build: Release v32.1 (bug fix release)

Signed-off-by: Bo Chen <chen.bo@intel.com>

.github/workflows/build.yaml

@@ -13,7 +13,7 @@ jobs:
           - stable
           - beta
           - nightly
-          - "1.65"
+          - "1.62"
         target:
           - x86_64-unknown-linux-gnu
           - x86_64-unknown-linux-musl
@@ -34,22 +34,22 @@ jobs:
           override: true
 
       - name: Build (default features)
-        run: cargo rustc --locked --bin cloud-hypervisor -- -D warnings
+        run: cargo rustc --locked --bin cloud-hypervisor -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Build (kvm)
-        run: cargo rustc --locked --bin cloud-hypervisor --no-default-features --features "kvm"  -- -D warnings
+        run: cargo rustc --locked --bin cloud-hypervisor --no-default-features --features "kvm"  -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Build (default features + tdx)
-        run: cargo rustc --locked --bin cloud-hypervisor --features "tdx" -- -D warnings
+        run: cargo rustc --locked --bin cloud-hypervisor --features "tdx" -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Build (default features + guest_debug)
-        run: cargo rustc --locked --bin cloud-hypervisor --features "guest_debug" -- -D warnings
+        run: cargo rustc --locked --bin cloud-hypervisor --features "guest_debug" -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Build (mshv)
-        run: cargo rustc --locked --bin cloud-hypervisor --no-default-features --features "mshv"  -- -D warnings
+        run: cargo rustc --locked --bin cloud-hypervisor --no-default-features --features "mshv"  -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Build (mshv + kvm)
-        run: cargo rustc --locked --bin cloud-hypervisor --no-default-features --features "mshv,kvm"  -- -D warnings
+        run: cargo rustc --locked --bin cloud-hypervisor --no-default-features --features "mshv,kvm"  -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Release Build (default features)
         run: cargo build --locked --all --release --target=${{ matrix.target }}

.github/workflows/docker-image.yaml

@@ -7,6 +7,10 @@ on:
   pull_request:
     paths: resources/Dockerfile
 
+env:
+  REGISTRY: ghcr.io
+  IMAGE_NAME: ${{ github.repository }}
+
 jobs:
   main:
     runs-on: ubuntu-latest
@@ -20,19 +24,19 @@ jobs:
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v1
 
-      - name: Login to DockerHub
-        if: ${{ github.event_name == 'push' }}
-        uses: docker/login-action@v1 
+      - name: Login to ghcr
+        uses: docker/login-action@v2
         with:
-          username: ${{ secrets.DOCKERHUB_USERNAME }}
-          password: ${{ secrets.DOCKERHUB_TOKEN }}
+          registry: ${{ env.REGISTRY }}
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
 
       - name: Docker meta
         id: meta
-        uses: docker/metadata-action@v3
+        uses: docker/metadata-action@v4
         with:
-          # list of Docker images to use as base name for tags
-          images: cloudhypervisor/dev
+          images: ${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}
           # generate Docker tags based on the following events/attributes
           tags: |
             type=raw,value={{date 'YYYYMMDD'}}-0

.github/workflows/quality.yaml

@@ -13,15 +13,24 @@ jobs:
         rust:
           - stable
         target:
-          - x86_64-unknown-linux-gnu
           - aarch64-unknown-linux-gnu
+          - aarch64-unknown-linux-musl
+          - x86_64-unknown-linux-gnu
+          - x86_64-unknown-linux-musl
+
         experimental: [false]
         include:
+          - rust: beta
+            target: aarch64-unknown-linux-gnu
+            experimental: true
+          - rust: beta
+            target: aarch64-unknown-linux-musl
+            experimental: true
           - rust: beta
             target: x86_64-unknown-linux-gnu
             experimental: true
           - rust: beta
-            target: aarch64-unknown-linux-gnu
+            target: x86_64-unknown-linux-musl
             experimental: true
     steps:
       - name: Code checkout
@@ -42,7 +51,7 @@ jobs:
         run: |
           set -e
           commits=$(git rev-list origin/${{ github.base_ref }}..${{ github.sha }})
-          for commit in $commits; do git checkout $commit; cargo check --tests --all --target=${{ matrix.target }}; done
+          for commit in $commits; do git checkout $commit; cargo check --tests --examples --all --target=${{ matrix.target }}; done
           git checkout ${{ github.sha }}
 
       - name: Formatting (rustfmt)
@@ -53,28 +62,28 @@ jobs:
         with:
           use-cross: ${{ matrix.target != 'x86_64-unknown-linux-gnu' }}
           command: clippy
-          args: --locked --all --all-targets --no-default-features --tests --features "kvm" -- -D warnings
+          args: --target=${{ matrix.target }} --locked --all --all-targets --no-default-features --tests --examples --features "kvm" -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Clippy (default features)
         uses: actions-rs/cargo@v1
         with:
           use-cross: ${{ matrix.target != 'x86_64-unknown-linux-gnu' }}
           command: clippy
-          args: --locked --all --all-targets --tests -- -D warnings
+          args: --target=${{ matrix.target }} --locked --all --all-targets --tests --examples -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Clippy (default features + guest_debug)
         uses: actions-rs/cargo@v1
         with:
           use-cross: ${{ matrix.target != 'x86_64-unknown-linux-gnu' }}
           command: clippy
-          args: --locked --all --all-targets --tests --features "guest_debug" -- -D warnings
+          args: --target=${{ matrix.target }} --locked --all --all-targets --tests --examples --features "guest_debug" -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Clippy (default features + tracing)
         uses: actions-rs/cargo@v1
         with:
           use-cross: ${{ matrix.target != 'x86_64-unknown-linux-gnu' }}
           command: clippy
-          args: --locked --all --all-targets --tests --features "tracing" -- -D warnings
+          args: --target=${{ matrix.target }} --locked --all --all-targets --tests --examples --features "tracing" -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Clippy (mshv)
         if: ${{ matrix.target == 'x86_64-unknown-linux-gnu' }}
@@ -82,7 +91,7 @@ jobs:
         with:
           use-cross: ${{ matrix.target != 'x86_64-unknown-linux-gnu' }}
           command: clippy
-          args: --locked --all --all-targets --no-default-features --tests --features "mshv" -- -D warnings
+          args: --target=${{ matrix.target }} --locked --all --all-targets --no-default-features --tests --examples --features "mshv" -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Clippy (mshv + kvm)
         if: ${{ matrix.target == 'x86_64-unknown-linux-gnu' }}
@@ -90,7 +99,7 @@ jobs:
         with:
           use-cross: ${{ matrix.target != 'x86_64-unknown-linux-gnu' }}
           command: clippy
-          args: --locked --all --all-targets --no-default-features --tests --features "mshv,kvm" -- -D warnings
+          args: --target=${{ matrix.target }} --locked --all --all-targets --no-default-features --tests --examples --features "mshv,kvm" -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Clippy (kvm + tdx)
         if: ${{ matrix.target == 'x86_64-unknown-linux-gnu' }}
@@ -98,7 +107,7 @@ jobs:
         with:
           use-cross: ${{ matrix.target != 'x86_64-unknown-linux-gnu' }}
           command: clippy
-          args: --locked --all --all-targets --no-default-features --tests --features "tdx,kvm" -- -D warnings
+          args: --target=${{ matrix.target }} --locked --all --all-targets --no-default-features --tests --examples --features "tdx,kvm" -- -D warnings -D clippy::undocumented_unsafe_blocks
 
       - name: Check build did not modify any files
         run: test -z "$(git status --porcelain)"

.github/workflows/release.yaml

@@ -16,29 +16,29 @@ jobs:
       - name: Install Rust toolchain (x86_64-unknown-linux-gnu)
         uses: actions-rs/toolchain@v1
         with:
-          toolchain: "1.65"
+          toolchain: "1.67.1"
           target: x86_64-unknown-linux-gnu
       - name: Install Rust toolchain (x86_64-unknown-linux-musl)
         uses: actions-rs/toolchain@v1
         with:
-          toolchain: "1.65"
+          toolchain: "1.67.1"
           target: x86_64-unknown-linux-musl
       - name: Build
         uses: actions-rs/cargo@v1
         with:
-            toolchain: "1.65"
-            command: build
-            args: --all --release --target=x86_64-unknown-linux-gnu
+          toolchain: "1.67.1"
+          command: build
+          args: --all --release --no-default-features --features "kvm,mshv" --target=x86_64-unknown-linux-gnu
       - name: Static Build
         uses: actions-rs/cargo@v1
         with:
-            toolchain: "1.65"
-            command: build
-            args: --all --release --target=x86_64-unknown-linux-musl
+          toolchain: "1.67.1"
+          command: build
+          args: --all --release --no-default-features --features "kvm,mshv" --target=x86_64-unknown-linux-musl
       - name: Install Rust toolchain (aarch64-unknown-linux-musl)
         uses: actions-rs/toolchain@v1
         with:
-          toolchain: "1.65"
+          toolchain: "1.67.1"
           target: aarch64-unknown-linux-musl
           override: true
       - name: Create Release

.gitignore

@@ -5,3 +5,4 @@
 **/Cargo.lock
 **/rusty-tags.vi
 /rpm/SOURCES
+/.vscode

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "cloud-hypervisor"
-version = "28.4.0"
+version = "32.1.0"
 authors = ["The Cloud Hypervisor Authors"]
 edition = "2021"
 default-run = "cloud-hypervisor"
@@ -12,10 +12,10 @@ homepage = "https://github.com/cloud-hypervisor/cloud-hypervisor"
 # Keep in sync with version in .github/workflows/build.yaml
 # Policy on MSRV (see #4318):
 # Can only be bumped by:
-# a.) A dependency requires it, 
+# a.) A dependency requires it,
 # b.) If we want to use a new feature and that MSRV is at least 6 months old,
 # c.) There is a security issue that is addressed by the toolchain update.
-rust-version = "1.65"
+rust-version = "1.62"
 
 [profile.release]
 lto = true
@@ -23,29 +23,32 @@ codegen-units = 1
 opt-level = "s"
 strip = true
 
+[profile.profiling]
+inherits = "release"
+strip = false
+debug = true
+
 [dependencies]
-anyhow = "1.0.66"
+anyhow = "1.0.70"
 api_client = { path = "api_client" }
-clap = { version = "4.0.29", features = ["wrap_help","cargo","string"] }
+argh = "0.1.9"
+dhat = { version = "0.3.2", optional = true }
 epoll = "4.3.1"
 event_monitor = { path = "event_monitor" }
 hypervisor = { path = "hypervisor" }
-libc = "0.2.138"
+libc = "0.2.139"
 log = { version = "0.4.17", features = ["std"] }
 option_parser = { path = "option_parser" }
 seccompiler = "0.3.0"
-serde_json = "1.0.89"
-signal-hook = "0.3.14"
-thiserror = "1.0.37"
+serde_json = "1.0.96"
+signal-hook = "0.3.15"
+thiserror = "1.0.39"
 tpm = { path = "tpm"}
 tracer = { path = "tracer" }
 vmm = { path = "vmm" }
 vmm-sys-util = "0.11.0"
 vm-memory = "0.10.0"
 
-[build-dependencies]
-clap = { version = "4.0.29", features = ["cargo"] }
-
 # List of patched crates
 [patch.crates-io]
 kvm-bindings = { git = "https://github.com/cloud-hypervisor/kvm-bindings", branch = "ch-v0.6.0-tdx" }
@@ -53,15 +56,16 @@ kvm-ioctls = { git = "https://github.com/rust-vmm/kvm-ioctls", branch = "main" }
 versionize_derive = { git = "https://github.com/cloud-hypervisor/versionize_derive", branch = "ch" }
 
 [dev-dependencies]
-dirs = "4.0.0"
+dirs = "5.0.0"
 net_util = { path = "net_util" }
-once_cell = "1.16.0"
-serde_json = "1.0.89"
+once_cell = "1.17.1"
+serde_json = "1.0.96"
 test_infra = { path = "test_infra" }
 wait-timeout = "0.2.0"
 
 [features]
 default = ["kvm"]
+dhat-heap = ["dhat"] # For heap profiling
 guest_debug = ["vmm/guest_debug"]
 kvm = ["vmm/kvm"]
 mshv = ["vmm/mshv"]
@@ -70,7 +74,6 @@ tracing = ["vmm/tracing", "tracer/tracing"]
 
 [workspace]
 members = [
-    "acpi_tables",
     "api_client",
     "arch",
     "block_util",
@@ -87,7 +90,6 @@ members = [
     "serial_buffer",
     "test_infra",
     "tracer",
-    "vfio_user",
     "vhdx",
     "vhost_user_block",
     "vhost_user_net",

Jenkinsfile

@@ -1,6 +1,9 @@
 def runWorkers = true
 pipeline {
     agent none
+    options {
+        timeout(time: 4, unit: 'HOURS')
+    }
     stages {
         stage('Early checks') {
             agent { node { label 'built-in' } }
@@ -10,29 +13,16 @@ pipeline {
                         checkout scm
                     }
                 }
-                stage('Check for documentation only changes') {
+                stage('Check if worker build can be skipped') {
                     when {
                         expression {
-                            return docsFileOnly()
+                            return skipWorkerBuild()
                         }
                     }
                     steps {
                         script {
                             runWorkers = false
-                            echo 'Documentation only changes, no need to run the CI'
-                        }
-                    }
-                }
-                stage('Check for fuzzer files only changes') {
-                    when {
-                        expression {
-                            return fuzzFileOnly()
-                        }
-                    }
-                    steps {
-                        script {
-                            runWorkers = false
-                            echo 'Fuzzer cargo files only changes, no need to run the CI'
+                            echo 'No changes requring a build'
                         }
                     }
                 }
@@ -202,52 +192,52 @@ pipeline {
                         }
                     }
                 }
-                stage('Worker build - Windows guest') {
-                    agent { node { label 'jammy' } }
-                    when {
-                        beforeAgent true
-                        expression {
-                            return runWorkers
-                        }
-                    }
-                    environment {
-                            AZURE_CONNECTION_STRING = credentials('46b4e7d6-315f-4cc1-8333-b58780863b9b')
-                    }
-                    stages {
-                        stage('Checkout') {
-                            steps {
-                                checkout scm
-                            }
-                        }
-                        stage('Install azure-cli') {
-                            steps {
-                                installAzureCli('jammy', 'amd64')
-                            }
-                        }
-                        stage('Download assets') {
-                            steps {
-                                sh "mkdir ${env.HOME}/workloads"
-                                sh 'az storage blob download --container-name private-images --file "$HOME/workloads/windows-server-2019.raw" --name windows-server-2019.raw --connection-string "$AZURE_CONNECTION_STRING"'
-                            }
-                        }
-                        stage('Run Windows guest integration tests') {
-                            options {
-                                timeout(time: 1, unit: 'HOURS')
-                            }
-                            steps {
-                                sh 'scripts/dev_cli.sh tests --integration-windows'
-                            }
-                        }
-                        stage('Run Windows guest integration tests for musl') {
-                            options {
-                                timeout(time: 1, unit: 'HOURS')
-                            }
-                            steps {
-                                sh 'scripts/dev_cli.sh tests --integration-windows --libc musl'
-                            }
-                        }
-                    }
-                }
+                // stage('Worker build - Windows guest') {
+                //     agent { node { label 'jammy' } }
+                //     when {
+                //         beforeAgent true
+                //         expression {
+                //             return runWorkers
+                //         }
+                //     }
+                //     environment {
+                //             AZURE_CONNECTION_STRING = credentials('46b4e7d6-315f-4cc1-8333-b58780863b9b')
+                //     }
+                //     stages {
+                //         stage('Checkout') {
+                //             steps {
+                //                 checkout scm
+                //             }
+                //         }
+                //         stage('Install azure-cli') {
+                //             steps {
+                //                 installAzureCli('jammy', 'amd64')
+                //             }
+                //         }
+                //         stage('Download assets') {
+                //             steps {
+                //                 sh "mkdir ${env.HOME}/workloads"
+                //                 sh 'az storage blob download --container-name private-images --file "$HOME/workloads/windows-server-2022-amd64-2.raw" --name windows-server-2022-amd64-2.raw --connection-string "$AZURE_CONNECTION_STRING"'
+                //             }
+                //         }
+                //         stage('Run Windows guest integration tests') {
+                //             options {
+                //                 timeout(time: 1, unit: 'HOURS')
+                //             }
+                //             steps {
+                //                 sh 'scripts/dev_cli.sh tests --integration-windows'
+                //             }
+                //         }
+                //         stage('Run Windows guest integration tests for musl') {
+                //             options {
+                //                 timeout(time: 1, unit: 'HOURS')
+                //             }
+                //             steps {
+                //                 sh 'scripts/dev_cli.sh tests --integration-windows --libc musl'
+                //             }
+                //         }
+                //     }
+                // }
                 stage('Worker build - Metrics') {
                     agent { node { label 'jammy-metrics' } }
                     when {
@@ -281,6 +271,113 @@ pipeline {
                         }
                     }
                 }
+                stage('Worker build - Rate Limiter') {
+                    agent { node { label 'focal-metrics' } }
+                    when {
+                        branch 'main'
+                        beforeAgent true
+                        expression {
+                            return runWorkers
+                        }
+                    }
+                    stages {
+                        stage('Checkout') {
+                            steps {
+                                checkout scm
+                            }
+                        }
+                        stage('Run rate-limiter integration tests') {
+                            options {
+                                timeout(time: 10, unit: 'MINUTES')
+                            }
+                            steps {
+                                sh 'scripts/dev_cli.sh tests --integration-rate-limiter'
+                            }
+                        }
+                    }
+                }
+                stage('Worker build - SGX') {
+                    agent { node { label 'jammy-sgx' } }
+                    when {
+                        beforeAgent true
+                        allOf {
+                            branch 'main'
+                            expression {
+                                return runWorkers
+                            }
+                        }
+                    }
+                    stages {
+                        stage('Checkout') {
+                            steps {
+                                checkout scm
+                            }
+                        }
+                        stage('Run SGX integration tests') {
+                            options {
+                                timeout(time: 1, unit: 'HOURS')
+                            }
+                            steps {
+                                sh 'scripts/dev_cli.sh tests --integration-sgx'
+                            }
+                        }
+                        stage('Run SGX integration tests for musl') {
+                            options {
+                                timeout(time: 1, unit: 'HOURS')
+                            }
+                            steps {
+                                sh 'scripts/dev_cli.sh tests --integration-sgx --libc musl'
+                            }
+                        }
+                    }
+                    post {
+                        always {
+                            sh "sudo chown -R jenkins.jenkins ${WORKSPACE}"
+                            deleteDir()
+                        }
+                    }
+                }
+                stage('Worker build - VFIO') {
+                    agent { node { label 'jammy-vfio' } }
+                    when {
+                        beforeAgent true
+                        allOf {
+                            branch 'main'
+                            expression {
+                                return runWorkers
+                            }
+                        }
+                    }
+                    stages {
+                        stage('Checkout') {
+                            steps {
+                                checkout scm
+                            }
+                        }
+                        stage('Run VFIO integration tests') {
+                            options {
+                                timeout(time: 1, unit: 'HOURS')
+                            }
+                            steps {
+                                sh 'scripts/dev_cli.sh tests --integration-vfio'
+                            }
+                        }
+                        stage('Run VFIO integration tests for musl') {
+                            options {
+                                timeout(time: 1, unit: 'HOURS')
+                            }
+                            steps {
+                                sh 'scripts/dev_cli.sh tests --integration-vfio --libc musl'
+                            }
+                        }
+                    }
+                    post {
+                        always {
+                            sh "sudo chown -R jenkins.jenkins ${WORKSPACE}"
+                            deleteDir()
+                        }
+                    }
+                }
             }
         }
     }
@@ -325,24 +422,31 @@ def installAzureCli(distro, arch) {
     sh 'sudo apt install -y azure-cli'
 }
 
