Linux – Host CPU does not scale frequency when KVM guest needs it

clockcpu-frequencyfreebsdkvmlinux

Observation:
I have an HP server with an AMD dual core CPU (Turion II Neo N40L) which can scale frequencies from 800 to 1500 MHz. The frequency scaling works under FreeBSD 9 and under Ubuntu 12.04 with the Linux kernel 3.5.
However, when I put FreeBSD 9 in a KVM environment on top of Ubuntu the frequency scaling does not work. The guest (thus FreeBSD) does not detect the minimum and maximum frequencies and thus does not scale anything when CPU occupation gets higher. On the host (thus Ubuntu) the KVM process uses between 80 and 140 % of the CPU resource but no frequency scaling happens, the frequency stays at 800 MHz, although when I run any other process on the same Ubuntu box, the ondemand governor quickly scales the frequency to 1500 MHz!

Concern and question:
I don't understand how the CPU is perhaps virtualised, and if it is up to the guest to perform the proper scaling. Does it require some CPU features to be exposed to the guest for this to work?

Apendix:
The following Red Hat release note tends to suggest that frequency scaling out to work even in a virtualised environment (see chapter 6.2.2 and 6.2.3), thought the note fails to address which virtualisation technology this work with (kvm, xen, etc.?)

For information, the cpufreq-info output on Ubuntu is:

$ cpufreq-info
cpufrequtils 007: cpufreq-info (C) Dominik Brodowski 2004-2009
Report errors and bugs to cpufreq@vger.kernel.org, please.
analyzing CPU 0:
  driver: powernow-k8
  CPUs which run at the same hardware frequency: 0
  CPUs which need to have their frequency coordinated by software: 0
  maximum transition latency: 8.0 us.
  hardware limits: 800 MHz - 1.50 GHz
  available frequency steps: 1.50 GHz, 1.30 GHz, 1000 MHz, 800 MHz
  available cpufreq governors: conservative, ondemand, userspace, powersave, performance
  current policy: frequency should be within 800 MHz and 1.50 GHz.
                  The governor "ondemand" may decide which speed to use
                  within this range.
  current CPU frequency is 800 MHz.
  cpufreq stats: 1.50 GHz:14.79%, 1.30 GHz:1.07%, 1000 MHz:0.71%, 800 MHz:83.43%  (277433)
analyzing CPU 1:
  driver: powernow-k8
  CPUs which run at the same hardware frequency: 1
  CPUs which need to have their frequency coordinated by software: 1
  maximum transition latency: 8.0 us.
  hardware limits: 800 MHz - 1.50 GHz
  available frequency steps: 1.50 GHz, 1.30 GHz, 1000 MHz, 800 MHz
  available cpufreq governors: conservative, ondemand, userspace, powersave, performance
  current policy: frequency should be within 800 MHz and 1.50 GHz.
                  The governor "ondemand" may decide which speed to use
                  within this range.
  current CPU frequency is 800 MHz.
  cpufreq stats: 1.50 GHz:14.56%, 1.30 GHz:1.06%, 1000 MHz:0.79%, 800 MHz:83.59%  (384089)

The reason I want this feature to work is: save energy, run quieter (less hot) and also simple curiosity to understand better why this is not working and how to make it work.

Best Answer

I have found the solution thanks to the tip given by Nils and a nice article.

Tuning the ondemand CPU DVFS governor

The ondemand governor has a set of parameters to control when it is kicking the dynamic frequency scaling (or DVFS for dynamic voltage and frequency scaling). Those parameters are located under the sysfs tree: /sys/devices/system/cpu/cpufreq/ondemand/

One of this parameters is up_threshold which like the name suggest is a threshold (unit is % of CPU, I haven't find out though if this is per core or merged cores) above which the ondemand governor kicks in and start changing dynamically the frequency.

To change it to 50% (for example) using sudo is simple:
sudo bash -c "echo 50 > /sys/devices/system/cpu/cpufreq/ondemand/up_threshold"

If you are root, an even simpler command is possible:
echo 50 > /sys/devices/system/cpu/cpufreq/ondemand/up_threshold

Note: those changes will be lost after the next host reboot. You should add them to a configuration file that is read during boot, like /etc/init.d/rc.local on Ubuntu.

