time: Revert to calling clock_was_set_delayed() while in irq context
authorJohn Stultz <john.stultz@linaro.org>
Thu, 27 Mar 2014 23:30:49 +0000 (16:30 -0700)
committerIngo Molnar <mingo@kernel.org>
Fri, 28 Mar 2014 07:07:07 +0000 (08:07 +0100)
In commit 47a1b796306356f35 ("tick/timekeeping: Call
update_wall_time outside the jiffies lock"), we moved to calling
clock_was_set() due to the fact that we were no longer holding
the timekeeping or jiffies lock.

However, there is still the problem that clock_was_set()
triggers an IPI, which cannot be done from the timer's hard irq
context, and will generate WARN_ON warnings.

Apparently in my earlier testing, I'm guessing I didn't bump the
dmesg log level, so I somehow missed the WARN_ONs.

Thus we need to revert back to calling clock_was_set_delayed().

Signed-off-by: John Stultz <john.stultz@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1395963049-11923-1-git-send-email-john.stultz@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/time/timekeeping.c

index 0aa4ce8..5b40279 100644 (file)
@@ -1435,7 +1435,8 @@ void update_wall_time(void)
 out:
        raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
        if (clock_set)
-               clock_was_set();
+               /* Have to call _delayed version, since in irq context*/
+               clock_was_set_delayed();
 }
 
 /**