-def boolean docsFileOnly() {
+def boolean skipWorkerBuild() {
     if (env.CHANGE_TARGET == null) {
         return false
     }
 
-    return sh(
+    if (sh(
         returnStatus: true,
         script: "git diff --name-only origin/${env.CHANGE_TARGET}... | grep -v '\\.md'"
-    ) != 0
-}
-
-def boolean fuzzFileOnly() {
-    if (env.CHANGE_TARGET == null) {
-        return false
+    ) != 0) {
+        return true
     }
 
-    return sh(
+    if (sh(
         returnStatus: true,
         script: "git diff --name-only origin/${env.CHANGE_TARGET}... | grep -v -E 'fuzz/'"
-    ) != 0
+    ) != 0) {
+        return true
+    }
+
+    if (sh(
+        returnStatus: true,
+        script: "git diff --name-only origin/${env.CHANGE_TARGET}... | grep -v -E '.github/'"
+    ) != 0) {
+        return true
+    }
+
+    return false
 }

MAINTAINERS.md

@@ -2,6 +2,7 @@
 
 - Sebastien Boeuf - @sboeuf
 - Robert Bradford - @rbradford
+- Bo Chen - @likebreath
 - Samuel Ortiz - @sameo
 - Wei Liu - @liuw
 - Michael Zhao - @michael2012z

README.md

@@ -69,10 +69,12 @@ Cloud Hypervisor supports `64-bit Linux` and Windows 10/Windows Server 2019.
 
 # 2. Getting Started
 
-The following sections describe how to build and run Cloud Hypervisor on the
-`x86-64` platform. For getting started on the `AArch64` platform, please refer
-to the
-[AArch64 documentation](docs/arm64.md).
+The following sections describe how to build and run Cloud Hypervisor.
+
+## Prerequisites for AArch64
+
+- AArch64 servers (recommended) or development boards equipped with the GICv3
+  interrupt controller.
 
 ## Host OS
 
@@ -105,16 +107,13 @@ do not wish to use the pre-built binaries.
 
 ## Booting Linux
 
-The instructions below are for the `x86-64` platform. For `AArch64` please see
-the [AArch64 specific documentation](docs/arm64.md).
-
-Cloud Hypervisor supports direct kernel boot (if the kernel is built with PVH
-support) or booting via a firmware (either [Rust Hypervisor
+Cloud Hypervisor supports direct kernel boot (the x86-64 kernel requires the kernel
+built with PVH support) or booting via a firmware (either [Rust Hypervisor
 Firmware](https://github.com/cloud-hypervisor/rust-hypervisor-firmware) or an
-edk2 UEFI firmware called `CLOUDHV`.)
+edk2 UEFI firmware called `CLOUDHV` / `CLOUDHV_EFI`.)
 
 Binary builds of the firmware files are available for the latest release of
-[Rust Hyperivor
+[Rust Hypervisor
 Firmware](https://github.com/cloud-hypervisor/rust-hypervisor-firmware/releases/latest)
 and [our edk2
 repository](https://github.com/cloud-hypervisor/edk2/releases/latest)
@@ -125,7 +124,7 @@ may be required.
 ### Firmware Booting
 
 Cloud Hypervisor supports booting disk images containing all needed components
-to run cloud workloads, a.k.a. cloud images. 
+to run cloud workloads, a.k.a. cloud images.
 
 The following sample commands will download an Ubuntu Cloud image, converting
 it into a format that Cloud Hypervisor can use and a firmware to boot the image
@@ -148,7 +147,7 @@ $ sudo setcap cap_net_admin+ep ./cloud-hypervisor
 $ ./create-cloud-init.sh
 $ ./cloud-hypervisor \
 	--kernel ./hypervisor-fw \
-	--disk path=focal-server-cloudimg-amd64.raw path=/tmp/ubuntu-cloudinit.img \
+	--disk path=focal-server-cloudimg-amd64.raw --disk path=/tmp/ubuntu-cloudinit.img \
 	--cpus boot=4 \
 	--memory size=1024M \
 	--net "tap=,mac=,ip=,mask="
@@ -161,7 +160,7 @@ GRUB) is required then it necessary to switch to the serial console instead of
 ```shell
 $ ./cloud-hypervisor \
 	--kernel ./hypervisor-fw \
-	--disk path=focal-server-cloudimg-amd64.raw path=/tmp/ubuntu-cloudinit.img \
+	--disk path=focal-server-cloudimg-amd64.raw --disk path=/tmp/ubuntu-cloudinit.img \
 	--cpus boot=4 \
 	--memory size=1024M \
 	--net "tap=,mac=,ip=,mask=" \
@@ -173,23 +172,31 @@ $ ./cloud-hypervisor \
 
 #### Building your Kernel
 
-Cloud Hypervisor also supports direct kernel boot into a `vmlinux` ELF kernel (compiled with PVH support). In order to support development there is a custom branch; however provided the required options are enabled any recent kernel will suffice.
+Cloud Hypervisor also supports direct kernel boot. For x86-64, a `vmlinux` ELF kernel (compiled with PVH support) is needed. In order to support development there is a custom branch; however provided the required options are enabled any recent kernel will suffice.
 
 To build the kernel:
 
 ```shell
 # Clone the Cloud Hypervisor Linux branch
-$ git clone --depth 1 https://github.com/cloud-hypervisor/linux.git -b ch-5.15.12 linux-cloud-hypervisor
+$ git clone --depth 1 https://github.com/cloud-hypervisor/linux.git -b ch-6.2 linux-cloud-hypervisor
 $ pushd linux-cloud-hypervisor
-# Use the cloud-hypervisor kernel config to build your kernel
+# Use the x86-64 cloud-hypervisor kernel config to build your kernel for x86-64
 $ wget https://raw.githubusercontent.com/cloud-hypervisor/cloud-hypervisor/main/resources/linux-config-x86_64
-$ cp linux-config-x86_64 .config
+# Use the AArch64 cloud-hypervisor kernel config to build your kernel for AArch64
+$ wget https://raw.githubusercontent.com/cloud-hypervisor/cloud-hypervisor/main/resources/linux-config-aarch64
+$ cp linux-config-x86_64 .config  # x86-64
+$ cp linux-config-aarch64 .config # AArch64
+# Do native build of the x86-64 kernel
 $ KCFLAGS="-Wa,-mx86-used-note=no" make bzImage -j `nproc`
+# Do native build of the AArch64 kernel
+$ make -j `nproc`
 $ popd
 ```
 
-The `vmlinux` kernel image will then be located at
+For x86-64, the `vmlinux` kernel image will then be located at
 `linux-cloud-hypervisor/arch/x86/boot/compressed/vmlinux.bin`.
+For AArch64, the `Image` kernel image will then be located at
+`linux-cloud-hypervisor/arch/arm64/boot/Image`.
 
 #### Disk image
 
@@ -197,8 +204,10 @@ For the disk image the same Ubuntu image as before can be used. This contains
 an `ext4` root filesystem.
 
 ```shell
-$ wget https://cloud-images.ubuntu.com/focal/current/focal-server-cloudimg-amd64.img
-$ qemu-img convert -p -f qcow2 -O raw focal-server-cloudimg-amd64.img focal-server-cloudimg-amd64.raw
+$ wget https://cloud-images.ubuntu.com/focal/current/focal-server-cloudimg-amd64.img # x86-64
+$ wget https://cloud-images.ubuntu.com/focal/current/focal-server-cloudimg-arm64.img # AArch64
+$ qemu-img convert -p -f qcow2 -O raw focal-server-cloudimg-amd64.img focal-server-cloudimg-amd64.raw # x86-64
+$ qemu-img convert -p -f qcow2 -O raw focal-server-cloudimg-arm64.img focal-server-cloudimg-arm64.raw # AArch64
 ```
 
 #### Booting the guest VM
@@ -206,12 +215,28 @@ $ qemu-img convert -p -f qcow2 -O raw focal-server-cloudimg-amd64.img focal-serv
 These sample commands boot the disk image using the custom kernel whilst also
 supplying the desired kernel command line.
 
+- x86-64
+
 ```shell
 $ sudo setcap cap_net_admin+ep ./cloud-hypervisor
 $ ./create-cloud-init.sh
 $ ./cloud-hypervisor \
 	--kernel ./linux-cloud-hypervisor/arch/x86/boot/compressed/vmlinux.bin \
-	--disk path=focal-server-cloudimg-amd64.raw path=/tmp/ubuntu-cloudinit.img \
+	--disk path=focal-server-cloudimg-amd64.raw --disk path=/tmp/ubuntu-cloudinit.img \
+	--cmdline "console=hvc0 root=/dev/vda1 rw" \
+	--cpus boot=4 \
+	--memory size=1024M \
+	--net "tap=,mac=,ip=,mask="
+```
+
+- AArch64
+
+```shell
+$ sudo setcap cap_net_admin+ep ./cloud-hypervisor
+$ ./create-cloud-init.sh
+$ ./cloud-hypervisor \
+	--kernel ./linux-cloud-hypervisor/arch/arm64/boot/Image \
+	--disk path=focal-server-cloudimg-arm64.raw --disk path=/tmp/ubuntu-cloudinit.img \
 	--cmdline "console=hvc0 root=/dev/vda1 rw" \
 	--cpus boot=4 \
 	--memory size=1024M \
@@ -220,7 +245,10 @@ $ ./cloud-hypervisor \
 
 If earlier kernel messages are required the serial console should be used instead of `virtio-console`.
 