I have found out that my guest VM, although consuming a lot of CPU (80-140%) on the host was distributing the load on both cores, so no single core was above 95%, thus the CPU, to my exasperation, was staying at 800 MHz. Now with the above patch, the CPU dynamically changes it frequency per core much faster, which suits better my needs, 50% seems a better threshold for my guest usage, your mileage may vary.

Optionally, verify if you are using HPET

It is possible that some applicable which incorrectly implement timers might get affected by DVFS. This can be a problem on the host and/or guest environment, though the host can have some convoluted algorithm to try to minimise this. However, modern CPU have newer TSC (Time Stamp Counter) which are independent of the current CPU/core frequency, those are: constant (constant_tsc), invariant (invariant_tsc) or non-stop (nonstop_tsc), see this Chromium article about TSC resynchronisation for more information on each. So if your CPU is equipped with one of this TSC, you don't need to force HPET. To verify if your host CPU supports them, use a similar command (change the grep parameter to the corresponding CPU feature, here we test for the constant TSC):

$ grep constant_tsc /proc/cpuinfo

If you do not have one of this modern TSC, you should either:

Active HPET, this is described here after;
Not use CPU DVFS if you have any applications in the VM that rely on precise timing, which is the one recommended by Red Hat.

A safe solution is to enable HPET timers (see below for more details), they are slower to query than TSC ones (TSC are in the CPU, vs. HPET are in the motherboard) and perhaps not has precise (HPET >10MHz; TSC often the max CPU clock) but they are much more reliable especially in a DVFS configuration where each core could have a different frequency. Linux is clever enough to use the best available timer, it will rely on first the TSC, but if found too unreliable, it will use the HPET one. This work good on host (bare metal) systems, but due to not all information properly exported by the hypervisor, this is more of a challenge for the guest VM to detect badly behaving TSC. The trick is then to force to use HPET in the guest, although you would need the hypervisor to make this clock source available to the guests!

Below you can find how to configure and/or enable HPET on Linux and FreeBSD.

Linux HPET configuration

HPET, or high-precision event timer, is a hardware timer that you can find in most commodity PC since 2005. This timer can be used efficiently by modern OS (Linux kernel supports it since 2.6, stable support on FreeBSD since latest 9.x but was introduced in 6.3) to provide consistent timing invariably to CPU power management. It allows to build also easier tick-less scheduler implementations.

Basically HPET is like a safety barrier which even if the host has DVFS active, the host and guest timing events will be less affected.

There is a good article from IBM regarding enabling HPET, it explains how to verify which hardware timer your kernel is using, and which are available. I provide here a brief summary:

Checking the available hardware timer(s):
cat /sys/devices/system/clocksource/clocksource0/available_clocksource

Checking the current active timer:
cat /sys/devices/system/clocksource/clocksource0/current_clocksource

Simpler way to force usage of HPET if you have it available is to modify your boot loader to ask to enable it (since kernel 2.6.16). This configuration is distribution dependant, so please refer to your own distribution documentation to set it properly. You should enable hpet=enable or clocksource=hpet on the kernel boot line (this again depends on the kernel version or distribution, I did not find any coherent information).
This make sure that the guest is using the HPET timer.

Note: on my kernel 3.5, Linux seems to pick-up automatically the hpet timer.

FreeBSD guest HPET configuration

On FreeBSD one can check which timers are available by running:
sysctl kern.timecounter.choice

The currently chosen timer can be verified with:
sysctl kern.timecounter.hardware

FreeBSD 9.1 seems to automatically prefer HPET over other timer provider.

Todo: how to force HPET on FreeBSD.

Hypervisor HPET export

KVM seems to export HPET automatically when the host has support for it. However, for Linux guest they will prefer the other automatically exported clock which is kvm-clock (a paravirtualised version of the host TSC). Some people reports trouble with the preferred clock, your mileage may vary. If you want to force HPET in the guest, refer to the above section.

