On Monday, February 24, 2014 12:05:39 PM Nicolas Pitre wrote:

On Mon, 24 Feb 2014, Tuukka Tikkanen wrote:

quoted

Sometimes (fairly often) when the cpuidle menu governor is making a decision
about idle state to enter the next timer for the cpu appears to expire in
the past. The menu governor expects the expiry to always be in the future
and in fact stores the time delta in an unsigned variable. However, when
the expiry is in the past, the value returned by tick_nohz_get_sleep_length
can be negative. This patch prevents using negative values, instead making
the governor return immediately similar to having latency requirement set
to 0.

Note: As with latency == 0, the return value is 0 with no check to see if
the state 0 has been disabled or not.

In your cover letter you mention some occurrences of the negative result 
being observed on x86.  That information is worth capturing in the 
commit log as well to distinguish between theoretical problems from 
actual observations.

In my opinion that applies to the other patches in the series too.  In all
cases it would be good to know whether or not the problems addressed are
theoretical or they can be reproduced in practice (and if so, then how), or
they already have been reported by somebody.

In the two latter cases the patches would be good candidates for -stable,
in particular.

Thanks,
Rafael