-```./cloud-hypervisor \
+- x86-64
+
+```shell
+$ ./cloud-hypervisor \
 	--kernel ./linux-cloud-hypervisor/arch/x86/boot/compressed/vmlinux.bin \
 	--console off \
 	--serial tty \
@@ -231,6 +259,20 @@ If earlier kernel messages are required the serial console should be used instea
 	--net "tap=,mac=,ip=,mask="
 ```
 
+- AArch64
+
+```shell
+$ ./cloud-hypervisor \
+	--kernel ./linux-cloud-hypervisor/arch/arm64/boot/Image \
+	--console off \
+	--serial tty \
+	--disk path=focal-server-cloudimg-arm64.raw \
+	--cmdline "console=ttyAMA0 root=/dev/vda1 rw" \
+	--cpus boot=4 \
+	--memory size=1024M \
+	--net "tap=,mac=,ip=,mask="
+```
+
 # 3. Status
 
 Cloud Hypervisor is under active development. The following stability
@@ -254,12 +296,11 @@ Currently the following items are **not** guaranteed across updates:
 
 Further details can be found in the [release documentation](docs/releases.md).
 
-As of 2022-10-13, the following cloud images are supported:
+As of 2023-01-03, the following cloud images are supported:
 
-- [Ubuntu Bionic](https://cloud-images.ubuntu.com/bionic/current/) (bionic-server-cloudimg-amd64.img)
-- [Ubuntu Focal](https://cloud-images.ubuntu.com/focal/current/) (focal-server-cloudimg-amd64.img)
-- [Ubuntu Jammy](https://cloud-images.ubuntu.com/jammy/current/) (jammy-server-cloudimg-amd64.img )
-- [Fedora 36](https://fedora.mirrorservice.org/fedora/linux/releases/36/Cloud/x86_64/images/) (Fedora-Cloud-Base-36-1.5.x86_64.raw.xz)
+- [Ubuntu Focal](https://cloud-images.ubuntu.com/focal/current/) (focal-server-cloudimg-{amd64,arm64}.img)
+- [Ubuntu Jammy](https://cloud-images.ubuntu.com/jammy/current/) (jammy-server-cloudimg-{amd64,arm64}.img )
+- [Fedora 36](https://fedora.mirrorservice.org/fedora/linux/releases/36/Cloud/) ([Fedora-Cloud-Base-36-1.5.x86_64.raw.xz](https://fedora.mirrorservice.org/fedora/linux/releases/36/Cloud/x86_64/images/) / [Fedora-Cloud-Base-36-1.5.aarch64.raw.xz](https://fedora.mirrorservice.org/fedora/linux/releases/36/Cloud/aarch64/images/))
 
 Direct kernel boot to userspace should work with a rootfs from most
 distributions although you may need to enable exotic filesystem types in the
@@ -331,8 +372,8 @@ are all equal and welcome means of contribution. See the
 
 ## Slack
 
-Get an [invite to our Slack channel](https://join.slack.com/t/cloud-hypervisor/shared_invite/enQtNjY3MTE3MDkwNDQ4LWQ1MTA1ZDVmODkwMWQ1MTRhYzk4ZGNlN2UwNTI3ZmFlODU0OTcwOWZjMTkwZDExYWE3YjFmNzgzY2FmNDAyMjI)
-and [join us on Slack](https://cloud-hypervisor.slack.com/).
+Get an [invite to our Slack channel](https://join.slack.com/t/cloud-hypervisor/shared_invite/enQtNjY3MTE3MDkwNDQ4LWQ1MTA1ZDVmODkwMWQ1MTRhYzk4ZGNlN2UwNTI3ZmFlODU0OTcwOWZjMTkwZDExYWE3YjFmNzgzY2FmNDAyMjI),
+ [join us on Slack](https://cloud-hypervisor.slack.com/), and [participate in our community activities](https://cloud-hypervisor.slack.com/archives/C04R5DUQVBN).
 
 ## Mailing list
 

acpi_tables/Cargo.toml

@@ -1,8 +0,0 @@
-[package]
-name = "acpi_tables"
-version = "0.1.0"
-authors = ["The Cloud Hypervisor Authors"]
-edition = "2021"
-
-[dependencies]
-vm-memory = "0.10.0"

acpi_tables/src/lib.rs

@@ -1,12 +0,0 @@
-// Copyright © 2019 Intel Corporation
-//
-// SPDX-License-Identifier: Apache-2.0
-//
-
-pub mod aml;
-pub mod rsdp;
-pub mod sdt;
-
-fn generate_checksum(data: &[u8]) -> u8 {
-    (255 - data.iter().fold(0u8, |acc, x| acc.wrapping_add(*x))).wrapping_add(1)
-}

acpi_tables/src/rsdp.rs

@@ -1,68 +0,0 @@
-// Copyright © 2019 Intel Corporation
-//
-// SPDX-License-Identifier: Apache-2.0
-//
-
-use vm_memory::ByteValued;
-
-#[repr(packed)]
-#[derive(Clone, Copy, Default)]
-pub struct Rsdp {
-    pub signature: [u8; 8],
-    pub checksum: u8,
-    pub oem_id: [u8; 6],
-    pub revision: u8,
-    _rsdt_addr: u32,
-    pub length: u32,
-    pub xsdt_addr: u64,
-    pub extended_checksum: u8,
-    _reserved: [u8; 3],
-}
-
-// SAFETY: Rsdp only contains a series of integers
-unsafe impl ByteValued for Rsdp {}
-
-impl Rsdp {
-    pub fn new(oem_id: [u8; 6], xsdt_addr: u64) -> Self {
-        let mut rsdp = Rsdp {
-            signature: *b"RSD PTR ",
-            checksum: 0,
-            oem_id,
-            revision: 2,
-            _rsdt_addr: 0,
-            length: std::mem::size_of::<Rsdp>() as u32,
-            xsdt_addr,
-            extended_checksum: 0,
-            _reserved: [0; 3],
-        };
-
-        rsdp.checksum = super::generate_checksum(&rsdp.as_slice()[0..19]);
-        rsdp.extended_checksum = super::generate_checksum(rsdp.as_slice());
-        rsdp
-    }
-
-    pub fn len() -> usize {
-        std::mem::size_of::<Rsdp>()
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::Rsdp;
-    use vm_memory::bytes::ByteValued;
-
-    #[test]
-    fn test_rsdp() {
-        let rsdp = Rsdp::new(*b"CHYPER", 0xdead_beef);
-        let sum = rsdp
-            .as_slice()
-            .iter()
-            .fold(0u8, |acc, x| acc.wrapping_add(*x));
-        assert_eq!(sum, 0);
-        let sum: u8 = rsdp
-            .as_slice()
-            .iter()
-            .fold(0u8, |acc, x| acc.wrapping_add(*x));
-        assert_eq!(sum, 0);
-    }
-}

acpi_tables/src/sdt.rs

@@ -1,147 +0,0 @@
-// Copyright © 2019 Intel Corporation
-//
-// SPDX-License-Identifier: Apache-2.0
-//
-
-#[repr(packed)]
-#[derive(Clone, Copy)]
-pub struct GenericAddress {
-    pub address_space_id: u8,
-    pub register_bit_width: u8,
-    pub register_bit_offset: u8,
-    pub access_size: u8,
-    pub address: u64,
-}
-
-impl GenericAddress {
-    pub fn io_port_address<T>(address: u16) -> Self {
-        GenericAddress {
-            address_space_id: 1,
-            register_bit_width: 8 * std::mem::size_of::<T>() as u8,
-            register_bit_offset: 0,
-            access_size: std::mem::size_of::<T>() as u8,
-            address: u64::from(address),
-        }
-    }
-    pub fn mmio_address<T>(address: u64) -> Self {
-        GenericAddress {
-            address_space_id: 0,
-            register_bit_width: 8 * std::mem::size_of::<T>() as u8,
-            register_bit_offset: 0,
-            access_size: std::mem::size_of::<T>() as u8,
-            address,
-        }
-    }
-}
-
-pub struct Sdt {
-    data: Vec<u8>,
-}
-
-#[allow(clippy::len_without_is_empty)]
-impl Sdt {
-    pub fn new(
-        signature: [u8; 4],
-        length: u32,
-        revision: u8,
-        oem_id: [u8; 6],
-        oem_table: [u8; 8],
-        oem_revision: u32,
-    ) -> Self {
-        assert!(length >= 36);
-
-        let mut data = Vec::with_capacity(length as usize);
-        data.extend_from_slice(&signature);
-        data.extend_from_slice(&length.to_le_bytes());
-        data.push(revision);
-        data.push(0); // checksum
-        data.extend_from_slice(&oem_id);
-        data.extend_from_slice(&oem_table);
-        data.extend_from_slice(&oem_revision.to_le_bytes());
-        data.extend_from_slice(b"CLDH");
-        data.extend_from_slice(&0u32.to_le_bytes());
-        assert_eq!(data.len(), 36);
-
-        data.resize(length as usize, 0);
-        let mut sdt = Sdt { data };
-
-        sdt.update_checksum();
-        sdt
-    }
-
-    pub fn update_checksum(&mut self) {
-        self.data[9] = 0;
-        let checksum = super::generate_checksum(self.data.as_slice());
-        self.data[9] = checksum
-    }
-
-    pub fn as_slice(&self) -> &[u8] {
-        self.data.as_slice()
-    }
-
-    pub fn append<T>(&mut self, value: T) {
-        let orig_length = self.data.len();
-        let new_length = orig_length + std::mem::size_of::<T>();
-        self.data.resize(new_length, 0);
-        self.write_u32(4, new_length as u32);
-        self.write(orig_length, value);
-    }
-
-    pub fn append_slice(&mut self, data: &[u8]) {
-        let orig_length = self.data.len();
-        let new_length = orig_length + data.len();
-        self.write_u32(4, new_length as u32);
-        self.data.extend_from_slice(data);
-        self.update_checksum();
-    }
-
-    /// Write a value at the given offset
-    pub fn write<T>(&mut self, offset: usize, value: T) {
-        assert!((offset + (std::mem::size_of::<T>() - 1)) < self.data.len());
-        unsafe {
-            *(((self.data.as_mut_ptr() as usize) + offset) as *mut T) = value;
-        }
-        self.update_checksum();
-    }
-
-    pub fn write_u8(&mut self, offset: usize, val: u8) {
-        self.write(offset, val);
-    }
-
-    pub fn write_u16(&mut self, offset: usize, val: u16) {
-        self.write(offset, val);
-    }
-
-    pub fn write_u32(&mut self, offset: usize, val: u32) {
-        self.write(offset, val);
-    }
-
-    pub fn write_u64(&mut self, offset: usize, val: u64) {
-        self.write(offset, val);
-    }
-
-    pub fn len(&self) -> usize {
-        self.data.len()
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::Sdt;
-
-    #[test]
-    fn test_sdt() {
-        let mut sdt = Sdt::new(*b"TEST", 40, 1, *b"CLOUDH", *b"TESTTEST", 1);
-        let sum: u8 = sdt
-            .as_slice()
-            .iter()
-            .fold(0u8, |acc, x| acc.wrapping_add(*x));
-        assert_eq!(sum, 0);
-        sdt.write_u32(36, 0x12345678);
-        let sum: u8 = sdt
-            .as_slice()
-            .iter()
-            .fold(0u8, |acc, x| acc.wrapping_add(*x));
-        assert_eq!(sum, 0);
-    }
-}

arch/Cargo.toml

@@ -9,16 +9,16 @@ default = []
 tdx = []
 
 [dependencies]
-anyhow = "1.0.66"
+anyhow = "1.0.70"
 byteorder = "1.4.3"
 hypervisor = { path = "../hypervisor" }
-libc = "0.2.138"
+libc = "0.2.139"
 linux-loader = { version = "0.8.1", features = ["elf", "bzimage", "pe"] }
 log = "0.4.17"
-serde = { version = "1.0.150", features = ["rc", "derive"] }
-thiserror = "1.0.37"
-uuid = "1.2.2"
-versionize = "0.1.9"
+serde = { version = "1.0.156", features = ["rc", "derive"] }
+thiserror = "1.0.39"
+uuid = "1.3.0"
+versionize = "0.1.10"
 versionize_derive = "0.1.4"
 vm-memory = { version = "0.10.0", features = ["backend-mmap", "backend-bitmap"] }
 vm-migration = { path = "../vm-migration" }

arch/src/aarch64/fdt.rs

@@ -191,9 +191,13 @@ fn create_cpu_nodes(
         let cpu_map_node = fdt.begin_node("cpu-map")?;
 
         // Create device tree nodes with regard of above mapping.
-        for cluster_idx in 0..packages {
-            let cluster_name = format!("cluster{cluster_idx:x}");
-            let cluster_node = fdt.begin_node(&cluster_name)?;
+        for package_idx in 0..packages {
+            let package_name = format!("socket{package_idx:x}");
+            let package_node = fdt.begin_node(&package_name)?;
+
+            // Cluster is the container of cores, and it is mandatory in the CPU topology.
+            // Add a default "cluster0" in each socket/package.
+            let cluster_node = fdt.begin_node("cluster0")?;
 
             for core_idx in 0..cores_per_package {
                 let core_name = format!("core{core_idx:x}");
@@ -202,7 +206,7 @@ fn create_cpu_nodes(
                 for thread_idx in 0..threads_per_core {
                     let thread_name = format!("thread{thread_idx:x}");
                     let thread_node = fdt.begin_node(&thread_name)?;
-                    let cpu_idx = threads_per_core * cores_per_package * cluster_idx
+                    let cpu_idx = threads_per_core * cores_per_package * package_idx
                         + threads_per_core * core_idx
                         + thread_idx;
                     fdt.property_u32("cpu", cpu_idx as u32 + FIRST_VCPU_PHANDLE)?;
@@ -212,6 +216,7 @@ fn create_cpu_nodes(
                 fdt.end_node(core_node)?;
             }
             fdt.end_node(cluster_node)?;
+            fdt.end_node(package_node)?;
         }
         fdt.end_node(cpu_map_node)?;
     } else {
@@ -255,35 +260,86 @@ fn create_memory_node(
             fdt.end_node(memory_node)?;
         }
     } else {
-        let last_addr = guest_mem.last_addr().raw_value();
-        if last_addr < super::layout::MEM_32BIT_RESERVED_START.raw_value() {
-            // Case 1: all RAM is under the hole
-            let mem_size = last_addr - super::layout::RAM_START.raw_value() + 1;
-            let mem_reg_prop = [super::layout::RAM_START.raw_value() as u64, mem_size as u64];
-            let memory_node = fdt.begin_node("memory")?;
-            fdt.property_string("device_type", "memory")?;
-            fdt.property_array_u64("reg", &mem_reg_prop)?;
-            fdt.end_node(memory_node)?;
-        } else {
-            // Case 2: RAM is split by the hole
-            // Region 1: RAM before the hole
-            let mem_size = super::layout::MEM_32BIT_RESERVED_START.raw_value()
-                - super::layout::RAM_START.raw_value();
-            let mem_reg_prop = [super::layout::RAM_START.raw_value() as u64, mem_size as u64];
-            let memory_node_name = format!("memory@{:x}", super::layout::RAM_START.raw_value());
+        // Note: memory regions from "GuestMemory" are sorted and non-zero sized.
+        let ram_regions = {
+            let mut ram_regions = Vec::new();
+            let mut current_start = guest_mem
+                .iter()
+                .next()
+                .map(GuestMemoryRegion::start_addr)
+                .expect("GuestMemory must have one memory region at least")
+                .raw_value();
+            let mut current_end = current_start;
+
+            for (start, size) in guest_mem
+                .iter()
+                .map(|m| (m.start_addr().raw_value(), m.len()))
+            {
+                if current_end == start {
+                    // This zone is continuous with the previous one.
+                    current_end += size;
+                } else {
+                    ram_regions.push((current_start, current_end));
+
+                    current_start = start;
+                    current_end = start + size;
+                }
+            }
+
+            ram_regions.push((current_start, current_end));
+
+            ram_regions
+        };
+
+        if ram_regions.len() > 2 {
+            panic!(
+                "There should be up to two non-continuous regions, devidided by the
+                    gap at the end of 32bit address space."
+            );
+        }
+
+        // Create the memory node for memory region before the gap
+        {
+            let (first_region_start, first_region_end) = ram_regions
+                .first()
+                .expect("There should be at last one memory region");
+            let ram_start = super::layout::RAM_START.raw_value();
+            let mem_32bit_reserved_start = super::layout::MEM_32BIT_RESERVED_START.raw_value();
+
+            if !((first_region_start <= &ram_start)
+                && (first_region_end > &ram_start)
+                && (first_region_end <= &mem_32bit_reserved_start))
+            {
+                panic!(
+                    "Unexpected first memory region layout: (start: 0x{:08x}, end: 0x{:08x}).
+                    ram_start: 0x{:08x}, mem_32bit_reserved_start: 0x{:08x}",
+                    first_region_start, first_region_end, ram_start, mem_32bit_reserved_start
+                );
+            }
+
+            let mem_size = first_region_end - ram_start;
+            let mem_reg_prop = [ram_start, mem_size];
+            let memory_node_name = format!("memory@{:x}", ram_start);
             let memory_node = fdt.begin_node(&memory_node_name)?;
             fdt.property_string("device_type", "memory")?;
             fdt.property_array_u64("reg", &mem_reg_prop)?;
             fdt.end_node(memory_node)?;
+        }
 
-            // Region 2: RAM after the hole
-            let mem_size = last_addr - super::layout::RAM_64BIT_START.raw_value() + 1;
-            let mem_reg_prop = [
-                super::layout::RAM_64BIT_START.raw_value() as u64,
-                mem_size as u64,
-            ];
-            let memory_node_name =
-                format!("memory@{:x}", super::layout::RAM_64BIT_START.raw_value());
+        // Create the memory map entry for memory region after the gap if any
+        if let Some((second_region_start, second_region_end)) = ram_regions.get(1) {
+            let ram_64bit_start = super::layout::RAM_64BIT_START.raw_value();
+
+            if second_region_start != &ram_64bit_start {
+                panic!(
+                    "Unexpected second memory region layout: start: 0x{:08x}, ram_64bit_start: 0x{:08x}",
+                    second_region_start, ram_64bit_start
+                );
+            }
+
+            let mem_size = second_region_end - ram_64bit_start;
+            let mem_reg_prop = [ram_64bit_start, mem_size];
+            let memory_node_name = format!("memory@{:x}", ram_64bit_start);
             let memory_node = fdt.begin_node(&memory_node_name)?;
             fdt.property_string("device_type", "memory")?;
             fdt.property_array_u64("reg", &mem_reg_prop)?;
@@ -303,7 +359,7 @@ fn create_chosen_node(
     fdt.property_string("bootargs", cmdline)?;
 
     if let Some(initrd_config) = initrd {
-        let initrd_start = initrd_config.address.raw_value() as u64;
+        let initrd_start = initrd_config.address.raw_value();
         let initrd_end = initrd_config.address.raw_value() + initrd_config.size as u64;
         fdt.property_u64("linux,initrd-start", initrd_start)?;
         fdt.property_u64("linux,initrd-end", initrd_end)?;

arch/src/aarch64/mod.rs

@@ -19,7 +19,9 @@ use std::collections::HashMap;
 use std::convert::TryInto;
 use std::fmt::Debug;
 use std::sync::{Arc, Mutex};
