[WIP] feat: add fault injection scripts

Makefile

@@ -6,6 +6,8 @@ DOCKER_PREFIX              := ${LIB_NAME}
 DOCKER_TAG                 := ${DOCKER_PREFIX}-yocto-builder
 DOCKER_SUSE_TAG            := ${DOCKER_PREFIX}-open-suse-env
 DOCKER_PYTHON_TAG          := ${DOCKER_PREFIX}-analyzer
+DOCKER_YOCTO_VM_TAG         := ${DOCKER_PREFIX}-yocto-vm
+DOCKER_SUSE_VM_TAG         := ${DOCKER_PREFIX}-suse-vm
 DOCKER_BUILD_VOLUME        := ${DOCKER_PREFIX}-yocto-build
 DOCKER_DOWNLOADS_VOLUME    := ${DOCKER_PREFIX}-yocto-downloads
 DOCKER_SSTATE_VOLUME       := ${DOCKER_PREFIX}-yocto-sstate

conf/packages.conf

@@ -59,6 +59,7 @@ IMAGE_INSTALL:append = " \
     jq \
     json-c \
     kbd \
+    kernel-devsrc \
     kmod \
     ldconfig \
     less \
@@ -92,6 +93,7 @@ IMAGE_INSTALL:append = " \
     libusb1 \
     libxcrypt \
     libxml2 \
+    linux-libc-headers \
     libyaml \
     logrotate \
     lshw \
@@ -140,6 +142,7 @@ IMAGE_INSTALL:append = " \
     python3-setuptools \
     readline \
     rpm \
+    ruby \
     rsync \
     rsyslog \
     screen \

docker/compose.yml

@@ -23,7 +23,34 @@ services:
       BUILD_DIR: ${BUILD_DIR}
       POKY_DIR: ${POKY_DIR}
       OS_IMAGE: ${OS_IMAGE}
-      VM_RAM: ${QEMU_VM_RAM}
+      USER: ${USER}
+      GROUP: ${GROUP}
+
+  imgtests-yocto-vm:
+    build:
+      dockerfile: docker/vm-yocto.dockerfile
+      args:
+        USER: ${USER}
+        GROUP: ${GROUP}
+        PASSWORD: ${PASSWORD}
+        POKY_DIR: ${POKY_DIR}
+    image: ${DOCKER_YOCTO_VM_TAG}
+    privileged: true
+    devices:
+      - /dev/kvm
+    volumes:
+      - ${DOCKER_BUILD_VOLUME}:${BUILD_DIR}
+      - ${DOCKER_DOWNLOADS_VOLUME}:${POKY_DIR}/downloads
+      - ${DOCKER_SSTATE_VOLUME}:${POKY_DIR}/sstate-cache
+      - ${HOST_CONF_PATH}/local.conf:${BUILD_DIR}/conf/local.conf
+      - ${HOST_CONF_PATH}/packages.conf:${BUILD_DIR}/conf/packages.conf
+      - ${HOST_CONF_PATH}/added_packages.conf:${BUILD_DIR}/conf/added_packages.conf
+      - ${HOST_LAYERS_PATH}/meta-image-tests:${POKY_DIR}/meta-image-tests
+      - ${HOST_SCRIPTS_PATH}/add-layers.sh:${POKY_DIR}/add-layers.sh
+    environment:
+      BUILD_DIR: ${BUILD_DIR}
+      POKY_DIR: ${POKY_DIR}
+      OS_IMAGE: ${OS_IMAGE}
       CPUS: ${CPUS}
       USER: ${USER}
       GROUP: ${GROUP}
@@ -37,6 +64,9 @@ services:
     networks:
       yocto-network:
         ipv4_address: ${YOCTO_ADDRESS}
+    depends_on:
+      imgtests-yocto:
+        condition: service_completed_successfully
 
