Rootcaused the bug to a deadlock in cpufreq and ondemand. Due to non-existent
ordering between cpu_hotplug lock and dbs_mutex. Basically a race condition
between cpu_down() and do_dbs_timer().

cpu_down() flow:
* cpu_down() call for CPU 1
* Takes hot plug lock
* Calls pre down notifier
*     cpufreq notifier handler calls cpufreq_driver_target() which takes
      cpu_hotplug lock again. OK as cpu_hotplug lock is recursive in same
      process context
* CPU 1 goes down
* Calls post down notifier
*     cpufreq notifier handler calls ondemand event stop which takes dbs_mutex

So, cpu_hotplug lock is taken before dbs_mutex in this flow.

do_dbs_timer is triggerred by a periodic timer event.
It first takes dbs_mutex and then takes cpu_hotplug lock in
cpufreq_driver_target().
Note the reverse order here compared to above. So, if this timer event happens
at right moment during cpu_down, system will deadlok.

Attached patch fixes the issue for both ondemand and conservative.
Signed-off-by: NVenkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: NDave Jones <davej@redhat.com>

Taking the cpu hotplug semaphore in a normal events workqueue
is unsafe because other tasks can wait for any workqueues with
it hold. This results in a deadlock.

Move the DBS timer into its own work queue which is not
affected by other work queue flushes to avoid this.

Has been acked by Venkatesh.

Cc: venkatesh.pallipadi@intel.com
Cc: cpufreq@lists.linux.org.uk
Signed-off-by: NAndi Kleen <ak@suse.de>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Assert that cpufreq_target is, at least, called with the minimum frequency
allowed by this policy, not something lower. It triggered problems on ARM.
Signed-off-by: NDominik Brodowski <linux@dominikbrodowski.net>

Keep the value of ignore_nice_load of the ondemand governor even after
the governor has been deselected and selected back. This is the behavior
of the other exported values of the ondemand governor and it's much more
user-friendly.
Signed-off-by: NEric Piel <eric.piel@tremplin-utc.net>
Acked-by: NVenkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: NDominik Brodowski <linux@dominikbrodowski.net>

Display a warning if the ondemand governor can not be selected due to a
transition latency of the cpufreq driver which is too long.
Signed-off-by: NEric Piel <eric.piel@tremplin-utc.net>
Acked-by: NVenkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: NDominik Brodowski <linux@dominikbrodowski.net>

Signed-off-by: NDave Jones <davej@redhat.com>

Semaphore to mutex conversion.

The conversion was generated via scripts, and the result was validated
automatically via a script as well.
Signed-off-by: NArjan van de Ven <arjan@infradead.org>
Signed-off-by: NIngo Molnar <mingo@elte.hu>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NDave Jones <davej@redhat.com>

The use of the 'ignore_nice' sysfs file is confusing to anyone using it.
This removes the sysfs file 'ignore_nice' and in its place creates a
'ignore_nice_load' entry that defaults to '0'; meaning nice'd processes
_are_ counted towards the 'business' calculation.

WARNING: this obvious breaks any userland tools that expected ignore_nice'
to exist, to draw attention to this fact it was concluded on the mailing
list that the entry should be removed altogether so the userland app breaks
and so the author can build simple to detect workaround.  Having said that
it seems currently very few tools even make use of this functionality; all
I could find was a Gentoo Wiki entry.
Signed-off-by: NAlexander Clouter <alex-kernel@digriz.org.uk>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NDave Jones <davej@redhat.com>

The problem is in the ondemand governor, there is a periodic measurement
of the CPU usage. This CPU usage is updated by the scheduler after every
tick (basically, by adding 1 either to "idle" or to "user" or to
"system"). So if the frequency of the governor is too high, the stat
will be meaningless (as mostly no number have changed).

So this patch checks that the measurements are separated by at least 10
ticks. It means that by default, stats will have about 5% error (20
ticks). Of course those numbers can be argued but, IMHO, they look sane.
The patch also includes a small clean-up to check more explictly the
result of the conversion from ns to µs being null.

Let's note that (on x86) this has never been really needed before 2.6.13
because HZ was always 1000. Now that HZ can be 100, some CPU might be
affected by this problem. For instance when HZ=100, the centrino ,which
has a 10µs transition latency, would lead to the governor allowing to
read stats every tick (10ms)!
Signed-off-by: NEric Piel <eric.piel@tremplin-utc.net>
Signed-off-by: NDave Jones <davej@redhat.com>

[PATCH] [5/5] ondemand governor default sampling downfactor as 1

Make default sampling downfactor 1.
This works better with earlier auto downscaling change in ondemand governor.
Signed-off-by: NVenkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: NDave Jones <davej@redhat.com>

[PATCH] [4/5] ondemand governor automatic downscaling

Here is a change of policy for the ondemand governor. The modification
concerns the frequency downscaling. Instead of decreasing to a lower
frequency when the CPU usage is under 20%, this new policy automatically
scales to the optimal frequency. The optimal frequency being the lowest
frequency which provides enough power to not trigger the upscaling policy.
Signed-off-by: NEric Piel <eric.piel@tremplin-utc.net>
Signed-off-by: NVenkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: NDave Jones <davej@redhat.com>

[PATCH] [3/5] ondemand,conservative governor idle_tick clean-up

Ondemand and conservative governor clean-up, it factorises the idle ticks 
measurement.
Signed-off-by: NEric Piel <eric.piel@tremplin-utc.net>
Signed-off-by: NVenkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: NDave Jones <davej@redhat.com>

[PATCH] [2/5] ondemand,conservative governor store the idle ticks for all cpus

Ondemand, conservative governor did not store prev_cpu_idle_up into 
prev_cpu_idle_down for other CPUs than the current CPU.
Signed-off-by: NEric Piel <eric.piel@tremplin-utc.net>
Signed-off-by: NVenkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: NDave Jones <davej@redhat.com>

[PATCH] [1/5] ondemand,conservative minor bug-fix and cleanup

Attached patch fixes some minor issues with Alexander's patch and related
cleanup in both ondemand and conservative governor.
Signed-off-by: NVenkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: NDave Jones <davej@redhat.com>

Adds support so that the cpufreq change stepping is no longer fixed at 5% and
can be changed dynamically by the user
Signed-off-by: NAlexander Clouter <alex-kernel@digriz.org.uk>
Signed-off-by: NDave Jones <davej@redhat.com>

Signed-off-by: NAlexander Clouter <alex-kernel@digriz.org.uk>
Signed-off-by: NDave Jones <davej@redhat.com>

Signed-off-by: NAlexander Clouter <alex-kernel@digriz.org.uk>
Signed-off-by: NDave Jones <davej@redhat.com>

This patch makes a needlessly global and EXPORT_SYMBOL'ed struct static.
Signed-off-by: NAdrian Bunk <bunk@stusta.de>
Signed-off-by: NDave Jones <davej@redhat.com>

Trivial ondemand governor clean-ups:
- change from sampling_rate_in_HZ() to the official function
usecs_to_jiffies().
- use for_each_online_cpu() to instead of using "if (cpu_online(i))"
Signed-off-by: NEric Piel <eric.piel@tremplin-utc.net>
Signed-off-by: NVenkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: NDominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: NDave Jones <davej@redhat.com>

Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.

Let it rip!