-use vm_memory::{Address, GuestAddress, GuestMemory, GuestUsize};
+use vm_memory::{Address, GuestAddress, GuestMemory, GuestMemoryAtomic};
+
+pub const _NSIG: i32 = 65;
 
 /// Errors thrown while configuring aarch64 system.
 #[derive(Debug)]
@@ -64,9 +66,9 @@ pub struct EntryPoint {
 pub fn configure_vcpu(
     vcpu: &Arc<dyn hypervisor::Vcpu>,
     id: u8,
-    kernel_entry_point: Option<EntryPoint>,
+    boot_setup: Option<(EntryPoint, &GuestMemoryAtomic<GuestMemoryMmap>)>,
 ) -> super::Result<u64> {
-    if let Some(kernel_entry_point) = kernel_entry_point {
+    if let Some((kernel_entry_point, _guest_memory)) = boot_setup {
         vcpu.setup_regs(
             id,
             kernel_entry_point.entry_addr.raw_value(),
@@ -81,8 +83,8 @@ pub fn configure_vcpu(
     Ok(mpidr)
 }
 
-pub fn arch_memory_regions(size: GuestUsize) -> Vec<(GuestAddress, usize, RegionType)> {
-    let mut regions = vec![
+pub fn arch_memory_regions() -> Vec<(GuestAddress, usize, RegionType)> {
+    vec![
         // 0 MiB ~ 256 MiB: UEFI, GIC and legacy devices
         (
             GuestAddress(0),
@@ -101,39 +103,21 @@ pub fn arch_memory_regions(size: GuestUsize) -> Vec<(GuestAddress, usize, Region
             layout::PCI_MMCONFIG_SIZE as usize,
             RegionType::Reserved,
         ),
-    ];
-
-    let ram_32bit_space_size =
-        layout::MEM_32BIT_RESERVED_START.unchecked_offset_from(layout::RAM_START);
-
-    // RAM space
-    // Case1: guest memory fits before the gap
-    if size as u64 <= ram_32bit_space_size {
-        regions.push((layout::RAM_START, size as usize, RegionType::Ram));
-    // Case2: guest memory extends beyond the gap
-    } else {
-        // Push memory before the gap
-        regions.push((
+        // 1GiB ~ 4032 MiB: RAM before the gap
+        (
             layout::RAM_START,
-            ram_32bit_space_size as usize,
+            layout::MEM_32BIT_RESERVED_START.unchecked_offset_from(layout::RAM_START) as usize,
             RegionType::Ram,
-        ));
-        // Other memory is placed after 4GiB
-        regions.push((
-            layout::RAM_64BIT_START,
-            (size - ram_32bit_space_size) as usize,
-            RegionType::Ram,
-        ));
-    }
-
-    // Add the 32-bit reserved memory hole as a reserved region
-    regions.push((
-        layout::MEM_32BIT_RESERVED_START,
-        layout::MEM_32BIT_RESERVED_SIZE as usize,
-        RegionType::Reserved,
-    ));
-
-    regions
+        ),
+        // 4GiB ~ inf: RAM after the gap
+        (layout::RAM_64BIT_START, usize::MAX, RegionType::Ram),
+        // Add the 32-bit reserved memory hole as a reserved region
+        (
+            layout::MEM_32BIT_RESERVED_START,
+            layout::MEM_32BIT_RESERVED_SIZE as usize,
+            RegionType::Reserved,
+        ),
+    ]
 }
 
 /// Configures the system and should be called once per vm before starting vcpu threads.
@@ -215,26 +199,12 @@ mod tests {
     use super::*;
 
     #[test]
-    fn test_arch_memory_regions_dram_2gb() {
-        let regions = arch_memory_regions((1usize << 31) as u64); //2GB
-        assert_eq!(5, regions.len());
-        assert_eq!(layout::RAM_START, regions[3].0);
-        assert_eq!((1usize << 31), regions[3].1);
-        assert_eq!(RegionType::Ram, regions[3].2);
-        assert_eq!(RegionType::Reserved, regions[4].2);
-    }
-
-    #[test]
-    fn test_arch_memory_regions_dram_4gb() {
-        let regions = arch_memory_regions((1usize << 32) as u64); //4GB
-        let ram_32bit_space_size =
-            layout::MEM_32BIT_RESERVED_START.unchecked_offset_from(layout::RAM_START) as usize;
+    fn test_arch_memory_regions_dram() {
+        let regions = arch_memory_regions();
         assert_eq!(6, regions.len());
         assert_eq!(layout::RAM_START, regions[3].0);
-        assert_eq!(ram_32bit_space_size as usize, regions[3].1);
         assert_eq!(RegionType::Ram, regions[3].2);
         assert_eq!(RegionType::Reserved, regions[5].2);
         assert_eq!(RegionType::Ram, regions[4].2);
-        assert_eq!(((1usize << 32) - ram_32bit_space_size), regions[4].1);
     }
 }

arch/src/aarch64/regs.rs

@@ -1,18 +1,17 @@
 // Copyright 2022 Arm Limited (or its affiliates). All rights reserved.
 // SPDX-License-Identifier: Apache-2.0
 
-///
-/// AArch64 system register encoding:
-/// See https://developer.arm.com/documentation/ddi0487 (chapter D12)
-///
-///   31      22  21 20 19 18 16 15 12 11  8 7   5 4  0
-///  +----------+---+-----+-----+-----+-----+-----+----+
-///  |1101010100| L | op0 | op1 | CRn | CRm | op2 | Rt |
-///  +----------+---+-----+-----+-----+-----+-----+----+
-///
-/// Notes:
-/// - L and Rt are reserved as implementation defined fields, ignored.
-///
+// AArch64 system register encoding:
+// See https://developer.arm.com/documentation/ddi0487 (chapter D12)
+//
+//   31      22  21 20 19 18 16 15 12 11  8 7   5 4  0
+//  +----------+---+-----+-----+-----+-----+-----+----+
+//  |1101010100| L | op0 | op1 | CRn | CRm | op2 | Rt |
+//  +----------+---+-----+-----+-----+-----+-----+----+
+//
+// Notes:
+// - L and Rt are reserved as implementation defined fields, ignored.
+
 const SYSREG_HEAD: u32 = 0b1101010100u32 << 22;
 const SYSREG_OP0_SHIFT: u32 = 19;
 const SYSREG_OP0_MASK: u32 = 0b11u32 << 19;

arch/src/lib.rs

@@ -82,7 +82,7 @@ pub mod aarch64;
 pub use aarch64::{
     arch_memory_regions, configure_system, configure_vcpu, fdt::DeviceInfoForFdt,
     get_host_cpu_phys_bits, initramfs_load_addr, layout, layout::CMDLINE_MAX_SIZE,
-    layout::IRQ_BASE, uefi, EntryPoint,
+    layout::IRQ_BASE, uefi, EntryPoint, _NSIG,
 };
 
 #[cfg(target_arch = "x86_64")]
@@ -92,14 +92,14 @@ pub mod x86_64;
 pub use x86_64::{
     arch_memory_regions, configure_system, configure_vcpu, generate_common_cpuid,
     get_host_cpu_phys_bits, initramfs_load_addr, layout, layout::CMDLINE_MAX_SIZE,
-    layout::CMDLINE_START, regs, CpuidFeatureEntry, EntryPoint,
+    layout::CMDLINE_START, regs, CpuidFeatureEntry, EntryPoint, _NSIG,
 };
 
 /// Safe wrapper for `sysconf(_SC_PAGESIZE)`.
 #[cfg(target_arch = "x86_64")]
 #[inline(always)]
 fn pagesize() -> usize {
-    // Trivially safe
+    // SAFETY: Trivially safe
     unsafe { libc::sysconf(libc::_SC_PAGESIZE) as usize }
 }
 

arch/src/x86_64/mod.rs

@@ -53,6 +53,8 @@ const KVM_FEATURE_ASYNC_PF_VMEXIT_BIT: u8 = 10;
 #[cfg(feature = "tdx")]
 const KVM_FEATURE_STEAL_TIME_BIT: u8 = 5;
 
+pub const _NSIG: i32 = 65;
+
 #[derive(Debug, Copy, Clone)]
 /// Specifies the entry point address where the guest must start
 /// executing code, as well as which of the supported boot protocols
@@ -126,9 +128,11 @@ struct MemmapTableEntryWrapper(hvm_memmap_table_entry);
 #[derive(Copy, Clone, Default)]
 struct ModlistEntryWrapper(hvm_modlist_entry);
 
-// SAFETY: These data structures only contain a series of integers
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for StartInfoWrapper {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for MemmapTableEntryWrapper {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for ModlistEntryWrapper {}
 
 // This is a workaround to the Rust enforcement specifying that any implementation of a foreign
@@ -206,7 +210,6 @@ impl From<Error> for super::Error {
     }
 }
 
-#[allow(clippy::upper_case_acronyms)]
 #[derive(Copy, Clone, Debug)]
 pub enum CpuidReg {
     EAX,
@@ -549,13 +552,30 @@ impl CpuidFeatureEntry {
 }
 
 pub fn generate_common_cpuid(
-    hypervisor: Arc<dyn hypervisor::Hypervisor>,
+    hypervisor: &Arc<dyn hypervisor::Hypervisor>,
     topology: Option<(u8, u8, u8)>,
     sgx_epc_sections: Option<Vec<SgxEpcSection>>,
     phys_bits: u8,
     kvm_hyperv: bool,
     #[cfg(feature = "tdx")] tdx_enabled: bool,
 ) -> super::Result<Vec<CpuIdEntry>> {
+    // SAFETY: cpuid called with valid leaves
+    if unsafe { x86_64::__cpuid(1) }.ecx & 1 << HYPERVISOR_ECX_BIT == 1 << HYPERVISOR_ECX_BIT {
+        // SAFETY: cpuid called with valid leaves
+        let hypervisor_cpuid = unsafe { x86_64::__cpuid(0x4000_0000) };
+
+        let mut identifier: [u8; 12] = [0; 12];
+        identifier[0..4].copy_from_slice(&hypervisor_cpuid.ebx.to_le_bytes()[..]);
+        identifier[4..8].copy_from_slice(&hypervisor_cpuid.ecx.to_le_bytes()[..]);
+        identifier[8..12].copy_from_slice(&hypervisor_cpuid.edx.to_le_bytes()[..]);
+
+        info!(
+            "Running under nested virtualisation. Hypervisor string: {}",
+            String::from_utf8_lossy(&identifier)
+        );
+    }
+
+    info!("Generating guest CPUID for with physical address size: {phys_bits}");
     let cpuid_patches = vec![
         // Patch tsc deadline timer bit
         CpuidPatch {
@@ -590,7 +610,9 @@ pub fn generate_common_cpuid(
     ];
 
     // Supported CPUID
-    let mut cpuid = hypervisor.get_cpuid().map_err(Error::CpuidGetSupported)?;
+    let mut cpuid = hypervisor
+        .get_supported_cpuid()
+        .map_err(Error::CpuidGetSupported)?;
 
     CpuidPatch::patch_cpuid(&mut cpuid, cpuid_patches);
 
@@ -679,6 +701,7 @@ pub fn generate_common_cpuid(
     // Copy CPU identification string
     for i in 0x8000_0002..=0x8000_0004 {
         cpuid.retain(|c| c.function != i);
+        // SAFETY: call cpuid with valid leaves
         let leaf = unsafe { std::arch::x86_64::__cpuid(i) };
         cpuid.push(CpuIdEntry {
             function: i,
@@ -743,8 +766,7 @@ pub fn generate_common_cpuid(
 pub fn configure_vcpu(
     vcpu: &Arc<dyn hypervisor::Vcpu>,
     id: u8,
-    kernel_entry_point: Option<EntryPoint>,
-    vm_memory: &GuestMemoryAtomic<GuestMemoryMmap>,
+    boot_setup: Option<(EntryPoint, &GuestMemoryAtomic<GuestMemoryMmap>)>,
     cpuid: Vec<CpuIdEntry>,
     kvm_hyperv: bool,
 ) -> super::Result<()> {
@@ -796,12 +818,12 @@ pub fn configure_vcpu(
     }
 
     regs::setup_msrs(vcpu).map_err(Error::MsrsConfiguration)?;
-    if let Some(kernel_entry_point) = kernel_entry_point {
+    if let Some((kernel_entry_point, guest_memory)) = boot_setup {
         if let Some(entry_addr) = kernel_entry_point.entry_addr {
             // Safe to unwrap because this method is called after the VM is configured
             regs::setup_regs(vcpu, entry_addr.raw_value()).map_err(Error::RegsConfiguration)?;
             regs::setup_fpu(vcpu).map_err(Error::FpuConfiguration)?;
-            regs::setup_sregs(&vm_memory.memory(), vcpu).map_err(Error::SregsConfiguration)?;
+            regs::setup_sregs(&guest_memory.memory(), vcpu).map_err(Error::SregsConfiguration)?;
         }
     }
     interrupts::set_lint(vcpu).map_err(|e| Error::LocalIntConfiguration(e.into()))?;
@@ -812,47 +834,29 @@ pub fn configure_vcpu(
 /// These should be used to configure the GuestMemory structure for the platform.
 /// For x86_64 all addresses are valid from the start of the kernel except a
 /// carve out at the end of 32bit address space.
-pub fn arch_memory_regions(size: GuestUsize) -> Vec<(GuestAddress, usize, RegionType)> {
-    let reserved_memory_gap_start = layout::MEM_32BIT_RESERVED_START
-        .checked_add(layout::MEM_32BIT_DEVICES_SIZE)
-        .expect("32-bit reserved region is too large");
-
-    let requested_memory_size = GuestAddress(size);
-    let mut regions = Vec::new();
-
-    // case1: guest memory fits before the gap
-    if size <= layout::MEM_32BIT_RESERVED_START.raw_value() {
-        regions.push((GuestAddress(0), size as usize, RegionType::Ram));
-    // case2: guest memory extends beyond the gap
-    } else {
-        // push memory before the gap
-        regions.push((
+pub fn arch_memory_regions() -> Vec<(GuestAddress, usize, RegionType)> {
+    vec![
+        // 0 GiB ~ 3GiB: memory before the gap
+        (
             GuestAddress(0),
             layout::MEM_32BIT_RESERVED_START.raw_value() as usize,
             RegionType::Ram,
-        ));
-        regions.push((
-            layout::RAM_64BIT_START,
-            requested_memory_size.unchecked_offset_from(layout::MEM_32BIT_RESERVED_START) as usize,
-            RegionType::Ram,
-        ));
-    }
-
-    // Add the 32-bit device memory hole as a sub region.
-    regions.push((
-        layout::MEM_32BIT_RESERVED_START,
-        layout::MEM_32BIT_DEVICES_SIZE as usize,
-        RegionType::SubRegion,
-    ));
-
-    // Add the 32-bit reserved memory hole as a sub region.
-    regions.push((
-        reserved_memory_gap_start,
-        (layout::MEM_32BIT_RESERVED_SIZE - layout::MEM_32BIT_DEVICES_SIZE) as usize,
-        RegionType::Reserved,
-    ));
-
-    regions
+        ),
+        // 4 GiB ~ inf: memory after the gap
+        (layout::RAM_64BIT_START, usize::MAX, RegionType::Ram),
+        // 3 GiB ~ 3712 MiB: 32-bit device memory hole
+        (
+            layout::MEM_32BIT_RESERVED_START,
+            layout::MEM_32BIT_DEVICES_SIZE as usize,
+            RegionType::SubRegion,
+        ),
+        // 3712 MiB ~ 3968 MiB: 32-bit reserved memory hole
+        (
+            layout::MEM_32BIT_RESERVED_START.unchecked_add(layout::MEM_32BIT_DEVICES_SIZE),
+            (layout::MEM_32BIT_RESERVED_SIZE - layout::MEM_32BIT_DEVICES_SIZE) as usize,
+            RegionType::Reserved,
+        ),
+    ]
 }
 
 /// Configures the system and should be called once per vm before starting vcpu threads.
@@ -950,30 +954,102 @@ fn configure_pvh(
     // Create the memory map entries.
     add_memmap_entry(&mut memmap, 0, layout::EBDA_START.raw_value(), E820_RAM);
 
-    let mem_end = guest_mem.last_addr();
+    // Merge continuous memory regions into one region.
+    // Note: memory regions from "GuestMemory" are sorted and non-zero sized.
+    let ram_regions = {
+        let mut ram_regions = Vec::new();
+        let mut current_start = guest_mem
+            .iter()
+            .next()
+            .map(GuestMemoryRegion::start_addr)
+            .expect("GuestMemory must have one memory region at least")
+            .raw_value();
+        let mut current_end = current_start;
 
-    if mem_end < layout::MEM_32BIT_RESERVED_START {
-        add_memmap_entry(
-            &mut memmap,
-            layout::HIGH_RAM_START.raw_value(),
-            mem_end.unchecked_offset_from(layout::HIGH_RAM_START) + 1,
-            E820_RAM,
-        );
-    } else {
-        add_memmap_entry(
-            &mut memmap,
-            layout::HIGH_RAM_START.raw_value(),
-            layout::MEM_32BIT_RESERVED_START.unchecked_offset_from(layout::HIGH_RAM_START),
-            E820_RAM,
-        );
-        if mem_end > layout::RAM_64BIT_START {
-            add_memmap_entry(
-                &mut memmap,
-                layout::RAM_64BIT_START.raw_value(),
-                mem_end.unchecked_offset_from(layout::RAM_64BIT_START) + 1,
-                E820_RAM,
-            );
+        for (start, size) in guest_mem
+            .iter()
+            .map(|m| (m.start_addr().raw_value(), m.len()))
+        {
+            if current_end == start {
+                // This zone is continuous with the previous one.
+                current_end += size;
+            } else {
+                ram_regions.push((current_start, current_end));
+
+                current_start = start;
+                current_end = start + size;
+            }
         }
+
+        ram_regions.push((current_start, current_end));
+
+        ram_regions
+    };
+
+    if ram_regions.len() > 2 {
+        error!(
+            "There should be up to two non-continuous regions, devidided by the
+            gap at the end of 32bit address space (e.g. between 3G and 4G)."
+        );
+        return Err(super::Error::MemmapTableSetup);
+    }
+
+    // Create the memory map entry for memory region before the gap
+    {
+        let (first_region_start, first_region_end) =
+            ram_regions.first().ok_or(super::Error::MemmapTableSetup)?;
+        let high_ram_start = layout::HIGH_RAM_START.raw_value();
+        let mem_32bit_reserved_start = layout::MEM_32BIT_RESERVED_START.raw_value();
+
+        if !((first_region_start <= &high_ram_start)
+            && (first_region_end > &high_ram_start)
+            && (first_region_end <= &mem_32bit_reserved_start))
+        {
+            error!(
+                "Unexpected first memory region layout: (start: 0x{:08x}, end: 0x{:08x}).
+                high_ram_start: 0x{:08x}, mem_32bit_reserved_start: 0x{:08x}",
+                first_region_start, first_region_end, high_ram_start, mem_32bit_reserved_start
+            );
+
+            return Err(super::Error::MemmapTableSetup);
+        }
+
+        info!(
+            "create_memmap_entry, start: 0x{:08x}, end: 0x{:08x})",
+            high_ram_start, first_region_end
+        );
+
+        add_memmap_entry(
+            &mut memmap,
+            high_ram_start,
+            first_region_end - high_ram_start,
+            E820_RAM,
+        );
+    }
+
+    // Create the memory map entry for memory region after the gap if any
+    if let Some((second_region_start, second_region_end)) = ram_regions.get(1) {
+        let ram_64bit_start = layout::RAM_64BIT_START.raw_value();
+
+        if second_region_start != &ram_64bit_start {
+            error!(
+                "Unexpected second memory region layout: start: 0x{:08x}, ram_64bit_start: 0x{:08x}",
+                second_region_start, ram_64bit_start
+            );
+
+            return Err(super::Error::MemmapTableSetup);
+        }
+
+        info!(
+            "create_memmap_entry, start: 0x{:08x}, end: 0x{:08x})",
+            ram_64bit_start, second_region_end
+        );
+        add_memmap_entry(
+            &mut memmap,
+            ram_64bit_start,
