Skip to content

Latest commit

 

History

History
64 lines (53 loc) · 2.83 KB

README.md

File metadata and controls

64 lines (53 loc) · 2.83 KB
Raw

System Call Performance

The experiments in this folder will evaluate the costs of a noop system call and a function call on a Linux KVM VM and Unikraft KVM VM. The same benchmark ELF binary (sources under src/) is executed on both kernels. Because we want to measure a baseline: the time needed to enter and exit a system call handler, the kernels are instrumented with a noop system call (number 500) that only returns a constant value. Under the folder linux/ you can find the binary and build configuration of a patched vanilla Linux 5.11.0 kernel. You find also build instructions there.

Please note that these experiments do their measurement with the TSC. You need to make sure that your machine does not dynamically scale the CPU frequency while the experiment is running. This makes sure that the TSC clock can be considered stable. On recent Intel CPUs you can achieve pinning the CPU frequency by adding intel_pstate=disable to the Linux kernel arguments of the host (on Debian see /etc/default/grub). After a reboot, our script /tools/tunecpumax can help you enabling the userspace frequency scaling governor on your host and setting up the maximum non-turbo frequency permanently to your CPU.

Usage

  1. ./clone-deps.sh - Clones needed repositories from GitHub to build a Unikraft unikernel that can execute the benchmark ELF binary
  2. ./build-elfloader.sh - Compile the Unikraft unikernel for this experiment
  3. ./build-benchmark.sh - Builds the benchmark binary within src/
  4. ./build-fses.sh - Creates an initramdisk that only contains the benchmark program as init process for the Linux experiments.
  5. ./run.sh - Run the experiments, console outputs are stored under eval/parsed/ for later evaluation.
  6. ./parse.sh - Parses the console outputs, intermediate data within eval/
  7. ./process.sh - Computes the average, median, q1, and q3 over the data. It creates a table within results/

Alternatively, you can execute these steps with make all. All steps together take roughly 25 mins.

With ./clean.sh you can delete all compilation units, cloned repositories and intermediate measurement data. It keeps the final overview table of the experiment within results/.

Expected kernel panics

While the experiments are running, you may notice kernel panics for the Linux experiments:

[    1.317775] Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000000
[...]

This particular panic message (Attempted to kill init!) is expected. It happens because we do execute our benchmark program as system initialization process instead of /sbin/init. When our benchmark completed, the process exits and returns 0. As long as the exitcode is 0 (0x00000000), our experiment executed successfully.