Skip to content

AMDGPU: add parallel restore of BO content to accelerate restore #209

AMDGPU: add parallel restore of BO content to accelerate restore

AMDGPU: add parallel restore of BO content to accelerate restore #209

name: Actuated aarch64 test
on: [push, pull_request]
# Cancel any preceding run on the pull request.
concurrency:
group: actuated-test-${{ github.event.pull_request.number || github.ref }}
cancel-in-progress: ${{ github.ref != 'refs/heads/criu-dev' }}
jobs:
build:
# Actuated runners are not available in all repositories.
if: ${{ github.repository == 'checkpoint-restore/criu' }}
# The memory size and the number of CPUs can be freely selected.
# 3GB and 4 CPUs seems to be enough according to the result from 'vmmeter'.
runs-on: actuated-arm64-4cpu-3gb
strategy:
matrix:
target: [GCC=1, CLANG=1]
steps:
# https://gist.github.com/alexellis/1f33e581c75e11e161fe613c46180771#file-metering-gha-md
# vmmeter start
- name: Prepare arkade
uses: alexellis/arkade-get@master
with:
crane: latest
print-summary: false
- name: Install vmmeter
run: |
crane export --platform linux/arm64 ghcr.io/openfaasltd/vmmeter:latest | sudo tar -xvf - -C /usr/local/bin
- name: Run vmmeter
uses: self-actuated/vmmeter-action@master
# vmmeter end
- uses: actions/checkout@v4
- name: Run Tests ${{ matrix.target }}
# Following tests are failing on the actuated VMs:
# ./change_mnt_context --pidfile=change_mnt_context.pid --outfile=change_mnt_context.out
# 45: ERR: change_mnt_context.c:23: mount (errno = 22 (Invalid argument))
#
# In combination with '--remote-lazy-pages' following error occurs:
# 138: FAIL: maps05.c:84: Data corrupted at page 1639 (errno = 11 (Resource temporarily unavailable))
run: |
# The 'sched_policy00' needs the following:
sudo sysctl -w kernel.sched_rt_runtime_us=-1
# etc/hosts entry is needed for netns_lock_iptables
echo "127.0.0.1 localhost" | sudo tee -a /etc/hosts
sudo -E make -C scripts/ci local ${{ matrix.target }} RUN_TESTS=1 \
ZDTM_OPTS="-x zdtm/static/change_mnt_context -x zdtm/static/maps05"