+            second_region_end - ram_64bit_start,
+            E820_RAM,
+        );
     }
 
     add_memmap_entry(
@@ -1064,6 +1140,7 @@ pub fn initramfs_load_addr(
 }
 
 pub fn get_host_cpu_phys_bits() -> u8 {
+    // SAFETY: call cpuid with valid leaves
     unsafe {
         let leaf = x86_64::__cpuid(0x8000_0000);
 
@@ -1174,6 +1251,7 @@ fn update_cpuid_sgx(
 
     // Get host CPUID for leaf 0x12, subleaf 0x2. This is to retrieve EPC
     // properties such as confidentiality and integrity.
+    // SAFETY: call cpuid with valid leaves
     let leaf = unsafe { std::arch::x86_64::__cpuid_count(0x12, 0x2) };
 
     for (i, epc_section) in epc_sections.iter().enumerate() {
@@ -1207,16 +1285,8 @@ mod tests {
     use super::*;
 
     #[test]
-    fn regions_lt_4gb() {
-        let regions = arch_memory_regions(1 << 29);
-        assert_eq!(3, regions.len());
-        assert_eq!(GuestAddress(0), regions[0].0);
-        assert_eq!(1usize << 29, regions[0].1);
-    }
-
-    #[test]
-    fn regions_gt_4gb() {
-        let regions = arch_memory_regions((1 << 32) + 0x8000);
+    fn regions_base_addr() {
+        let regions = arch_memory_regions();
         assert_eq!(4, regions.len());
         assert_eq!(GuestAddress(0), regions[0].0);
         assert_eq!(GuestAddress(1 << 32), regions[1].0);
@@ -1240,11 +1310,10 @@ mod tests {
         assert!(config_err.is_err());
 
         // Now assigning some memory that falls before the 32bit memory hole.
-        let mem_size = 128 << 20;
-        let arch_mem_regions = arch_memory_regions(mem_size);
+        let arch_mem_regions = arch_memory_regions();
         let ram_regions: Vec<(GuestAddress, usize)> = arch_mem_regions
             .iter()
-            .filter(|r| r.2 == RegionType::Ram)
+            .filter(|r| r.2 == RegionType::Ram && r.1 != usize::MAX)
             .map(|r| (r.0, r.1))
             .collect();
         let gm = GuestMemoryMmap::from_ranges(&ram_regions).unwrap();
@@ -1262,48 +1331,18 @@ mod tests {
         )
         .unwrap();
 
-        // Now assigning some memory that is equal to the start of the 32bit memory hole.
-        let mem_size = 3328 << 20;
-        let arch_mem_regions = arch_memory_regions(mem_size);
-        let ram_regions: Vec<(GuestAddress, usize)> = arch_mem_regions
-            .iter()
-            .filter(|r| r.2 == RegionType::Ram)
-            .map(|r| (r.0, r.1))
-            .collect();
-        let gm = GuestMemoryMmap::from_ranges(&ram_regions).unwrap();
-        configure_system(
-            &gm,
-            GuestAddress(0),
-            &None,
-            no_vcpus,
-            None,
-            None,
-            None,
-            None,
-            None,
-        )
-        .unwrap();
-
-        configure_system(
-            &gm,
-            GuestAddress(0),
-            &None,
-            no_vcpus,
-            None,
-            None,
-            None,
-            None,
-            None,
-        )
-        .unwrap();
-
         // Now assigning some memory that falls after the 32bit memory hole.
-        let mem_size = 3330 << 20;
-        let arch_mem_regions = arch_memory_regions(mem_size);
+        let arch_mem_regions = arch_memory_regions();
         let ram_regions: Vec<(GuestAddress, usize)> = arch_mem_regions
             .iter()
             .filter(|r| r.2 == RegionType::Ram)
-            .map(|r| (r.0, r.1))
+            .map(|r| {
+                if r.1 == usize::MAX {
+                    (r.0, 128 << 20)
+                } else {
+                    (r.0, r.1)
+                }
+            })
             .collect();
         let gm = GuestMemoryMmap::from_ranges(&ram_regions).unwrap();
         configure_system(

arch/src/x86_64/mptable.rs

@@ -37,11 +37,17 @@ struct MpfIntelWrapper(mpspec::mpf_intel);
 
 // SAFETY: These `mpspec` wrapper types are only data, reading them from data is a safe initialization.
 unsafe impl ByteValued for MpcBusWrapper {}
+// SAFETY: see above
 unsafe impl ByteValued for MpcCpuWrapper {}
+// SAFETY: see above
 unsafe impl ByteValued for MpcIntsrcWrapper {}
+// SAFETY: see above
 unsafe impl ByteValued for MpcIoapicWrapper {}
+// SAFETY: see above
 unsafe impl ByteValued for MpcTableWrapper {}
+// SAFETY: see above
 unsafe impl ByteValued for MpcLintsrcWrapper {}
+// SAFETY: see above
 unsafe impl ByteValued for MpfIntelWrapper {}
 
 #[derive(Debug)]
@@ -95,7 +101,7 @@ const CPU_FEATURE_APIC: u32 = 0x200;
 const CPU_FEATURE_FPU: u32 = 0x001;
 
 fn compute_checksum<T: Copy>(v: &T) -> u8 {
-    // Safe because we are only reading the bytes within the size of the `T` reference `v`.
+    // SAFETY: we are only reading the bytes within the size of the `T` reference `v`.
     let v_slice = unsafe { slice::from_raw_parts(v as *const T as *const u8, mem::size_of::<T>()) };
     let mut checksum: u8 = 0;
     for i in v_slice.iter() {

arch/src/x86_64/smbios.rs

@@ -67,7 +67,7 @@ const PCI_SUPPORTED: u64 = 1 << 7;
 const IS_VIRTUAL_MACHINE: u8 = 1 << 4;
 
 fn compute_checksum<T: Copy>(v: &T) -> u8 {
-    // Safe because we are only reading the bytes within the size of the `T` reference `v`.
+    // SAFETY: we are only reading the bytes within the size of the `T` reference `v`.
     let v_slice = unsafe { slice::from_raw_parts(v as *const T as *const u8, mem::size_of::<T>()) };
     let mut checksum: u8 = 0;
     for i in v_slice.iter() {
@@ -145,11 +145,15 @@ struct SmbiosEndOfTable {
     handle: u16,
 }
 
-// SAFETY: These data structures only contain a series of integers
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for Smbios30Entrypoint {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for SmbiosBiosInfo {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for SmbiosSysInfo {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for SmbiosOemStrings {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for SmbiosEndOfTable {}
 
 fn write_and_incr<T: ByteValued>(

arch/src/x86_64/tdx/mod.rs

@@ -4,7 +4,9 @@
 use crate::GuestMemoryMmap;
 use std::fs::File;
 use std::io::{Read, Seek, SeekFrom};
+use std::str::FromStr;
 use thiserror::Error;
+use uuid::Uuid;
 use vm_memory::{ByteValued, Bytes, GuestAddress, GuestMemoryError};
 
 #[derive(Error, Debug)]
@@ -13,6 +15,8 @@ pub enum TdvfError {
     ReadDescriptor(#[source] std::io::Error),
     #[error("Failed read TDVF descriptor offset: {0}")]
     ReadDescriptorOffset(#[source] std::io::Error),
+    #[error("Failed read GUID table: {0}")]
+    ReadGuidTable(#[source] std::io::Error),
     #[error("Invalid descriptor signature")]
     InvalidDescriptorSignature,
     #[error("Invalid descriptor size")]
@@ -21,8 +25,13 @@ pub enum TdvfError {
     InvalidDescriptorVersion,
     #[error("Failed to write HOB details to guest memory: {0}")]
     GuestMemoryWriteHob(#[source] GuestMemoryError),
+    #[error("Failed to create Uuid: {0}")]
+    UuidCreation(#[source] uuid::Error),
 }
 
+const TABLE_FOOTER_GUID: &str = "96b582de-1fb2-45f7-baea-a366c55a082d";
+const TDVF_METADATA_OFFSET_GUID: &str = "e47a6535-984a-4798-865e-4685a7bf8ec2";
+
 // TDVF_DESCRIPTOR
 #[repr(packed)]
 #[derive(Default)]
@@ -59,7 +68,72 @@ pub enum TdvfSectionType {
     Reserved = 0xffffffff,
 }
 
-pub fn parse_tdvf_sections(file: &mut File) -> Result<Vec<TdvfSection>, TdvfError> {
+fn tdvf_descriptor_offset(file: &mut File) -> Result<(SeekFrom, bool), TdvfError> {
+    // Let's first try to identify the presence of the table footer GUID
+    file.seek(SeekFrom::End(-0x30))
+        .map_err(TdvfError::ReadGuidTable)?;
+    let mut table_footer_guid: [u8; 16] = [0; 16];
+    file.read_exact(&mut table_footer_guid)
+        .map_err(TdvfError::ReadGuidTable)?;
+    let uuid =
+        Uuid::from_slice_le(table_footer_guid.as_slice()).map_err(TdvfError::UuidCreation)?;
+    let expected_uuid = Uuid::from_str(TABLE_FOOTER_GUID).map_err(TdvfError::UuidCreation)?;
+    if uuid == expected_uuid {
+        // Retrieve the table size
+        file.seek(SeekFrom::End(-0x32))
+            .map_err(TdvfError::ReadGuidTable)?;
+        let mut table_size: [u8; 2] = [0; 2];
+        file.read_exact(&mut table_size)
+            .map_err(TdvfError::ReadGuidTable)?;
+        let table_size = u16::from_le_bytes(table_size) as usize;
+        let mut table: Vec<u8> = vec![0; table_size];
+
+        // Read the entire table
+        file.seek(SeekFrom::End(-(table_size as i64 + 0x20)))
+            .map_err(TdvfError::ReadGuidTable)?;
+        file.read_exact(table.as_mut_slice())
+            .map_err(TdvfError::ReadGuidTable)?;
+
+        // Let's start from the top and go backward down the table.
+        // We start after the footer GUID and the table length.
+        let mut offset = table_size - 18;
+
+        debug!("Parsing GUIDed structure");
+        while offset >= 18 {
+            let entry_uuid = Uuid::from_slice_le(&table[offset - 16..offset])
+                .map_err(TdvfError::UuidCreation)?;
+            let entry_size =
+                u16::from_le_bytes(table[offset - 18..offset - 16].try_into().unwrap()) as usize;
+            debug!(
+                "Entry GUID = {}, size = {}",
+                entry_uuid.hyphenated().to_string(),
+                entry_size
+            );
+
+            // Avoid going through an infinite loop if the entry size is 0
+            if entry_size == 0 {
+                break;
+            }
+
+            offset -= entry_size;
+
+            let expected_uuid =
+                Uuid::from_str(TDVF_METADATA_OFFSET_GUID).map_err(TdvfError::UuidCreation)?;
+            if entry_uuid == expected_uuid && entry_size == 22 {
+                return Ok((
+                    SeekFrom::End(
+                        -(u32::from_le_bytes(table[offset..offset + 4].try_into().unwrap()) as i64),
+                    ),
+                    true,
+                ));
+            }
+        }
+    }
+
+    // If we end up here, this means the firmware doesn't support the new way
+    // of exposing the TDVF descriptor offset through the table of GUIDs.
+    // That's why we fallback onto the deprecated method.
+
     // The 32-bit offset to the TDVF metadata is located 32 bytes from
     // the end of the file.
     // See "TDVF Metadata Pointer" in "TDX Virtual Firmware Design Guide
@@ -69,13 +143,21 @@ pub fn parse_tdvf_sections(file: &mut File) -> Result<Vec<TdvfSection>, TdvfErro
     let mut descriptor_offset: [u8; 4] = [0; 4];
     file.read_exact(&mut descriptor_offset)
         .map_err(TdvfError::ReadDescriptorOffset)?;
-    let descriptor_offset = u32::from_le_bytes(descriptor_offset) as u64;
 
-    file.seek(SeekFrom::Start(descriptor_offset))
+    Ok((
+        SeekFrom::Start(u32::from_le_bytes(descriptor_offset) as u64),
+        false,
+    ))
+}
+
+pub fn parse_tdvf_sections(file: &mut File) -> Result<(Vec<TdvfSection>, bool), TdvfError> {
+    let (descriptor_offset, guid_found) = tdvf_descriptor_offset(file)?;
+
+    file.seek(descriptor_offset)
         .map_err(TdvfError::ReadDescriptor)?;
 
     let mut descriptor: TdvfDescriptor = Default::default();
-    // Safe as we read exactly the size of the descriptor header
+    // SAFETY: we read exactly the size of the descriptor header
     file.read_exact(unsafe {
         std::slice::from_raw_parts_mut(
             &mut descriptor as *mut _ as *mut u8,
@@ -102,7 +184,7 @@ pub fn parse_tdvf_sections(file: &mut File) -> Result<Vec<TdvfSection>, TdvfErro
     let mut sections = Vec::new();
     sections.resize_with(descriptor.num_sections as usize, TdvfSection::default);
 
-    // Safe as we read exactly the advertised sections
+    // SAFETY: we read exactly the advertised sections
     file.read_exact(unsafe {
         std::slice::from_raw_parts_mut(
             sections.as_mut_ptr() as *mut u8,
@@ -111,7 +193,7 @@ pub fn parse_tdvf_sections(file: &mut File) -> Result<Vec<TdvfSection>, TdvfErro
     })
     .map_err(TdvfError::ReadDescriptor)?;
 
-    Ok(sections)
+    Ok((sections, guid_found))
 }
 
 #[repr(u16)]
@@ -196,12 +278,17 @@ struct TdPayload {
     payload_info: PayloadInfo,
 }
 
-// SAFETY: These data structures only contain a series of integers
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for HobHeader {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for HobHandoffInfoTable {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for HobResourceDescriptor {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for HobGuidType {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for PayloadInfo {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for TdPayload {}
 
 pub struct TdHob {
@@ -297,12 +384,19 @@ impl TdHob {
         physical_start: u64,
         resource_length: u64,
         ram: bool,
+        guid_found: bool,
     ) -> Result<(), TdvfError> {
         self.add_resource(
             mem,
             physical_start,
             resource_length,
             if ram {
+                if guid_found {
+                    0x7 /* EFI_RESOURCE_MEMORY_UNACCEPTED */
+                } else {
+                    0 /* EFI_RESOURCE_SYSTEM_MEMORY */
+                }
+            } else if guid_found {
                 0 /* EFI_RESOURCE_SYSTEM_MEMORY */
             } else {
                 0x5 /*EFI_RESOURCE_MEMORY_RESERVED */
@@ -430,7 +524,7 @@ mod tests {
     #[ignore]
     fn test_parse_tdvf_sections() {
         let mut f = std::fs::File::open("tdvf.fd").unwrap();
-        let sections = parse_tdvf_sections(&mut f).unwrap();
+        let (sections, _) = parse_tdvf_sections(&mut f).unwrap();
         for section in sections {
             eprintln!("{section:x?}")
         }

block_util/Cargo.toml

@@ -8,16 +8,17 @@ edition = "2021"
 default = []
 
 [dependencies]
-io-uring = "0.5.9"
-libc = "0.2.138"
+io-uring = "0.5.13"
+libc = "0.2.139"
 log = "0.4.17"
 qcow = { path = "../qcow" }
-thiserror = "1.0.37"
-versionize = "0.1.9"
+smallvec = "1.10.0"
+thiserror = "1.0.39"
+versionize = "0.1.10"
 versionize_derive = "0.1.4"
 vhdx = { path = "../vhdx" }
-virtio-bindings = { version = "0.1.0", features = ["virtio-v5_0_0"] }
-virtio-queue = "0.7.0"
+virtio-bindings = { version = "0.2.0", features = ["virtio-v5_0_0"] }
+virtio-queue = "0.7.1"
 vm-memory = { version = "0.10.0", features = ["backend-mmap", "backend-atomic", "backend-bitmap"] }
 vm-virtio = { path = "../vm-virtio" }
 vmm-sys-util = "0.11.0"

block_util/src/async_io.rs

@@ -3,7 +3,6 @@
 // SPDX-License-Identifier: Apache-2.0 AND BSD-3-Clause
 
 use libc::{ioctl, S_IFBLK, S_IFMT};
-use std::convert::TryInto;
 use std::fs::File;
 use std::os::unix::io::AsRawFd;
 use thiserror::Error;
@@ -54,19 +53,21 @@ enum BlockSize {
 impl DiskTopology {
     fn is_block_device(f: &mut File) -> std::io::Result<bool> {
         let mut stat = std::mem::MaybeUninit::<libc::stat>::uninit();
+        // SAFETY: FFI call with a valid fd and buffer
         let ret = unsafe { libc::fstat(f.as_raw_fd(), stat.as_mut_ptr()) };
         if ret != 0 {
             return Err(std::io::Error::last_os_error());
         }
 
+        // SAFETY: stat is valid at this point
         let is_block = unsafe { (*stat.as_ptr()).st_mode & S_IFMT == S_IFBLK };
         Ok(is_block)
     }
 
     // libc::ioctl() takes different types on different architectures
-    #[allow(clippy::useless_conversion)]
     fn query_block_size(f: &mut File, block_size_type: BlockSize) -> std::io::Result<u64> {
         let mut block_size = 0;
+        // SAFETY: FFI call with correct arguments
         let ret = unsafe {
             ioctl(
                 f.as_raw_fd(),
@@ -75,9 +76,7 @@ impl DiskTopology {
                     BlockSize::PhysicalBlock => BLKPBSZGET(),
                     BlockSize::MinimumIo => BLKIOMIN(),
                     BlockSize::OptimalIo => BLKIOOPT(),
-                }
-                .try_into()
-                .unwrap(),
+                } as _,
                 &mut block_size,
             )
         };
@@ -132,15 +131,15 @@ pub trait AsyncIo: Send {
     fn read_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()>;
     fn write_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()>;
     fn fsync(&mut self, user_data: Option<u64>) -> AsyncIoResult<()>;
-    fn complete(&mut self) -> Vec<(u64, i32)>;
+    fn next_completed_request(&mut self) -> Option<(u64, i32)>;
 }

block_util/src/fixed_vhd_async.rs

@@ -64,7 +64,7 @@ impl AsyncIo for FixedVhdAsync {
     fn read_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
         if offset as u64 >= self.size {
@@ -83,7 +83,7 @@ impl AsyncIo for FixedVhdAsync {
     fn write_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
         if offset as u64 >= self.size {
@@ -104,7 +104,7 @@ impl AsyncIo for FixedVhdAsync {
         self.raw_file_async.fsync(user_data)
     }
 
-    fn complete(&mut self) -> Vec<(u64, i32)> {
-        self.raw_file_async.complete()
+    fn next_completed_request(&mut self) -> Option<(u64, i32)> {
+        self.raw_file_async.next_completed_request()
     }
 }

block_util/src/fixed_vhd_sync.rs

@@ -62,7 +62,7 @@ impl AsyncIo for FixedVhdSync {
     fn read_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
         if offset as u64 >= self.size {
@@ -81,7 +81,7 @@ impl AsyncIo for FixedVhdSync {
     fn write_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
         if offset as u64 >= self.size {
@@ -101,7 +101,7 @@ impl AsyncIo for FixedVhdSync {
         self.raw_file_sync.fsync(user_data)
     }
 
-    fn complete(&mut self) -> Vec<(u64, i32)> {