   imgtests-suse-156:
     build:
@@ -51,13 +81,28 @@ services:
     environment:
       S_USER: ${S_USER}
       PASSWORD: ${PASSWORD}
-      VM_RAM: ${QEMU_VM_RAM}
-      CPUS: ${CPUS}
+    networks:
+      - yocto-network
+
+  imgtests-suse-vm:
+    build:
+      dockerfile: docker/vm-suse.dockerfile
+      args:
+        USER: ${USER}
+    image: ${DOCKER_SUSE_VM_TAG}
+    privileged: true
+    devices:
+      - /dev/kvm
+    volumes:
+      - ${DOCKER_OPENSUSE_VOLUME}:${SUSE_DIR}
+    environment:
       SSH_QEMU_PORT: ${SSH_TO_QEMU_PORT}
       SSH_SUSE_PORT: ${SSH_SUSE_PORT_156}
       SUSE_NE_PORT: ${SUSE_156_NE_PORT}
       DEFAULT_NE_PORT: ${DEFAULT_NE_PORT}
       IPERF3_PORT: ${IPERF3_PORT}
+      VM_RAM: ${QEMU_VM_RAM}
+      CPUS: ${CPUS}
     ports:
       - ${SSH_SUSE_PORT_156}:${SSH_SUSE_PORT_156}
     networks:

docker/image_builder.dockerfile

@@ -30,12 +30,14 @@ RUN groupadd -g 510 ${GROUP} && \
     echo "${USER}:${PASSWORD}" | chpasswd
 
 USER ${USER}
+
 RUN mkdir --parents ${POKY_DIR} && \
     git clone --depth 1 -b walnascar --recurse-submodules https://git.yoctoproject.org/poky ${POKY_DIR}
+
 WORKDIR ${POKY_DIR}
 
 COPY --chown=${USER}:${GROUP} layers ${POKY_DIR}
-COPY --chown=${USER}:${GROUP} scripts/entrypoint.sh ${POKY_DIR}/
-COPY --chown=${USER}:${GROUP} scripts/cmd_yocto.sh ${POKY_DIR}/
+COPY --chown=${USER}:${GROUP} scripts/entrypoint.sh ${POKY_DIR}
+COPY --chown=${USER}:${GROUP} scripts/cmd_yocto.sh ${POKY_DIR}
 
 ENTRYPOINT ["/home/user/poky/entrypoint.sh"]

docker/open-suse.dockerfile

@@ -7,9 +7,6 @@ ENV DEBIAN_FRONTEND=noninteractive \
 
 RUN apt-get update && \
     apt-get install -y --no-install-recommends \
-    qemu-system \
-    qemu-utils \
-    qemu-user-static \
     cloud-init \
     cloud-image-utils \
     wget \

docker/vm-suse.dockerfile

@@ -0,0 +1,19 @@
+FROM ubuntu:24.04
+
+ARG USER
+
+ENV DEBIAN_FRONTEND=noninteractive \
+    SUSE_DIR=/home/${USER}/suse
+
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends \
+    qemu-system \
+    qemu-utils \
+    qemu-user-static
+
+WORKDIR ${SUSE_DIR}
+
+COPY scripts/opensuse/start-vm.sh /start-vm.sh
+RUN chmod +x /start-vm.sh
+
+ENTRYPOINT ["/start-vm.sh"]

docker/vm-yocto.dockerfile

@@ -0,0 +1,38 @@
+FROM ubuntu:22.04
+
+ARG USER
+ARG GROUP
+ARG PASSWORD
+ARG POKY_DIR
+
+ENV DEBIAN_FRONTEND=noninteractive \
+    LANG=en_US.UTF-8
+
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends \
+        sudo wget git locales chrpath cpio diffstat gawk \
+        zstd liblz4-tool python3 python3-pip tar gcc make \
+        bzip2 file g++ patch python3-pexpect python3-git \
+        python3-jinja2 python3-subunit screen \
+        libtirpc-dev libtirpc3 pkg-config \
+        qemu qemu-system-x86 qemu-utils \
+        openssh-server && \
+    rm -rf /var/lib/apt/lists/* && \
+    locale-gen en_US.UTF-8
+
+RUN groupadd -g 510 ${GROUP} && \
+    groupadd -g 993 kvm && \
+    useradd -rm -d /home/${USER} -s /bin/bash -g ${GROUP} -u 1010 -G sudo,kvm ${USER} && \
+    echo "${USER} ALL=(ALL) NOPASSWD:ALL" >> /etc/sudoers.d/${USER} && \
+    echo "${USER}:${PASSWORD}" | chpasswd
+
+USER ${USER}
+
+RUN git clone --depth 1 -b walnascar --recurse-submodules https://git.yoctoproject.org/poky ${POKY_DIR}
+
+WORKDIR ${POKY_DIR}
+
+COPY --chown=${USER}:${GROUP} layers ${POKY_DIR}
+COPY --chown=${USER}:${GROUP} scripts/start-vm-yocto.sh ${POKY_DIR}
+
+ENTRYPOINT ["/home/user/poky/start-vm-yocto.sh"]

