Anubis

General

Please check fs/proc/kvm_new.c for debug and test

Most of the file change are in kernel/sched/fair.c

and arch/x86/kvm

The code is still prototype, Not fully clean and perfect coded.

the guest require the 4.14 guest

We will update more test case soon

Test

Preparation

---

1. Before start the VMs, in the host we need sudo echo 0xd300 > /proc/ct_offset.
2. Start the VMs(with correct kernel version).
3. While starting the VMs, do cat /proc/burrito_flag to open for 1 sec and cat /proc/burrito_flag again.
4. Use cat /proc/vcpu_list_show to check if success (all vCPUs value not zero)
5. Use virsh vcpupin to bind running VM's vCPU.
6. Create the CPU backgroud pressure sysbench cpu --time=99999999 --threads={Nr of vCPUs} run &

Example

---

sysbench fileio --file-total-size=10G --file-test-mode=seqrd --time=40 --max-requests=0 --file-extra-flags=direct --report-interval=5 --percentile=99 --threads=1 run

More details please refer to the paper Maximizing VMs' IO Performance on Overcommitted CPUs with Fairness

@inproceedings{10.1145/3620678.3624649, author = {Xing, Tong and Xiong, Cong and Ye, Chuan and Wei, Qi and Picorel, Javier and Barbalace, Antonio}, title = {Maximizing VMs' IO Performance on Overcommitted CPUs with Fairness}, year = {2023}, isbn = {9798400703874}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, url = {https://doi.org/10.1145/3620678.3624649}, doi = {10.1145/3620678.3624649}, pages = {93–108}, numpages = {16}, keywords = {IO performance, KVM, Linux, Overcommit, compute resources, fair scheduling, low-latency, virtualization}, location = {, Santa Cruz, CA, USA, }, series = {SoCC '23} }