-        self.raw_file_sync.complete()
+    fn next_completed_request(&mut self) -> Option<(u64, i32)> {
+        self.raw_file_sync.next_completed_request()
     }
 }

block_util/src/lib.rs

@@ -22,8 +22,10 @@ pub mod vhdx_sync;
 
 use crate::async_io::{AsyncIo, AsyncIoError, AsyncIoResult};
 use io_uring::{opcode, IoUring, Probe};
+use smallvec::SmallVec;
 use std::alloc::{alloc_zeroed, dealloc, Layout};
 use std::cmp;
+use std::collections::VecDeque;
 use std::convert::TryInto;
 use std::fs::File;
 use std::io::{self, IoSlice, IoSliceMut, Read, Seek, SeekFrom, Write};
@@ -32,10 +34,11 @@ use std::path::Path;
 use std::result;
 use std::sync::Arc;
 use std::sync::MutexGuard;
+use std::time::Instant;
 use thiserror::Error;
 use versionize::{VersionMap, Versionize, VersionizeResult};
 use versionize_derive::Versionize;
-use virtio_bindings::bindings::virtio_blk::*;
+use virtio_bindings::virtio_blk::*;
 use virtio_queue::DescriptorChain;
 use vm_memory::{
     bitmap::AtomicBitmap, bitmap::Bitmap, ByteValued, Bytes, GuestAddress, GuestMemory,
@@ -195,10 +198,11 @@ pub struct AlignedOperation {
 pub struct Request {
     pub request_type: RequestType,
     pub sector: u64,
-    pub data_descriptors: Vec<(GuestAddress, u32)>,
+    pub data_descriptors: SmallVec<[(GuestAddress, u32); 1]>,
     pub status_addr: GuestAddress,
     pub writeback: bool,
-    pub aligned_operations: Vec<AlignedOperation>,
+    pub aligned_operations: SmallVec<[AlignedOperation; 1]>,
+    pub start: Instant,
 }
 
 impl Request {
@@ -226,10 +230,11 @@ impl Request {
         let mut req = Request {
             request_type: request_type(desc_chain.memory(), hdr_desc_addr)?,
             sector: sector(desc_chain.memory(), hdr_desc_addr)?,
-            data_descriptors: Vec::new(),
+            data_descriptors: SmallVec::with_capacity(1),
             status_addr: GuestAddress(0),
             writeback: true,
-            aligned_operations: Vec::new(),
+            aligned_operations: SmallVec::with_capacity(1),
+            start: Instant::now(),
         };
 
         let status_desc;
@@ -249,6 +254,7 @@ impl Request {
                 return Err(Error::DescriptorChainTooShort);
             }
         } else {
+            req.data_descriptors.reserve_exact(1);
             while desc.has_next() {
                 if desc.is_write_only() && req.request_type == RequestType::Out {
                     return Err(Error::UnexpectedWriteOnlyDescriptor);
@@ -353,7 +359,8 @@ impl Request {
         let request_type = self.request_type;
         let offset = (sector << SECTOR_SHIFT) as libc::off_t;
 
-        let mut iovecs = Vec::new();
+        let mut iovecs: SmallVec<[libc::iovec; 1]> =
+            SmallVec::with_capacity(self.data_descriptors.len());
         for (data_addr, data_len) in &self.data_descriptors {
             if *data_len == 0 {
                 continue;
@@ -381,7 +388,7 @@ impl Request {
             let iov_base = if (origin_ptr as u64) % SECTOR_SIZE != 0 {
                 let layout =
                     Layout::from_size_align(*data_len as usize, SECTOR_SIZE as usize).unwrap();
-                // Safe because layout has non-zero size
+                // SAFETY: layout has non-zero size
                 let aligned_ptr = unsafe { alloc_zeroed(layout) };
                 if aligned_ptr.is_null() {
                     return Err(ExecuteError::TemporaryBufferAllocation(
@@ -392,7 +399,7 @@ impl Request {
                 // We need to perform the copy beforehand in case we're writing
                 // data out.
                 if request_type == RequestType::Out {
-                    // Safe because destination buffer has been allocated with
+                    // SAFETY: destination buffer has been allocated with
                     // the proper size.
                     unsafe {
                         std::ptr::copy(origin_ptr as *const u8, aligned_ptr, *data_len as usize)
@@ -430,12 +437,12 @@ impl Request {
                         .mark_dirty(0, *data_len as usize);
                 }
                 disk_image
-                    .read_vectored(offset, iovecs, user_data)
+                    .read_vectored(offset, &iovecs, user_data)
                     .map_err(ExecuteError::AsyncRead)?;
             }
             RequestType::Out => {
                 disk_image
-                    .write_vectored(offset, iovecs, user_data)
+                    .write_vectored(offset, &iovecs, user_data)
                     .map_err(ExecuteError::AsyncWrite)?;
             }
             RequestType::Flush => {
@@ -467,7 +474,7 @@ impl Request {
             // We need to perform the copy after the data has been read inside
             // the aligned buffer in case we're reading data in.
             if self.request_type == RequestType::In {
-                // Safe because origin buffer has been allocated with the
+                // SAFETY: origin buffer has been allocated with the
                 // proper size.
                 unsafe {
                     std::ptr::copy(
@@ -479,7 +486,7 @@ impl Request {
             }
 
             // Free the temporary aligned buffer.
-            // Safe because aligned_ptr was allocated by alloc_zeroed with the same
+            // SAFETY: aligned_ptr was allocated by alloc_zeroed with the same
             // layout
             unsafe {
                 dealloc(
@@ -528,8 +535,9 @@ pub struct VirtioBlockGeometry {
     pub sectors: u8,
 }
 
-// SAFETY: these data structures only contain a series of integers
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for VirtioBlockConfig {}
+// SAFETY: data structure only contain a series of integers
 unsafe impl ByteValued for VirtioBlockGeometry {}
 
 /// Check if io_uring for block device can be used on the current system, as
@@ -588,14 +596,15 @@ where
     fn read_vectored_sync(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
         eventfd: &EventFd,
-        completion_list: &mut Vec<(u64, i32)>,
+        completion_list: &mut VecDeque<(u64, i32)>,
     ) -> AsyncIoResult<()> {
         // Convert libc::iovec into IoSliceMut
-        let mut slices = Vec::new();
+        let mut slices: SmallVec<[IoSliceMut; 1]> = SmallVec::with_capacity(iovecs.len());
         for iovec in iovecs.iter() {
+            // SAFETY: on Linux IoSliceMut wraps around libc::iovec
             slices.push(IoSliceMut::new(unsafe { std::mem::transmute(*iovec) }));
         }
 
@@ -611,7 +620,7 @@ where
                 .map_err(AsyncIoError::ReadVectored)?
         };
 
-        completion_list.push((user_data, result as i32));
+        completion_list.push_back((user_data, result as i32));
         eventfd.write(1).unwrap();
 
         Ok(())
@@ -620,14 +629,15 @@ where
     fn write_vectored_sync(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
         eventfd: &EventFd,
-        completion_list: &mut Vec<(u64, i32)>,
+        completion_list: &mut VecDeque<(u64, i32)>,
     ) -> AsyncIoResult<()> {
         // Convert libc::iovec into IoSlice
-        let mut slices = Vec::new();
+        let mut slices: SmallVec<[IoSlice; 1]> = SmallVec::with_capacity(iovecs.len());
         for iovec in iovecs.iter() {
+            // SAFETY: on Linux IoSlice wraps around libc::iovec
             slices.push(IoSlice::new(unsafe { std::mem::transmute(*iovec) }));
         }
 
@@ -643,7 +653,7 @@ where
                 .map_err(AsyncIoError::WriteVectored)?
         };
 
-        completion_list.push((user_data, result as i32));
+        completion_list.push_back((user_data, result as i32));
         eventfd.write(1).unwrap();
 
         Ok(())
@@ -653,7 +663,7 @@ where
         &mut self,
         user_data: Option<u64>,
         eventfd: &EventFd,
-        completion_list: &mut Vec<(u64, i32)>,
+        completion_list: &mut VecDeque<(u64, i32)>,
     ) -> AsyncIoResult<()> {
         let result: i32 = {
             let mut file = self.file();
@@ -665,7 +675,7 @@ where
         };
 
         if let Some(user_data) = user_data {
-            completion_list.push((user_data, result));
+            completion_list.push_back((user_data, result));
             eventfd.write(1).unwrap();
         }
 

block_util/src/qcow_sync.rs

@@ -5,6 +5,7 @@
 use crate::async_io::{AsyncIo, AsyncIoResult, DiskFile, DiskFileError, DiskFileResult};
 use crate::AsyncAdaptor;
 use qcow::{QcowFile, RawFile, Result as QcowResult};
+use std::collections::VecDeque;
 use std::fs::File;
 use std::io::{Seek, SeekFrom};
 use std::sync::{Arc, Mutex, MutexGuard};
@@ -37,7 +38,7 @@ impl DiskFile for QcowDiskSync {
 pub struct QcowSync {
     qcow_file: Arc<Mutex<QcowFile>>,
     eventfd: EventFd,
-    completion_list: Vec<(u64, i32)>,
+    completion_list: VecDeque<(u64, i32)>,
 }
 
 impl QcowSync {
@@ -46,7 +47,7 @@ impl QcowSync {
             qcow_file,
             eventfd: EventFd::new(libc::EFD_NONBLOCK)
                 .expect("Failed creating EventFd for QcowSync"),
-            completion_list: Vec::new(),
+            completion_list: VecDeque::new(),
         }
     }
 }
@@ -65,7 +66,7 @@ impl AsyncIo for QcowSync {
     fn read_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
         self.qcow_file.read_vectored_sync(
@@ -80,7 +81,7 @@ impl AsyncIo for QcowSync {
     fn write_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
         self.qcow_file.write_vectored_sync(
@@ -97,7 +98,7 @@ impl AsyncIo for QcowSync {
             .fsync_sync(user_data, &self.eventfd, &mut self.completion_list)
     }
 
-    fn complete(&mut self) -> Vec<(u64, i32)> {
-        self.completion_list.drain(..).collect()
+    fn next_completed_request(&mut self) -> Option<(u64, i32)> {
+        self.completion_list.pop_front()
     }
 }

block_util/src/raw_async.rs

@@ -76,12 +76,12 @@ impl AsyncIo for RawFileAsync {
     fn read_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
         let (submitter, mut sq, _) = self.io_uring.split();
 
-        // Safe because we know the file descriptor is valid and we
+        // SAFETY: we know the file descriptor is valid and we
         // relied on vm-memory to provide the buffer address.
         let _ = unsafe {
             sq.push(
@@ -104,12 +104,12 @@ impl AsyncIo for RawFileAsync {
     fn write_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
         let (submitter, mut sq, _) = self.io_uring.split();
 
-        // Safe because we know the file descriptor is valid and we
+        // SAFETY: we know the file descriptor is valid and we
         // relied on vm-memory to provide the buffer address.
         let _ = unsafe {
             sq.push(
@@ -133,7 +133,7 @@ impl AsyncIo for RawFileAsync {
         if let Some(user_data) = user_data {
             let (submitter, mut sq, _) = self.io_uring.split();
 
-            // Safe because we know the file descriptor is valid.
+            // SAFETY: we know the file descriptor is valid.
             let _ = unsafe {
                 sq.push(
                     &opcode::Fsync::new(types::Fd(self.fd))
@@ -148,20 +148,17 @@ impl AsyncIo for RawFileAsync {
             sq.sync();
             submitter.submit().map_err(AsyncIoError::Fsync)?;
         } else {
+            // SAFETY: FFI call with a valid fd
             unsafe { libc::fsync(self.fd) };
         }
 
         Ok(())
     }
 
-    fn complete(&mut self) -> Vec<(u64, i32)> {
-        let mut completion_list = Vec::new();
-
-        let cq = self.io_uring.completion();
-        for cq_entry in cq {
-            completion_list.push((cq_entry.user_data(), cq_entry.result()));
-        }
-
-        completion_list
+    fn next_completed_request(&mut self) -> Option<(u64, i32)> {
+        self.io_uring
+            .completion()
+            .next()
+            .map(|entry| (entry.user_data(), entry.result()))
     }
 }

block_util/src/raw_sync.rs

@@ -5,6 +5,7 @@
 use crate::async_io::{
     AsyncIo, AsyncIoError, AsyncIoResult, DiskFile, DiskFileError, DiskFileResult, DiskTopology,
 };
+use std::collections::VecDeque;
 use std::fs::File;
 use std::io::{Seek, SeekFrom};
 use std::os::unix::io::{AsRawFd, RawFd};
@@ -44,7 +45,7 @@ impl DiskFile for RawFileDiskSync {
 pub struct RawFileSync {
     fd: RawFd,
     eventfd: EventFd,
-    completion_list: Vec<(u64, i32)>,
+    completion_list: VecDeque<(u64, i32)>,
 }
 
 impl RawFileSync {
@@ -52,7 +53,7 @@ impl RawFileSync {
         RawFileSync {
             fd,
             eventfd: EventFd::new(libc::EFD_NONBLOCK).expect("Failed creating EventFd for RawFile"),
-            completion_list: Vec::new(),
+            completion_list: VecDeque::new(),
         }
     }
 }
@@ -65,13 +66,14 @@ impl AsyncIo for RawFileSync {
     fn read_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
+        // SAFETY: FFI call with valid arguments
         let result = unsafe {
             libc::preadv(
                 self.fd as libc::c_int,
-                iovecs.as_ptr(),
+                iovecs.as_ptr() as *const libc::iovec,
                 iovecs.len() as libc::c_int,
                 offset,
             )
@@ -80,7 +82,7 @@ impl AsyncIo for RawFileSync {
             return Err(AsyncIoError::ReadVectored(std::io::Error::last_os_error()));
         }
 
-        self.completion_list.push((user_data, result as i32));
+        self.completion_list.push_back((user_data, result as i32));
         self.eventfd.write(1).unwrap();
 
         Ok(())
@@ -89,13 +91,14 @@ impl AsyncIo for RawFileSync {
     fn write_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
+        // SAFETY: FFI call with valid arguments
         let result = unsafe {
             libc::pwritev(
                 self.fd as libc::c_int,
-                iovecs.as_ptr(),
+                iovecs.as_ptr() as *const libc::iovec,
                 iovecs.len() as libc::c_int,
                 offset,
             )
@@ -104,27 +107,28 @@ impl AsyncIo for RawFileSync {
             return Err(AsyncIoError::WriteVectored(std::io::Error::last_os_error()));
         }
 
-        self.completion_list.push((user_data, result as i32));
+        self.completion_list.push_back((user_data, result as i32));
         self.eventfd.write(1).unwrap();
 
         Ok(())
     }
 
     fn fsync(&mut self, user_data: Option<u64>) -> AsyncIoResult<()> {
+        // SAFETY: FFI call
         let result = unsafe { libc::fsync(self.fd as libc::c_int) };
         if result < 0 {
             return Err(AsyncIoError::Fsync(std::io::Error::last_os_error()));
         }
 
         if let Some(user_data) = user_data {
-            self.completion_list.push((user_data, result));
+            self.completion_list.push_back((user_data, result));
             self.eventfd.write(1).unwrap();
         }
 
         Ok(())
     }
 
-    fn complete(&mut self) -> Vec<(u64, i32)> {
-        self.completion_list.drain(..).collect()
+    fn next_completed_request(&mut self) -> Option<(u64, i32)> {
+        self.completion_list.pop_front()
     }
 }

block_util/src/vhdx_sync.rs

@@ -4,6 +4,7 @@
 
 use crate::async_io::{AsyncIo, AsyncIoResult, DiskFile, DiskFileError, DiskFileResult};
 use crate::AsyncAdaptor;
+use std::collections::VecDeque;
 use std::fs::File;
 use std::sync::{Arc, Mutex, MutexGuard};
 use vhdx::vhdx::{Result as VhdxResult, Vhdx};
@@ -37,7 +38,7 @@ impl DiskFile for VhdxDiskSync {
 pub struct VhdxSync {
     vhdx_file: Arc<Mutex<Vhdx>>,
     eventfd: EventFd,
-    completion_list: Vec<(u64, i32)>,
+    completion_list: VecDeque<(u64, i32)>,
 }
 
 impl VhdxSync {
@@ -45,7 +46,7 @@ impl VhdxSync {
         Ok(VhdxSync {
             vhdx_file,
             eventfd: EventFd::new(libc::EFD_NONBLOCK)?,
-            completion_list: Vec::new(),
+            completion_list: VecDeque::new(),
         })
     }
 }
@@ -64,7 +65,7 @@ impl AsyncIo for VhdxSync {
     fn read_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
         self.vhdx_file.read_vectored_sync(
@@ -79,7 +80,7 @@ impl AsyncIo for VhdxSync {
     fn write_vectored(
         &mut self,
         offset: libc::off_t,
-        iovecs: Vec<libc::iovec>,
+        iovecs: &[libc::iovec],
         user_data: u64,
     ) -> AsyncIoResult<()> {
         self.vhdx_file.write_vectored_sync(
@@ -96,7 +97,7 @@ impl AsyncIo for VhdxSync {
             .fsync_sync(user_data, &self.eventfd, &mut self.completion_list)
     }
 
-    fn complete(&mut self) -> Vec<(u64, i32)> {
-        self.completion_list.drain(..).collect()
+    fn next_completed_request(&mut self) -> Option<(u64, i32)> {
+        self.completion_list.pop_front()
     }
 }

build.rs

@@ -3,13 +3,10 @@
 // SPDX-License-Identifier: Apache-2.0
 //
 
-#[macro_use(crate_version)]
-extern crate clap;
-
 use std::process::Command;
 
 fn main() {
-    let mut version = "v".to_owned() + crate_version!();
+    let mut version = "v".to_owned() + env!("CARGO_PKG_VERSION");
 
     if let Ok(git_out) = Command::new("git").args(["describe", "--dirty"]).output() {
         if git_out.status.success() {
@@ -20,8 +17,8 @@ fn main() {
     }
 
     // This println!() has a special behavior, as it will set the environment
-    // variable BUILT_VERSION, so that it can be reused from the binary.
+    // variable BUILD_VERSION, so that it can be reused from the binary.
     // Particularly, this is used from src/main.rs to display the exact
     // version.
-    println!("cargo:rustc-env=BUILT_VERSION={version}");
+    println!("cargo:rustc-env=BUILD_VERSION={version}");
 }

devices/Cargo.toml

@@ -5,18 +5,17 @@ authors = ["The Chromium OS Authors"]