layers/meta-image-tests/recipes-kernel/linux/linux-yocto/fault-injection.cfg

@@ -5,3 +5,4 @@ CONFIG_FAIL_PAGE_ALLOC=y
 CONFIG_FAIL_MAKE_REQUEST=y
 CONFIG_FAIL_IO_TIMEOUT=y
 CONFIG_FAIL_FUTEX=y
+CONFIG_BPF_KPROBE_OVERRIDE=y

scripts/cmd_yocto.sh

@@ -5,7 +5,3 @@ set -e
 cd "${POKY_DIR}"
 ./add-layers.sh
 bitbake "${OS_IMAGE}"
-ROOTFS_IMAGE="${POKY_DIR}/build/tmp/deploy/images/qemux86-64/${OS_IMAGE}-qemux86-64.rootfs.ext4.qcow2"
-
-QB_SLIRP_OPT="-netdev user,id=net0,hostfwd=tcp::8080-:80,hostfwd=tcp::${SSH_QEMU_PORT}-:${SSH_TO_QEMU_PORT},hostfwd=tcp::${YOCTO_NE_PORT}-:${DEFAULT_NE_PORT},hostfwd=tcp:0.0.0.0:${IPERF3_PORT}-:${IPERF3_PORT},hostfwd=udp:0.0.0.0:${IPERF3_PORT}-:${IPERF3_PORT}" \
-    runqemu qemux86-64 slirp nographic "${ROOTFS_IMAGE}" qemuparams="-monitor tcp:0.0.0.0:4444,server,nowait -smp ${CPUS}"

scripts/fault-injection/chaosblade.sh

@@ -0,0 +1,85 @@
+#!/bin/bash
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+
+source "${SCRIPT_DIR}/common.sh"
+
+CHAOSBLADE_VERSION="${CHAOSBLADE_VERSION:-1.8.0}"
+CPU_PERCENT="${CHAOSBLADE_CPU_PERCENT:-20}"
+TIMEOUT="${CHAOSBLADE_TIMEOUT_SEC:-30}"
+NET_DELAY_MS="${CHAOSBLADE_NETWORK_DELAY_MS:-}"
+INTERFACE="${CHAOSBLADE_INTERFACE:-eth0}"
+
+IDS=()
+
+cleanup() {
+    local id
+    for id in "${IDS[@]}"; do
+        run "$id" <<<"sudo /opt/chaosblade/blade destroy $id" || warn "Failed to destroy $id"
+    done
+}
+trap cleanup EXIT
+
+install_chao() {
+    has blade && { info "chaosblade already installed on ${TARGET}"; return; }
+    [[ "$TARGET" == yocto ]] && die "not installed on ${TARGET}; add the chaosblade to the poky image"
+    info "Installing chaosblade v${CHAOSBLADE_VERSION} on ${TARGET}..."
+    run "$CHAOSBLADE_VERSION" <<'EOF' || die "Failed to install chaosblade"
+set -euo pipefail
+v=$1
+d=$(mktemp -d)
+trap 'rm -rf "$d"' EXIT
+cd "$d"
+sudo zypper -n install --no-recommends wget tar
+wget -q "https://github.com/chaosblade-io/chaosblade/releases/download/v${v}/chaosblade-${v}-linux_amd64.tar.gz"
+tar -xzf "chaosblade-${v}-linux_amd64.tar.gz"
+mkdir -p /opt/chaosblade
+cp -r "chaosblade-${v}-linux_amd64"/. /opt/chaosblade/
+chmod 755 /opt/chaosblade/blade
+ln -sf /opt/chaosblade/blade /usr/local/bin/blade
+EOF
+}
+
+create_exp() {
+    local out id
+    out=$(run "$@" <<'EOF'
+set -euo pipefail
+sudo /opt/chaosblade/blade check os "$1" "$2" > /dev/null 2>&1 || true
+sudo /opt/chaosblade/blade create "$1" "$2" "${@:3}"
+EOF
+) || return 1
+    id=$(printf '%s\n' "$out" | grep -o '"result":"[^"]*"' | head -1 | sed 's/"result":"\([^"]*\)"/\1/') || return 1
+    [[ -n "$id" ]] || return 1
+    printf '%s' "$id"
+}
+
+show()    { run "$1" <<<'sudo /opt/chaosblade/blade status "$1"'; }
+destroy() { run "$1" <<<"sudo /opt/chaosblade/blade destroy $1" || warn "Failed to destroy $1"; }
+
+main() {
+    install_chao
+
+    cpu=$(create_exp cpu fullload --cpu-percent "$CPU_PERCENT" --timeout "$TIMEOUT") ||
+        die "Failed to create CPU experiment"
+    IDS+=("$cpu")
+    show "$cpu"
+    info "Running CPU experiment for 5s..."
+    sleep 5
+
+    if [[ -n "$NET_DELAY_MS" ]]; then
+        if net=$(create_exp network delay --time "$NET_DELAY_MS" --interface "$INTERFACE" --timeout "$TIMEOUT"); then
+            IDS+=("$net")
+            show "$net"
+            info "Running network experiment for 5s..."
+            sleep 5
+            destroy "$net"
+        else
+            warn "Skipping network experiment"
+        fi
+    fi
+
+    destroy "$cpu"
+    show "$cpu"
+    info "Done."
+}
+main "$@"

