[permalink] [raw]
Subject: Re: [PATCH] Make MIPS dynamic clocksource/clockevent clock code generic v2

Linus Walleij writes:
> This moves the clocksource_set_clock() and clockevent_set_clock()
> from the MIPS timer code into clockchips and clocksource where
> it belongs. The patch was triggered by code posted by Mikael
> Pettersson duplicating this code for the IOP ARM system. The
> function signatures where altered slightly to fit into their
> destination header files, unsigned int changed to u32 and inlined.
>
> Signed-off-by: Linus Walleij <[email protected]>
> Cc: Thomas Gleixner <[email protected]>
> Tested-by: Mikael Pettersson <[email protected]>
> Reviewed-by: Ralf Baechle <[email protected]>
> ---
> Changes v1->v2:
> - Fixed Mikaels comments: spelling, terminology.
> - Kept the functions inline: all uses and foreseen uses
> are once per kernel and all are in __init or __cpuinit sections.
> - Unable to break out common code - the code is not common and
> implementing two execution paths will be more awkward.
> - Hoping the tglx likes it anyway.

Very minor spelling nits below.

> --- a/include/linux/clockchips.h
> +++ b/include/linux/clockchips.h
> @@ -115,6 +115,28 @@ static inline unsigned long div_sc(unsigned long ticks, unsigned long nsec,
> return (unsigned long) tmp;
> }
>
> +/**
> + * clockevent_set_clock - calculates an appropriate shift
> + * and mult values for a clockevent given a

can't have 'an' and a plural form, so s/an //

> --- a/include/linux/clocksource.h
> +++ b/include/linux/clocksource.h
> @@ -257,6 +257,28 @@ static inline u32 clocksource_hz2mult(u32 hz, u32 shift_constant)
> }
>
> /**
> + * clocksource_set_clock - calculates an appropriate shift
> + * and mult values for a clocksource given a

ditto

2009-10-25 22:18:10

by Linus Walleij

[permalink] [raw]
Subject: Re: [PATCH] Make MIPS dynamic clocksource/clockevent clock code generic v2

2009/10/21 Linus Walleij <[email protected]>:

> Changes v1->v2:
> - Fixed Mikaels comments: spelling, terminology.
> - Kept the functions inline: all uses and foreseen uses
> ?are once per kernel and all are in __init or __cpuinit sections.
> - Unable to break out common code - the code is not common and
> ?implementing two execution paths will be more awkward.
> - Hoping the tglx likes it anyway.

Ping on tglx on this, I will uninline the functions if you think
it's better in the long run, just tell me, else can you pick this
patch?

Linus Walleij

2009-11-03 13:51:38

by Linus Walleij

[permalink] [raw]
Subject: Re: [PATCH] Make MIPS dynamic clocksource/clockevent clock code generic v2

2009/10/25 Linus Walleij <[email protected]>:
> 2009/10/21 Linus Walleij <[email protected]>:
>
>> Changes v1->v2:
>> - Fixed Mikaels comments: spelling, terminology.
>> - Kept the functions inline: all uses and foreseen uses
>> ?are once per kernel and all are in __init or __cpuinit sections.
>> - Unable to break out common code - the code is not common and
>> ?implementing two execution paths will be more awkward.
>> - Hoping the tglx likes it anyway.
>
> Ping on tglx on this, I will uninline the functions if you think
> it's better in the long run, just tell me, else can you pick this
> patch?

Ping again...

Linus Walleij

2009-11-03 15:32:26

by Thomas Gleixner

[permalink] [raw]
Subject: Re: [PATCH] Make MIPS dynamic clocksource/clockevent clock code generic v2

On Tue, 3 Nov 2009, Linus Walleij wrote:
> > Ping on tglx on this, I will uninline the functions if you think
> > it's better in the long run, just tell me, else can you pick this
> > patch?
>
> Ping again...

Still catching up with mail backlog. Will come to this in the next
days.

tglx

2009-11-10 17:03:54

by Thomas Gleixner

[permalink] [raw]
Subject: Re: [PATCH] Make MIPS dynamic clocksource/clockevent clock code generic v2

On Wed, 21 Oct 2009, Linus Walleij wrote:

> This moves the clocksource_set_clock() and clockevent_set_clock()
> from the MIPS timer code into clockchips and clocksource where
> it belongs. The patch was triggered by code posted by Mikael
> Pettersson duplicating this code for the IOP ARM system. The
> function signatures where altered slightly to fit into their
> destination header files, unsigned int changed to u32 and inlined.
>
> Signed-off-by: Linus Walleij <[email protected]>
> Cc: Thomas Gleixner <[email protected]>
> Tested-by: Mikael Pettersson <[email protected]>
> Reviewed-by: Ralf Baechle <[email protected]>
> ---
> Changes v1->v2:
> - Fixed Mikaels comments: spelling, terminology.
> - Kept the functions inline: all uses and foreseen uses
> are once per kernel and all are in __init or __cpuinit sections.

Not entirely true. We have the ability to dynamically switch on/off
clocksources/events which might also change the clock frequency and
requires recalculation of those factors.

> - Unable to break out common code - the code is not common and
> implementing two execution paths will be more awkward.

Come on. It's not rocket science to implement it as a common function
with two inline wrappers for clock sources and clock events.

> - Hoping the tglx likes it anyway.

I looked in more detail and I found a flaw which is already in the
MIPS implementation:

Both functions assume that the resulting shift/mult is chosen in way
that the input value to the runtime conversion functions is always
less than (1 << 32) nano seconds.

That limits NOHZ to sleeps of ~ 4 seconds. We had already discussions
about sleep times which are in the 10 to 20 seconds range and power
saving folks definitely want to extend this even further.

The tradeoff for the clock conversion is accuracy, but we don't want
to impose the maximum accuracy policy on everyone who wants to use the
runtime conversion factor setup functions.

I have a function coded which takes the maximum desired time span of
the conversion into account, but I need to think more about how we
define it. Runtime requested by the caller, some global Kconfig switch
or whatever is the best.

Thanks,

tglx