 edition = "2021"
 
 [dependencies]
-acpi_tables = { path = "../acpi_tables" }
-anyhow = "1.0.66"
+acpi_tables = { git = "https://github.com/rust-vmm/acpi_tables", branch = "main"  }
+anyhow = "1.0.70"
 arch = { path = "../arch" }
-bitflags = "2.4.1"
+bitflags = "1.3.2"
 byteorder = "1.4.3"
 hypervisor = { path = "../hypervisor" }
-libc = "0.2.138"
+libc = "0.2.139"
 log = "0.4.17"
-phf = { version = "0.11.1", features = ["macros"] }
-thiserror = "1.0.37"
+thiserror = "1.0.39"
 tpm = { path = "../tpm" }
-versionize = "0.1.9"
+versionize = "0.1.10"
 versionize_derive = "0.1.4"
 vm-device = { path = "../vm-device" }
 vm-memory = "0.10.0"

devices/src/acpi.rs

@@ -4,7 +4,7 @@
 //
 
 use super::AcpiNotificationFlags;
-use acpi_tables::{aml, aml::Aml};
+use acpi_tables::{aml, Aml, AmlSink};
 use std::sync::{Arc, Barrier};
 use std::time::Instant;
 use vm_device::interrupt::InterruptSourceGroup;
@@ -103,11 +103,11 @@ impl BusDevice for AcpiGedDevice {
 }
 
 impl Aml for AcpiGedDevice {
-    fn append_aml_bytes(&self, bytes: &mut Vec<u8>) {
+    fn to_aml_bytes(&self, sink: &mut dyn AmlSink) {
         aml::Device::new(
             "_SB_.GEC_".into(),
             vec![
-                &aml::Name::new("_HID".into(), &aml::EisaName::new("PNP0A06")),
+                &aml::Name::new("_HID".into(), &aml::EISAName::new("PNP0A06")),
                 &aml::Name::new("_UID".into(), &"Generic Event Controller"),
                 &aml::Name::new(
                     "_CRS".into(),
@@ -121,12 +121,13 @@ impl Aml for AcpiGedDevice {
                 &aml::OpRegion::new(
                     "GDST".into(),
                     aml::OpRegionSpace::SystemMemory,
-                    self.address.0 as usize,
-                    GED_DEVICE_ACPI_SIZE,
+                    &(self.address.0 as usize),
+                    &GED_DEVICE_ACPI_SIZE,
                 ),
                 &aml::Field::new(
                     "GDST".into(),
                     aml::FieldAccessType::Byte,
+                    aml::FieldLockRule::NoLock,
                     aml::FieldUpdateRule::WriteAsZeroes,
                     vec![aml::FieldEntry::Named(*b"GDAT", 8)],
                 ),
@@ -163,7 +164,7 @@ impl Aml for AcpiGedDevice {
                 ),
             ],
         )
-        .append_aml_bytes(bytes);
+        .to_aml_bytes(sink);
         aml::Device::new(
             "_SB_.GED_".into(),
             vec![
@@ -187,7 +188,7 @@ impl Aml for AcpiGedDevice {
                 ),
             ],
         )
-        .append_aml_bytes(bytes)
+        .to_aml_bytes(sink)
     }
 }
 

devices/src/gic.rs

@@ -8,7 +8,7 @@ use anyhow::anyhow;
 use arch::layout;
 use hypervisor::{
     arch::aarch64::gic::{Vgic, VgicConfig},
-    CpuState,
+    CpuState, GicState,
 };
 use std::result;
 use std::sync::{Arc, Mutex};
@@ -25,6 +25,7 @@ type Result<T> = result::Result<T, Error>;
 // Reserve 32 IRQs for legacy devices.
 pub const IRQ_LEGACY_BASE: usize = layout::IRQ_BASE as usize;
 pub const IRQ_LEGACY_COUNT: usize = 32;
+pub const GIC_SNAPSHOT_ID: &str = "gic-v3-its";
 
 // Gic (Generic Interupt Controller) struct provides all the functionality of a
 // GIC device. It wraps a hypervisor-emulated GIC device (Vgic) provided by the
@@ -39,8 +40,9 @@ pub struct Gic {
 
 impl Gic {
     pub fn new(
-        _vcpu_count: u8,
+        vcpu_count: u8,
         interrupt_manager: Arc<dyn InterruptManager<GroupConfig = MsiIrqGroupConfig>>,
+        vm: Arc<dyn hypervisor::Vm>,
     ) -> Result<Gic> {
         let interrupt_source_group = interrupt_manager
             .create_group(MsiIrqGroupConfig {
@@ -49,45 +51,34 @@ impl Gic {
             })
             .map_err(Error::CreateInterruptSourceGroup)?;
 
-        Ok(Gic {
+        let vgic = vm
+            .create_vgic(Gic::create_default_config(vcpu_count as u64))
+            .map_err(Error::CreateGic)?;
+
+        let gic = Gic {
             interrupt_source_group,
-            vgic: None,
-        })
+            vgic: Some(vgic),
+        };
+        gic.enable()?;
+
+        Ok(gic)
     }
 
-    /// Default config implied by arch::layout
-    pub fn create_default_config(vcpu_count: u64) -> VgicConfig {
-        let redists_size = layout::GIC_V3_REDIST_SIZE * vcpu_count;
-        let redists_addr = layout::GIC_V3_DIST_START.raw_value() - redists_size;
-        VgicConfig {
-            vcpu_count,
-            dist_addr: layout::GIC_V3_DIST_START.raw_value(),
-            dist_size: layout::GIC_V3_DIST_SIZE,
-            redists_addr,
-            redists_size,
-            msi_addr: redists_addr - layout::GIC_V3_ITS_SIZE,
-            msi_size: layout::GIC_V3_ITS_SIZE,
-            nr_irqs: layout::IRQ_NUM,
-        }
-    }
-
-    pub fn create_vgic(
+    pub fn restore_vgic(
         &mut self,
-        vm: &Arc<dyn hypervisor::Vm>,
-        config: VgicConfig,
-    ) -> Result<Arc<Mutex<dyn Vgic>>> {
-        let vgic = vm.create_vgic(config).map_err(Error::CreateGic)?;
-        self.vgic = Some(vgic.clone());
-        Ok(vgic.clone())
+        state: Option<GicState>,
+        saved_vcpu_states: &[CpuState],
+    ) -> Result<()> {
+        self.set_gicr_typers(saved_vcpu_states);
+        self.vgic
+            .clone()
+            .unwrap()
+            .lock()
+            .unwrap()
+            .set_state(&state.unwrap())
+            .map_err(Error::RestoreGic)
     }
 
-    pub fn set_gicr_typers(&mut self, vcpu_states: &[CpuState]) {
-        let vgic = self.vgic.as_ref().unwrap().clone();
-        vgic.lock().unwrap().set_gicr_typers(vcpu_states);
-    }
-}
-
-impl InterruptController for Gic {
     fn enable(&self) -> Result<()> {
         // Set irqfd for legacy interrupts
         self.interrupt_source_group
@@ -113,6 +104,33 @@ impl InterruptController for Gic {
         Ok(())
     }
 
+    /// Default config implied by arch::layout
+    pub fn create_default_config(vcpu_count: u64) -> VgicConfig {
+        let redists_size = layout::GIC_V3_REDIST_SIZE * vcpu_count;
+        let redists_addr = layout::GIC_V3_DIST_START.raw_value() - redists_size;
+        VgicConfig {
+            vcpu_count,
+            dist_addr: layout::GIC_V3_DIST_START.raw_value(),
+            dist_size: layout::GIC_V3_DIST_SIZE,
+            redists_addr,
+            redists_size,
+            msi_addr: redists_addr - layout::GIC_V3_ITS_SIZE,
+            msi_size: layout::GIC_V3_ITS_SIZE,
+            nr_irqs: layout::IRQ_NUM,
+        }
+    }
+
+    pub fn get_vgic(&mut self) -> Result<Arc<Mutex<dyn Vgic>>> {
+        Ok(self.vgic.clone().unwrap())
+    }
+
+    pub fn set_gicr_typers(&mut self, vcpu_states: &[CpuState]) {
+        let vgic = self.vgic.as_ref().unwrap().clone();
+        vgic.lock().unwrap().set_gicr_typers(vcpu_states);
+    }
+}
+
+impl InterruptController for Gic {
     // This should be called anytime an interrupt needs to be injected into the
     // running guest.
     fn service_irq(&mut self, irq: usize) -> Result<()> {
@@ -128,27 +146,15 @@ impl InterruptController for Gic {
     }
 }
 
-pub const GIC_V3_ITS_SNAPSHOT_ID: &str = "gic-v3-its";
 impl Snapshottable for Gic {
     fn id(&self) -> String {
-        GIC_V3_ITS_SNAPSHOT_ID.to_string()
+        GIC_SNAPSHOT_ID.to_string()
     }
 
     fn snapshot(&mut self) -> std::result::Result<Snapshot, MigratableError> {
         let vgic = self.vgic.as_ref().unwrap().clone();
         let state = vgic.lock().unwrap().state().unwrap();
-        Snapshot::new_from_state(&self.id(), &state)
-    }
-
-    fn restore(&mut self, snapshot: Snapshot) -> std::result::Result<(), MigratableError> {
-        let vgic = self.vgic.as_ref().unwrap().clone();
-        vgic.lock()
-            .unwrap()
-            .set_state(&snapshot.to_state(&self.id())?)
-            .map_err(|e| {
-                MigratableError::Restore(anyhow!("Could not restore GICv3ITS state {:?}", e))
-            })?;
-        Ok(())
+        Snapshot::new_from_state(&state)
     }
 }
 

devices/src/interrupt_controller.rs

@@ -24,8 +24,12 @@ pub enum Error {
     UpdateInterrupt(io::Error),
     /// Failed enabling the interrupt.
     EnableInterrupt(io::Error),
+    #[cfg(target_arch = "aarch64")]
     /// Failed creating GIC device.
     CreateGic(hypervisor::HypervisorVmError),
+    #[cfg(target_arch = "aarch64")]
+    /// Failed restoring GIC device.
+    RestoreGic(hypervisor::arch::aarch64::gic::Error),
 }
 
 type Result<T> = result::Result<T, Error>;
@@ -55,8 +59,6 @@ pub struct MsiMessage {
 // IOAPIC (X86) or GIC (Arm).
 pub trait InterruptController: Send {
     fn service_irq(&mut self, irq: usize) -> Result<()>;
-    #[cfg(target_arch = "aarch64")]
-    fn enable(&self) -> Result<()>;
     #[cfg(target_arch = "x86_64")]
     fn end_of_interrupt(&mut self, vec: u8);
     fn notifier(&self, irq: usize) -> Option<EventFd>;

devices/src/ioapic.rs

@@ -10,7 +10,6 @@
 // See https://pdos.csail.mit.edu/6.828/2016/readings/ia32/ioapic.pdf for a specification.
 
 use super::interrupt_controller::{Error, InterruptController};
-use anyhow::anyhow;
 use byteorder::{ByteOrder, LittleEndian};
 use std::result;
 use std::sync::{Arc, Barrier};
@@ -194,6 +193,7 @@ impl Ioapic {
         id: String,
         apic_address: GuestAddress,
         interrupt_manager: Arc<dyn InterruptManager<GroupConfig = MsiIrqGroupConfig>>,
+        state: Option<IoapicState>,
     ) -> Result<Ioapic> {
         let interrupt_source_group = interrupt_manager
             .create_group(MsiIrqGroupConfig {
@@ -202,17 +202,47 @@ impl Ioapic {
             })
             .map_err(Error::CreateInterruptSourceGroup)?;
 
+        let (id_reg, reg_sel, reg_entries, used_entries, apic_address) = if let Some(state) = &state
+        {
+            (
+                state.id_reg,
+                state.reg_sel,
+                state.reg_entries,
+                state.used_entries,
+                GuestAddress(state.apic_address),
+            )
+        } else {
+            (
+                0,
+                0,
+                [0x10000; NUM_IOAPIC_PINS],
+                [false; NUM_IOAPIC_PINS],
+                apic_address,
+            )
+        };
+
         // The IOAPIC is created with entries already masked. The guest will be
         // in charge of unmasking them if/when necessary.
-        Ok(Ioapic {
+        let ioapic = Ioapic {
             id,
-            id_reg: 0,
-            reg_sel: 0,
-            reg_entries: [0x10000; NUM_IOAPIC_PINS],
-            used_entries: [false; NUM_IOAPIC_PINS],
+            id_reg,
+            reg_sel,
+            reg_entries,
+            used_entries,
             apic_address,
             interrupt_source_group,
-        })
+        };
+
+        // When restoring the Ioapic, we must enable used entries.
+        if state.is_some() {
+            for (irq, entry) in ioapic.used_entries.iter().enumerate() {
+                if *entry {
+                    ioapic.update_entry(irq)?;
+                }
+            }
+        }
+
+        Ok(ioapic)
     }
 
     fn ioapic_write(&mut self, val: u32) {
@@ -299,21 +329,6 @@ impl Ioapic {
         }
     }
 
-    fn set_state(&mut self, state: &IoapicState) -> Result<()> {
-        self.id_reg = state.id_reg;
-        self.reg_sel = state.reg_sel;
-        self.reg_entries = state.reg_entries;
-        self.used_entries = state.used_entries;
-        self.apic_address = GuestAddress(state.apic_address);
-        for (irq, entry) in self.used_entries.iter().enumerate() {
-            if *entry {
-                self.update_entry(irq)?;
-            }
-        }
-
-        Ok(())
-    }
-
     fn update_entry(&self, irq: usize) -> Result<()> {
         let entry = self.reg_entries[irq];
 
@@ -423,18 +438,7 @@ impl Snapshottable for Ioapic {
     }
 
     fn snapshot(&mut self) -> std::result::Result<Snapshot, MigratableError> {
-        Snapshot::new_from_versioned_state(&self.id, &self.state())
-    }
-
-    fn restore(&mut self, snapshot: Snapshot) -> std::result::Result<(), MigratableError> {
-        self.set_state(&snapshot.to_versioned_state(&self.id)?)
-            .map_err(|e| {
-                MigratableError::Restore(anyhow!(
-                    "Could not restore state for {}: {:?}",
-                    self.id,
-                    e
-                ))
-            })
+        Snapshot::new_from_versioned_state(&self.state())
     }
 }
 

devices/src/legacy/cmos.rs

@@ -97,7 +97,7 @@ impl BusDevice for Cmos {
                 let day;
                 let month;
                 let year;
-                // The clock_gettime and gmtime_r calls are safe as long as the structs they are
+                // SAFETY: The clock_gettime and gmtime_r calls are safe as long as the structs they are
                 // given are large enough, and neither of them fail. It is safe to zero initialize
                 // the tm and timespec struct because it contains only plain data.
                 let update_in_progress = unsafe {

devices/src/legacy/gpio_pl061.rs

@@ -106,18 +106,39 @@ impl VersionMapped for GpioState {}
 
 impl Gpio {
     /// Constructs an PL061 GPIO device.
-    pub fn new(id: String, interrupt: Arc<dyn InterruptSourceGroup>) -> Self {
+    pub fn new(
+        id: String,
+        interrupt: Arc<dyn InterruptSourceGroup>,
+        state: Option<GpioState>,
+    ) -> Self {
+        let (data, old_in_data, dir, isense, ibe, iev, im, istate, afsel) =
+            if let Some(state) = state {
+                (
+                    state.data,
+                    state.old_in_data,
+                    state.dir,
+                    state.isense,
+                    state.ibe,
+                    state.iev,
+                    state.im,
+                    state.istate,
+                    state.afsel,
+                )
+            } else {
+                (0, 0, 0, 0, 0, 0, 0, 0, 0)
+            };
+
         Self {
             id,
-            data: 0,
-            old_in_data: 0,
-            dir: 0,
-            isense: 0,
-            ibe: 0,
-            iev: 0,
-            im: 0,
-            istate: 0,
-            afsel: 0,
+            data,
+            old_in_data,
+            dir,
+            isense,
+            ibe,
+            iev,
+            im,
+            istate,
+            afsel,
             interrupt,
         }
     }
@@ -136,24 +157,12 @@ impl Gpio {
         }
     }
 
-    fn set_state(&mut self, state: &GpioState) {
-        self.data = state.data;
-        self.old_in_data = state.old_in_data;
-        self.dir = state.dir;
-        self.isense = state.isense;
-        self.ibe = state.ibe;
-        self.iev = state.iev;
-        self.im = state.im;
-        self.istate = state.istate;
-        self.afsel = state.afsel;
-    }
-
     fn pl061_internal_update(&mut self) {
         // FIXME:
         //  Missing Output Interrupt Emulation.
 
         // Input Edging Interrupt Emulation.
-        let changed = ((self.old_in_data ^ self.data) & !self.dir) as u32;
+        let changed = (self.old_in_data ^ self.data) & !self.dir;
         if changed > 0 {
             self.old_in_data = self.data;
             for i in 0..N_GPIOS {
@@ -319,12 +328,7 @@ impl Snapshottable for Gpio {
     }
 
     fn snapshot(&mut self) -> std::result::Result<Snapshot, MigratableError> {
-        Snapshot::new_from_versioned_state(&self.id, &self.state())
-    }
-
-    fn restore(&mut self, snapshot: Snapshot) -> std::result::Result<(), MigratableError> {
-        self.set_state(&snapshot.to_versioned_state(&self.id)?);
-        Ok(())
+        Snapshot::new_from_versioned_state(&self.state())
     }
 }
 
@@ -378,6 +382,7 @@ mod tests {
         let mut gpio = Gpio::new(
             String::from(GPIO_NAME),
             Arc::new(TestInterrupt::new(intr_evt.try_clone().unwrap())),
+            None,
         );
         let mut data = [0; 4];
 

devices/src/legacy/rtc_pl031.rs

@@ -61,12 +61,10 @@ pub enum ClockType {
     /// Equivalent to `libc::CLOCK_MONOTONIC`.
     Monotonic,
     /// Equivalent to `libc::CLOCK_REALTIME`.
-    #[allow(dead_code)]
     Real,
     /// Equivalent to `libc::CLOCK_PROCESS_CPUTIME_ID`.
     ProcessCpu,
     /// Equivalent to `libc::CLOCK_THREAD_CPUTIME_ID`.
-    #[allow(dead_code)]
     ThreadCpu,
 }
 
@@ -101,7 +99,7 @@ pub struct LocalTime {
 
 impl LocalTime {
     /// Returns the [LocalTime](struct.LocalTime.html) structure for the calling moment.
-    #[allow(dead_code)]
+    #[cfg(test)]
     pub fn now() -> LocalTime {
         let mut timespec = libc::timespec {
             tv_sec: 0,
@@ -121,7 +119,7 @@ impl LocalTime {
             tm_zone: std::ptr::null(),
         };
 
-        // Safe because the parameters are valid.
+        // SAFETY: the parameters are valid.
         unsafe {
             libc::clock_gettime(libc::CLOCK_REALTIME, &mut timespec);
             libc::localtime_r(&timespec.tv_sec, &mut tm);
@@ -173,22 +171,6 @@ impl Default for TimestampUs {
     }
 }
 
-/// Returns a timestamp in nanoseconds from a monotonic clock.
-///
-/// Uses `_rdstc` on `x86_64` and [`get_time`](fn.get_time.html) on other architectures.
-#[allow(dead_code)]
-pub fn timestamp_cycles() -> u64 {
-    #[cfg(target_arch = "x86_64")]
-    // Safe because there's nothing that can go wrong with this call.
-    unsafe {
-        std::arch::x86_64::_rdtsc() as u64
-    }
-    #[cfg(not(target_arch = "x86_64"))]
-    {
-        get_time(ClockType::Monotonic)
-    }
-}
-
