Re: [PATCH v4 00/36] hrtimer: Provide softirq context hrtimers

[Ingo Molnar]

* Ingo Molnar <mingo@xxxxxxxxxx> wrote:

> No serious changes intended, but please double check the end result once I've 
> pushed it out after local testing.

I made the fix below for the !NO_HZ_COMMON || !HIGH_RES_TIMERS case, which would 
fail with a build failure due to non-available __hrtimer_next_event_base().

These functions won't get called, so the fix I picked was to make the code 
unconditionally available.

That's the model I'd encourage going forward: we should reduce the !hrtimer and 
hrtimer differences, in fact we should eventually implement !hrtimers as a 
compatibility mode of the hrtimers code, with no separate #ifdeffery.

Thanks,

	Ingo

diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c
index 5bd0fe9b6402..ba4674e9adc2 100644
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -487,7 +487,6 @@ __next_base(struct hrtimer_cpu_base *cpu_base, unsigned int *active)
 #define for_each_active_base(base, cpu_base, active)	\
 	while ((base = __next_base((cpu_base), &(active))))
 
-#if defined(CONFIG_NO_HZ_COMMON) || defined(CONFIG_HIGH_RES_TIMERS)
 static ktime_t __hrtimer_next_event_base(struct hrtimer_cpu_base *cpu_base,
 					 unsigned int active,
 					 ktime_t expires_next)
@@ -539,7 +538,6 @@ static ktime_t __hrtimer_get_next_event(struct hrtimer_cpu_base *cpu_base,
 
 	return expires_next;
 }
-#endif
 
 static inline ktime_t hrtimer_update_base(struct hrtimer_cpu_base *base)
 {