scripts/fault-injection/chaosblade_vm.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+set -euo pipefail
+
+CHAOSBLADE_VERSION="${CHAOSBLADE_VERSION:-1.8.0}"
+CPU_PERCENT="${CHAOSBLADE_CPU_PERCENT:-20}"
+TIMEOUT="${CHAOSBLADE_TIMEOUT_SEC:-30}"
+NET_DELAY_MS="${CHAOSBLADE_NETWORK_DELAY_MS:-}"
+INTERFACE="${CHAOSBLADE_INTERFACE:-eth0}"
+
+IDS=()
+
+cleanup() {
+    local id
+    for id in "${IDS[@]}"; do
+        sudo /opt/chaosblade/blade destroy "$id" || true
+    done
+}
+trap cleanup EXIT
+
+install_chao() {
+    command -v blade && { echo "[INFO]  chaosblade already installed"; return; }
+    echo "[INFO]  Installing chaosblade v${CHAOSBLADE_VERSION}..."
+    d=$(mktemp -d)
+    trap 'rm -rf "$d"' EXIT
+    cd "$d"
+    sudo zypper -n install --no-recommends wget tar
+    wget -q "https://github.com/chaosblade-io/chaosblade/releases/download/v${CHAOSBLADE_VERSION}/chaosblade-${CHAOSBLADE_VERSION}-linux_amd64.tar.gz"
+    tar -xzf "chaosblade-${CHAOSBLADE_VERSION}-linux_amd64.tar.gz"
+    sudo mkdir -p /opt/chaosblade
+    sudo cp -r "chaosblade-${CHAOSBLADE_VERSION}-linux_amd64"/. /opt/chaosblade/
+    sudo chmod 755 /opt/chaosblade/blade
+    sudo ln -sf /opt/chaosblade/blade /usr/local/bin/blade
+}
+
+create_exp() {
+    sudo /opt/chaosblade/blade check os "$1" "$2" > /dev/null 2>&1 || true
+    local out id
+    out=$(sudo /opt/chaosblade/blade create "$1" "$2" "${@:3}")
+    id=$(printf '%s\n' "$out" | grep -o '"result":"[^"]*"' | head -1 | sed 's/"result":"\([^"]*\)"/\1/')
+    [[ -n "$id" ]] || return 1
+    printf '%s' "$id"
+}
+
+show()    { sudo /opt/chaosblade/blade status "$1"; }
+destroy() { sudo /opt/chaosblade/blade destroy "$1" || true; }
+
+main() {
+    install_chao
+
+    local cpu net
+    cpu=$(create_exp cpu fullload --cpu-percent "$CPU_PERCENT" --timeout "$TIMEOUT") ||
+        { echo "[ERROR] Failed to create CPU experiment"; exit 1; }
+    IDS+=("$cpu")
+    show "$cpu"
+    echo "[INFO]  Running CPU experiment for 5s..."
+    sleep 5
+
+    if [[ -n "$NET_DELAY_MS" ]]; then
+        if net=$(create_exp network delay --time "$NET_DELAY_MS" --interface "$INTERFACE" --timeout "$TIMEOUT"); then
+            IDS+=("$net")
+            show "$net"
+            echo "[INFO]  Running network experiment for 5s..."
+            sleep 5
+            destroy "$net"
+        else
+            echo "[WARN]  Skipping network experiment"
+        fi
+    fi
+
+    destroy "$cpu"
+    show "$cpu"
+    echo "[INFO]  Done."
+}
+main "$@"