VirtualBox does not export the HPET clock to the guest by default, and there is no option to do so in the GUI. You need to use the command line and make sure the VM is powered off. the command is:

./VBoxManage modifyvm "VM NAME" --hpet on

If the guest keeps on selecting another source than HPET after the above change, please refer to the above section how to force the kernel to use HPET clock as a source.

Related Solutions

Linux – How to add CPU frequency governors to the Linux kernel

You'll have to find the code for that specific governor and add it to your kernel before recompiling it or you can also write the proper Makefile and compile the governor as a module. The code for the governor should be in drivers/cpufreq/. For example, for the lulzactive governor: drivers/cpufreq/cpufreq_lulzactive.c

Is it possible to set a constant lowest CPU frequency under the modern PSTATE driver

Well, well, turns out that the new pstate Intel driver is awesome, but first one needs to practice a bit of the ancient, lost art of reading the documentation.

I'll leave my question as it is because, judging by all the grief and frustration I see scattered all around the internet, I am not the first to have this issues.

The new CPU driver has lots of options, but I will restrict my explanation to something simple and, for me, more than enough and satisfactory. First of all:

sudo apt-get install linux-cpupower

(or the equivalent in your non-debian based distros)

There are now 2 behaviors (governors) with the names powersave and performance but that is quite an unfortunate naming scheme, because these governors have nothing to do with those that bear the same names in the old driver:

powersave means now variable frequency that depends on the load, i.e. this is essentially the old ondemand governor. You get to set the minimum and maximum frequency, and if that maximum frequency is set to the maximum frequency your CPU is able of (which I believe it's the default) then you don't save a crap. You may even crank up the minimum frequency to the second highest value, and the result will be the CPU nearly at full throttle 24/7 and the governor will still be named powersave. They should have named this governor VARIABLE or something similar, avoiding a lot of confusion among users and sparing the developers a lot of false reports of kernel bugs.
performance means here constant frequency, no matter the load, and this depends on what the user has set as maximum. When this governor is set, the mininum frequency is ignored and the CPU runs at the frequency you set as maximum. If as such you have set a very low frequency, then you will not see any special performance or anything, you just will get a slowed down CPU at constant frequency. So they'd better named this governor CONSTANT or something similar, sparing frustration to many people like me, used to the old scheme.

So, here are some examples that work like a charm, at least with kernel 4.14. I will use as minimum and maximum frequency the values for my CPU: 0.4 and 3.1 GHz. See yours with cpupower frequency-info

CONSTANT, LOWEST FREQUENCY IN ALL CORES

This IS what I wanted! We get that by setting the constant frequency governor and setting the lowest available frequency as the maximum:

sudo cpupower frequency-set -g performance
sudo cpupower frequency-set -u 400MHz

CONSTANT, HIGHEST FREQUENCY IN ALL CORES

(You see this asked very often for desktop computers where it makes sense, although there are also people out there willing to destroy their laptop fan)

sudo cpupower frequency-set -g performance
sudo cpupower frequency-set -u 3100MHz

VARIABLE FREQUENCY BETWEEN THE MINIMUM AND MAXIMUM POSSIBLE FREQUENCIES IN ALL CORES

(this was called ondemand with the old acpi-cpufreq driver)

sudo cpupower frequency-set -g powersave
sudo cpupower frequency-set -d 400MHz
sudo cpupower frequency-set -u 3100MHz

VARIABLE FREQUENCY BETWEEN THE MINIMUM AND A MODERATE FREQUENCY IN ALL CORES

(maybe because you want to get some more speed when required but you don't want to reach the maximum and hear the fan blowing like mad)

sudo cpupower frequency-set -g powersave
sudo cpupower frequency-set -d 400MHz
sudo cpupower frequency-set -u 1200MHz

And so on. It is very easy and works really well. You can set too a constant low frequency in one core where the heavy numerical stuff is running, while you leave a variable frequency in other core where you launch the more usual stuff (email, web browsing...). See taskset for more info.