Hi !
in 3.12.6-rt9:
patch timers-do-not-raise-softirq-unconditionally.patch
fails compile test for:
CONFIG_PREEMPT_RT_BASE=y
# CONFIG_PREEMPT_RT_FULL is not set
full config with seed 0xE3A03BD0.
<snip>
CC kernel/timer.o
kernel/timer.c: In function 'run_local_timers':
kernel/timer.c:1463: error: implicit declaration of function 'spin_do_trylock'
kernel/timer.c:1474: error: implicit declaration of function 'rt_spin_unlock_after_trylock_in_irq'
make[1]: *** [kernel/timer.o] Error 1
make: *** [kernel] Error 2
... RT specific API not ifdef'ed, a brute force ifdefification "fix" below
But what this might actually point to is that testing of patches with RT off
(and other configs ?) is being neglected a bit. rt6 also had a problem (seed
was 0xBE96A834 file:kernel/rtmutex.c:__mutex_lock_check_stamp:1007)
maybe a set of minimum test-configs should be defined to atleast catch
build-failurs.
thx!
hofrat
Signed-off-by: Nicholas Mc Guire <der.herr@xxxxxxx>
---
kernel/timer.c | 8 +++++++-
1 files changed, 7 insertions(+), 1 deletions(-)
diff --git a/kernel/timer.c b/kernel/timer.c
index 46467be..88951cb 100644
--- a/kernel/timer.c
+++ b/kernel/timer.c
@@ -1452,9 +1452,13 @@ static void run_timer_softirq(struct softirq_action *h)
*/
void run_local_timers(void)
{
+#ifdef CONFIG_PREEMPT_RT_FULL
struct tvec_base *base = __this_cpu_read(tvec_bases);
+#endif
hrtimer_run_queues();
+
+#ifdef CONFIG_PREEMPT_RT_FULL
/*
* We can access this lockless as we are in the timer
* interrupt. If there are no timers queued, nothing to do in
@@ -1472,7 +1476,9 @@ void run_local_timers(void)
raise_softirq(TIMER_SOFTIRQ);
out:
rt_spin_unlock_after_trylock_in_irq(&base->lock);
-
+#else
+ raise_softirq(TIMER_SOFTIRQ);
+#endif
}
#ifdef __ARCH_WANT_SYS_ALARM
--
1.7.2.5
--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
