On Mon, Aug 20, 2012 at 06:49:52PM -0700, John Stultz wrote:

On 08/20/2012 06:40 PM, Fengguang Wu wrote:

quoted

On Mon, Aug 20, 2012 at 06:15:00PM -0700, John Stultz wrote:

quoted

On 08/20/2012 06:10 PM, John Stultz wrote:

quoted

On 08/20/2012 06:04 PM, Fengguang Wu wrote:

quoted

Hi John,

The below oops happens in v3.5..v3.6-rc2 and it's bisected down
to commit
2a8c0883c ("time: Move xtime_nsec adjustment underflow handling
timekeeping_adjust").

However linux-next is working fine. Do you have any fixes not
yet sent to Linus?

Yea, there's a fix pending in tip/timers/urgent
(4e8b14526ca7fb046a81c94002c1c43b6fdf0e9b) to catch crazy values
from settimeofday or the cmos clock that might overflow a ktime_t.
Out of curiosity, how are you triggering/reproducing this?

Looking at the commit you pointed out, it seems more likely Ingo's
fix (1d17d17484d40f2d5b35c79518597a2b25296996) might be related, but
that should have landed in v3.6-rc2.  So I'm not sure.

Ah ok, looking back into the logs, I didn't find v3.6-rc2 being
tested. Sorry I should have confused it with another bug.

I'll test v3.6-rc2.

Ok. If v3.6-rc2 doesn't show the problem that would make me feel a
bit better that Ingo's fix is the one that is resolving this for
you.

Got the results: v3.6-rc2 tests fine. So it's a fixed problem in the upstream.

Sorry for the noises!

Thanks,
Fengguang