RAPL MSR support

The RAPL interface (Running Average Power Limit) is advertising the accumulated energy consumption of various power domains (e.g. CPU packages, DRAM, etc.).

The consumption is reported via MSRs (model specific registers) like MSR_PKG_ENERGY_STATUS for the CPU package power domain. These MSRs are 64 bits registers that represent the accumulated energy consumption in micro Joules.

Thanks to the MSR Filtering patch [1] not all MSRs are handled by KVM. Some of them can now be handled by the userspace (QEMU). It uses a mechanism called “MSR filtering” where a list of MSRs is given at init time of a VM to KVM so that a callback is put in place. The design of this patch uses only this mechanism for handling the MSRs between guest/host.

At the moment the following MSRs are involved:

#define MSR_RAPL_POWER_UNIT             0x00000606
#define MSR_PKG_POWER_LIMIT             0x00000610
#define MSR_PKG_ENERGY_STATUS           0x00000611
#define MSR_PKG_POWER_INFO              0x00000614

The *_POWER_UNIT, *_POWER_LIMIT, *_POWER INFO are part of the RAPL spec and specify the power limit of the package, provide range of parameter(min power, max power,..) and also the information of the multiplier for the energy counter to calculate the power. Those MSRs are populated once at the beginning by reading the host CPU MSRs and are given back to the guest 1:1 when requested.

The MSR_PKG_ENERGY_STATUS is a counter; it represents the total amount of energy consumed since the last time the register was cleared. If you multiply it with the UNIT provided above you’ll get the power in micro-joules. This counter is always increasing and it increases more or less faster depending on the consumption of the package. This counter is supposed to overflow at some point.

Each core belonging to the same Package reading the MSR_PKG_ENERGY_STATUS (i.e “rdmsr 0x611”) will retrieve the same value. The value represents the energy for the whole package. Whatever Core reading it will get the same value and a core that belongs to PKG-0 will not be able to get the value of PKG-1 and vice-versa.

High level implementation

In order to update the value of the virtual MSR, a QEMU thread is created. The thread is basically just an infinity loop that does:

  1. Snapshot of the time metrics of all QEMU threads (Time spent scheduled in Userspace and System)

  2. Snapshot of the actual MSR_PKG_ENERGY_STATUS counter of all packages where the QEMU threads are running on.

  3. Sleep for 1 second - During this pause the vcpu and other non-vcpu threads will do what they have to do and so the energy counter will increase.

  4. Repeat 2. and 3. and calculate the delta of every metrics representing the time spent scheduled for each QEMU thread and the energy spent by the packages during the pause.

  5. Filter the vcpu threads and the non-vcpu threads.

  6. Retrieve the topology of the Virtual Machine. This helps identify which vCPU is running on which virtual package.

  7. The total energy spent by the non-vcpu threads is divided by the number of vcpu threads so that each vcpu thread will get an equal part of the energy spent by the QEMU workers.

  8. Calculate the ratio of energy spent per vcpu threads.

  9. Calculate the energy for each virtual package.

  10. The virtual MSRs are updated for each virtual package. Each vCPU that belongs to the same package will return the same value when accessing the the MSR.

  11. Loop back to 1.

Ratio calculation

In Linux, a process has an execution time associated with it. The scheduler is dividing the time in clock ticks. The number of clock ticks per second can be found by the sysconf system call. A typical value of clock ticks per second is 100. So a core can run a process at the maximum of 100 ticks per second. If a package has 4 cores, 400 ticks maximum can be scheduled on all the cores of the package for a period of 1 second.

The /proc/[pid]/stat [2] is a sysfs file that can give the executed time of a process with the [pid] as the process ID. It gives the amount of ticks the process has been scheduled in userspace (utime) and kernel space (stime).

By reading those metrics for a thread, one can calculate the ratio of time the package has spent executing the thread.

Example:

A 4 cores package can schedule a maximum of 400 ticks per second with 100 ticks per second per core. If a thread was scheduled for 100 ticks between a second on this package, that means my thread has been scheduled for 1/4 of the whole package. With that, the calculation of the energy spent by the thread on this package during this whole second is 1/4 of the total energy spent by the package.

Usage

Currently this feature is only working on an Intel CPU that has the RAPL driver mounted and available in the sysfs. if not, QEMU fails at start-up.

This feature is activated with -accel kvm,rapl=true,rapl-helper-socket=/path/sock.sock

It is important that the socket path is the same as the one qemu-vmsr-helper is listening to.

qemu-vmsr-helper

The qemu-vmsr-helper is working very much like the qemu-pr-helper. Instead of making persistent reservation, qemu-vmsr-helper is here to overcome the CVE-2020-8694 which remove user access to the rapl msr attributes.

A socket communication is established between QEMU processes that has the RAPL MSR support activated and the qemu-vmsr-helper. A systemd service and socket activation is provided in contrib/systemd/qemu-vmsr-helper.(service/socket).

The systemd socket uses 600, like contrib/systemd/qemu-pr-helper.socket. The socket can be passed via SCM_RIGHTS by libvirt, or its permissions can be changed (e.g. 660 and root:kvm for a Debian system for example). Libvirt could also start a separate helper if needed. All in all, the policy is left to the user.

See the qemu-pr-helper documentation or manpage for further details.

Current Limitations

  • Works only on Intel host CPUs because AMD CPUs are using different MSR addresses.

  • Only the Package Power-Plane (MSR_PKG_ENERGY_STATUS) is reported at the moment.

References