 /// Returns a timestamp in nanoseconds based on the provided clock type.
 ///
 /// # Arguments
@@ -199,7 +181,7 @@ pub fn get_time(clock_type: ClockType) -> u64 {
         tv_sec: 0,
         tv_nsec: 0,
     };
-    // Safe because the parameters are valid.
+    // SAFETY: the parameters are valid.
     unsafe { libc::clock_gettime(clock_type.into(), &mut time_struct) };
     seconds_to_nanoseconds(time_struct.tv_sec).unwrap() as u64 + (time_struct.tv_nsec as u64)
 }
@@ -525,7 +507,6 @@ mod tests {
         ($test_name: ident, $write_fn_name: ident, $read_fn_name: ident, $is_be: expr, $data_type: ty) => {
             #[test]
             fn $test_name() {
-                #[allow(overflowing_literals)]
                 let test_cases = [
                     (
                         0x0123_4567_89AB_CDEF as u64,

devices/src/legacy/serial.rs

@@ -63,7 +63,6 @@ pub struct Serial {
     id: String,
     interrupt_enable: u8,
     interrupt_identification: u8,
-    interrupt: Arc<dyn InterruptSourceGroup>,
     line_control: u8,
     line_status: u8,
     modem_control: u8,
@@ -71,6 +70,7 @@ pub struct Serial {
     scratch: u8,
     baud_divisor: u16,
     in_buffer: VecDeque<u8>,
+    interrupt: Arc<dyn InterruptSourceGroup>,
     out: Option<Box<dyn io::Write + Send>>,
 }
 
@@ -93,19 +93,56 @@ impl Serial {
         id: String,
         interrupt: Arc<dyn InterruptSourceGroup>,
         out: Option<Box<dyn io::Write + Send>>,
+        state: Option<SerialState>,
     ) -> Serial {
+        let (
+            interrupt_enable,
+            interrupt_identification,
+            line_control,
+            line_status,
+            modem_control,
+            modem_status,
+            scratch,
+            baud_divisor,
+            in_buffer,
+        ) = if let Some(state) = state {
+            (
+                state.interrupt_enable,
+                state.interrupt_identification,
+                state.line_control,
+                state.line_status,
+                state.modem_control,
+                state.modem_status,
+                state.scratch,
+                state.baud_divisor,
+                state.in_buffer.into(),
+            )
+        } else {
+            (
+                0,
+                DEFAULT_INTERRUPT_IDENTIFICATION,
+                DEFAULT_LINE_CONTROL,
+                DEFAULT_LINE_STATUS,
+                DEFAULT_MODEM_CONTROL,
+                DEFAULT_MODEM_STATUS,
+                0,
+                DEFAULT_BAUD_DIVISOR,
+                VecDeque::new(),
+            )
+        };
+
         Serial {
             id,
-            interrupt_enable: 0,
-            interrupt_identification: DEFAULT_INTERRUPT_IDENTIFICATION,
+            interrupt_enable,
+            interrupt_identification,
+            line_control,
+            line_status,
+            modem_control,
+            modem_status,
+            scratch,
+            baud_divisor,
+            in_buffer,
             interrupt,
-            line_control: DEFAULT_LINE_CONTROL,
-            line_status: DEFAULT_LINE_STATUS,
-            modem_control: DEFAULT_MODEM_CONTROL,
-            modem_status: DEFAULT_MODEM_STATUS,
-            scratch: 0,
-            baud_divisor: DEFAULT_BAUD_DIVISOR,
-            in_buffer: VecDeque::new(),
             out,
         }
     }
@@ -115,13 +152,18 @@ impl Serial {
         id: String,
         interrupt: Arc<dyn InterruptSourceGroup>,
         out: Box<dyn io::Write + Send>,
+        state: Option<SerialState>,
     ) -> Serial {
-        Self::new(id, interrupt, Some(out))
+        Self::new(id, interrupt, Some(out), state)
     }
 
     /// Constructs a Serial port with no connected output.
-    pub fn new_sink(id: String, interrupt: Arc<dyn InterruptSourceGroup>) -> Serial {
-        Self::new(id, interrupt, None)
+    pub fn new_sink(
+        id: String,
+        interrupt: Arc<dyn InterruptSourceGroup>,
+        state: Option<SerialState>,
+    ) -> Serial {
+        Self::new(id, interrupt, None, state)
     }
 
     pub fn set_out(&mut self, out: Box<dyn io::Write + Send>) {
@@ -242,18 +284,6 @@ impl Serial {
             in_buffer: self.in_buffer.clone().into(),
         }
     }
-
-    fn set_state(&mut self, state: &SerialState) {
-        self.interrupt_enable = state.interrupt_enable;
-        self.interrupt_identification = state.interrupt_identification;
-        self.line_control = state.line_control;
-        self.line_status = state.line_status;
-        self.modem_control = state.modem_control;
-        self.modem_status = state.modem_status;
-        self.scratch = state.scratch;
-        self.baud_divisor = state.baud_divisor;
-        self.in_buffer = state.in_buffer.clone().into();
-    }
 }
 
 impl BusDevice for Serial {
@@ -304,12 +334,7 @@ impl Snapshottable for Serial {
     }
 
     fn snapshot(&mut self) -> std::result::Result<Snapshot, MigratableError> {
-        Snapshot::new_from_versioned_state(&self.id, &self.state())
-    }
-
-    fn restore(&mut self, snapshot: Snapshot) -> std::result::Result<(), MigratableError> {
-        self.set_state(&snapshot.to_versioned_state(&self.id)?);
-        Ok(())
+        Snapshot::new_from_versioned_state(&self.state())
     }
 }
 
@@ -384,6 +409,7 @@ mod tests {
             String::from(SERIAL_NAME),
             Arc::new(TestInterrupt::new(intr_evt.try_clone().unwrap())),
             Box::new(serial_out.clone()),
+            None,
         );
 
         serial.write(0, DATA as u64, &[b'x', b'y']);
@@ -404,6 +430,7 @@ mod tests {
             String::from(SERIAL_NAME),
             Arc::new(TestInterrupt::new(intr_evt.try_clone().unwrap())),
             Box::new(serial_out),
+            None,
         );
 
         // write 1 to the interrupt event fd, so that read doesn't block in case the event fd
@@ -440,6 +467,7 @@ mod tests {
         let mut serial = Serial::new_sink(
             String::from(SERIAL_NAME),
             Arc::new(TestInterrupt::new(intr_evt.try_clone().unwrap())),
+            None,
         );
 
         // write 1 to the interrupt event fd, so that read doesn't block in case the event fd
@@ -462,6 +490,7 @@ mod tests {
         let mut serial = Serial::new_sink(
             String::from(SERIAL_NAME),
             Arc::new(TestInterrupt::new(intr_evt.try_clone().unwrap())),
+            None,
         );
 
         serial.write(0, LCR as u64, &[LCR_DLAB_BIT]);
@@ -483,6 +512,7 @@ mod tests {
         let mut serial = Serial::new_sink(
             String::from(SERIAL_NAME),
             Arc::new(TestInterrupt::new(intr_evt.try_clone().unwrap())),
+            None,
         );
 
         serial.write(0, MCR as u64, &[MCR_LOOP_BIT]);
@@ -509,6 +539,7 @@ mod tests {
         let mut serial = Serial::new_sink(
             String::from(SERIAL_NAME),
             Arc::new(TestInterrupt::new(intr_evt.try_clone().unwrap())),
+            None,
         );
 
         serial.write(0, SCR as u64, &[0x12]);

devices/src/legacy/uart_pl011.rs

@@ -122,24 +122,79 @@ impl Pl011 {
         irq: Arc<dyn InterruptSourceGroup>,
         out: Option<Box<dyn io::Write + Send>>,
         timestamp: Instant,
+        state: Option<Pl011State>,
     ) -> Self {
+        let (
+            flags,
+            lcr,
+            rsr,
+            cr,
+            dmacr,
+            debug,
+            int_enabled,
+            int_level,
+            read_fifo,
+            ilpr,
+            ibrd,
+            fbrd,
+            ifl,
+            read_count,
+            read_trigger,
+        ) = if let Some(state) = state {
+            (
+                state.flags,
+                state.lcr,
+                state.rsr,
+                state.cr,
+                state.dmacr,
+                state.debug,
+                state.int_enabled,
+                state.int_level,
+                state.read_fifo.into(),
+                state.ilpr,
+                state.ibrd,
+                state.fbrd,
+                state.ifl,
+                state.read_count,
+                state.read_trigger,
+            )
+        } else {
+            (
+                0x90,
+                0,
+                0,
+                0x300,
+                0,
+                0,
+                0,
+                0,
+                VecDeque::new(),
+                0,
+                0,
+                0,
+                0x12,
+                0,
+                1,
+            )
+        };
+
         Self {
             id,
-            flags: 0x90u32,
-            lcr: 0u32,
-            rsr: 0u32,
-            cr: 0x300u32,
-            dmacr: 0u32,
-            debug: 0u32,
-            int_enabled: 0u32,
-            int_level: 0u32,
-            read_fifo: VecDeque::new(),
-            ilpr: 0u32,
-            ibrd: 0u32,
-            fbrd: 0u32,
-            ifl: 0x12u32,
-            read_count: 0u32,
-            read_trigger: 1u32,
+            flags,
+            lcr,
+            rsr,
+            cr,
+            dmacr,
+            debug,
+            int_enabled,
+            int_level,
+            read_fifo,
+            ilpr,
+            ibrd,
+            fbrd,
+            ifl,
+            read_count,
+            read_trigger,
             irq,
             out,
             timestamp,
@@ -170,24 +225,6 @@ impl Pl011 {
         }
     }
 
-    fn set_state(&mut self, state: &Pl011State) {
-        self.flags = state.flags;
-        self.lcr = state.lcr;
-        self.rsr = state.rsr;
-        self.cr = state.cr;
-        self.dmacr = state.dmacr;
-        self.debug = state.debug;
-        self.int_enabled = state.int_enabled;
-        self.int_level = state.int_level;
-        self.read_fifo = state.read_fifo.clone().into();
-        self.ilpr = state.ilpr;
-        self.ibrd = state.ibrd;
-        self.fbrd = state.fbrd;
-        self.ifl = state.ifl;
-        self.read_count = state.read_count;
-        self.read_trigger = state.read_trigger;
-    }
-
     /// Queues raw bytes for the guest to read and signals the interrupt
     pub fn queue_input_bytes(&mut self, c: &[u8]) -> vmm_sys_util::errno::Result<()> {
         self.read_fifo.extend(c);
@@ -417,12 +454,7 @@ impl Snapshottable for Pl011 {
     }
 
     fn snapshot(&mut self) -> std::result::Result<Snapshot, MigratableError> {
-        Snapshot::new_from_versioned_state(&self.id, &self.state())
-    }
-
-    fn restore(&mut self, snapshot: Snapshot) -> std::result::Result<(), MigratableError> {
-        self.set_state(&snapshot.to_versioned_state(&self.id)?);
-        Ok(())
+        Snapshot::new_from_versioned_state(&self.state())
     }
 }
 
@@ -498,12 +530,13 @@ mod tests {
             Arc::new(TestInterrupt::new(intr_evt.try_clone().unwrap())),
             Some(Box::new(pl011_out.clone())),
             Instant::now(),
+            None,
         );
 
-        pl011.write(0, UARTDR as u64, &[b'x', b'y']);
-        pl011.write(0, UARTDR as u64, &[b'a']);
-        pl011.write(0, UARTDR as u64, &[b'b']);
-        pl011.write(0, UARTDR as u64, &[b'c']);
+        pl011.write(0, UARTDR, &[b'x', b'y']);
+        pl011.write(0, UARTDR, &[b'a']);
+        pl011.write(0, UARTDR, &[b'b']);
+        pl011.write(0, UARTDR, &[b'c']);
         assert_eq!(
             pl011_out.buf.lock().unwrap().as_slice(),
             &[b'x', b'a', b'b', b'c']
@@ -519,6 +552,7 @@ mod tests {
             Arc::new(TestInterrupt::new(intr_evt.try_clone().unwrap())),
             Some(Box::new(pl011_out)),
             Instant::now(),
+            None,
         );
 
         // write 1 to the interrupt event fd, so that read doesn't block in case the event fd
@@ -529,11 +563,11 @@ mod tests {
         assert_eq!(intr_evt.read().unwrap(), 2);
 
         let mut data = [0u8];
-        pl011.read(0, UARTDR as u64, &mut data);
+        pl011.read(0, UARTDR, &mut data);
         assert_eq!(data[0], b'a');
-        pl011.read(0, UARTDR as u64, &mut data);
+        pl011.read(0, UARTDR, &mut data);
         assert_eq!(data[0], b'b');
-        pl011.read(0, UARTDR as u64, &mut data);
+        pl011.read(0, UARTDR, &mut data);
         assert_eq!(data[0], b'c');
     }
 }

devices/src/lib.rs

@@ -33,11 +33,9 @@ bitflags! {
     }
 }
 
-#[allow(unused_macros)]
 #[cfg(target_arch = "aarch64")]
 macro_rules! generate_read_fn {
     ($fn_name: ident, $data_type: ty, $byte_type: ty, $type_size: expr, $endian_type: ident) => {
-        #[allow(dead_code)]
         pub fn $fn_name(input: &[$byte_type]) -> $data_type {
             assert!($type_size == std::mem::size_of::<$data_type>());
             let mut array = [0u8; $type_size];
@@ -49,11 +47,9 @@ macro_rules! generate_read_fn {
     };
 }
 
-#[allow(unused_macros)]
 #[cfg(target_arch = "aarch64")]
 macro_rules! generate_write_fn {
     ($fn_name: ident, $data_type: ty, $byte_type: ty, $endian_type: ident) => {
-        #[allow(dead_code)]
         pub fn $fn_name(buf: &mut [$byte_type], n: $data_type) {
             for (byte, read) in buf
                 .iter_mut()

devices/src/tpm.rs

@@ -8,13 +8,11 @@ use anyhow::anyhow;
 use arch::aarch64::layout::{TPM_SIZE, TPM_START};
 #[cfg(target_arch = "x86_64")]
 use arch::x86_64::layout::{TPM_SIZE, TPM_START};
-use phf::phf_map;
 use std::cmp;
 use std::sync::{Arc, Barrier};
 use thiserror::Error;
 use tpm::emulator::{BackendCmd, Emulator};
 use tpm::TPM_CRB_BUFFER_MAX;
-use tpm::TPM_SUCCESS;
 use vm_device::BusDevice;
 
 #[derive(Error, Debug)]
@@ -23,62 +21,135 @@ pub enum Error {
     CheckCaps(#[source] anyhow::Error),
     #[error("Failed to initialize tpm: {0}")]
     Init(#[source] anyhow::Error),
-    #[error("Failed to deliver tpm Command: {0}")]
-    DeliverRequest(#[source] anyhow::Error),
 }
 type Result<T> = anyhow::Result<T, Error>;
 
+#[allow(dead_code)]
+enum LocStateFields {
+    TpmEstablished,
+    LocAssigned,
+    ActiveLocality,
+    Reserved,
+    TpmRegValidSts,
+}
+
+enum LocStsFields {
+    Granted,
+    BeenSeized,
+}
+
+#[allow(dead_code)]
+enum IntfIdFields {
+    InterfaceType,
+    InterfaceVersion,
+    CapLocality,
+    CapCRBIdleBypass,
+    Reserved1,
+    CapDataXferSizeSupport,
+    CapFIFO,
+    CapCRB,
+    CapIFRes,
+    InterfaceSelector,
+    IntfSelLock,
+    Reserved2,
+    Rid,
+}
+
+#[allow(dead_code)]
+enum IntfId2Fields {
+    Vid,
+    Did,
+}
+
+enum CtrlStsFields {
+    TpmSts,
+    TpmIdle,
+}
+
+enum CrbRegister {
+    LocState(LocStateFields),
+    LocSts(LocStsFields),
+    IntfId(IntfIdFields),
+    IntfId2(IntfId2Fields),
+    CtrlSts(CtrlStsFields),
+}
+
 /* crb 32-bit registers */
 const CRB_LOC_STATE: u32 = 0x0;
 //Register Fields
-// Field => (start, length)
-// start: lowest bit in the bit field numbered from 0
+// Field => (base, offset, length)
+// base:   starting position of the register
+// offset: lowest bit in the bit field numbered from 0
 // length: length of the bit field
-const CRB_LOC_STATE_FIELDS: phf::Map<&str, [u32; 2]> = phf_map! {
-    "tpmEstablished" => [0, 1],
-    "locAssigned" => [1,1],
-    "activeLocality"=> [2, 3],
-    "reserved" => [5, 2],
-    "tpmRegValidSts" => [7, 1]
-};
+const fn get_crb_loc_state_field(f: LocStateFields) -> (u32, u32, u32) {
+    let (offset, len) = match f {
+        LocStateFields::TpmEstablished => (0, 1),
+        LocStateFields::LocAssigned => (1, 1),
+        LocStateFields::ActiveLocality => (2, 3),
+        LocStateFields::Reserved => (5, 2),
+        LocStateFields::TpmRegValidSts => (7, 1),
+    };
+
+    (CRB_LOC_STATE, offset, len)
+}
+
 const CRB_LOC_CTRL: u32 = 0x08;
 const CRB_LOC_CTRL_REQUEST_ACCESS: u32 = 1 << 0;
 const CRB_LOC_CTRL_RELINQUISH: u32 = 1 << 1;
 const CRB_LOC_CTRL_RESET_ESTABLISHMENT_BIT: u32 = 1 << 3;
 const CRB_LOC_STS: u32 = 0x0C;
-const CRB_LOC_STS_FIELDS: phf::Map<&str, [u32; 2]> = phf_map! {
-    "Granted" => [0, 1],
-    "beenSeized" => [1,1]
-};
+const fn get_crb_loc_sts_field(f: LocStsFields) -> (u32, u32, u32) {
+    let (offset, len) = match f {
+        LocStsFields::Granted => (0, 1),
+        LocStsFields::BeenSeized => (1, 1),
+    };
+
+    (CRB_LOC_STS, offset, len)
+}
+
 const CRB_INTF_ID: u32 = 0x30;
-const CRB_INTF_ID_FIELDS: phf::Map<&str, [u32; 2]> = phf_map! {
-    "InterfaceType" => [0, 4],
-    "InterfaceVersion" => [4, 4],
-    "CapLocality" =>  [8, 1],
-    "CapCRBIdleBypass" => [9, 1],
-    "Reserved1" => [10, 1],
-    "CapDataXferSizeSupport" => [11, 2],
-    "CapFIFO" =>  [13, 1],
-    "CapCRB" => [14, 1],
-    "CapIFRes" => [15, 2],
-    "InterfaceSelector" => [17, 2],
-    "IntfSelLock" =>  [19, 1],
-    "Reserved2" => [20, 4],
-    "RID" => [24, 8]
-};
+const fn get_crb_intf_id_field(f: IntfIdFields) -> (u32, u32, u32) {
+    let (offset, len) = match f {
+        IntfIdFields::InterfaceType => (0, 4),
+        IntfIdFields::InterfaceVersion => (4, 4),
+        IntfIdFields::CapLocality => (8, 1),
+        IntfIdFields::CapCRBIdleBypass => (9, 1),
+        IntfIdFields::Reserved1 => (10, 1),
+        IntfIdFields::CapDataXferSizeSupport => (11, 2),
+        IntfIdFields::CapFIFO => (13, 1),
+        IntfIdFields::CapCRB => (14, 1),
+        IntfIdFields::CapIFRes => (15, 2),
+        IntfIdFields::InterfaceSelector => (17, 2),
+        IntfIdFields::IntfSelLock => (19, 1),
+        IntfIdFields::Reserved2 => (20, 4),
+        IntfIdFields::Rid => (24, 8),
+    };
+
+    (CRB_INTF_ID, offset, len)
+}
+
 const CRB_INTF_ID2: u32 = 0x34;
-const CRB_INTF_ID2_FIELDS: phf::Map<&str, [u32; 2]> = phf_map! {
-    "VID" => [0, 16],
-    "DID" => [16, 16]
-};
+const fn get_crb_intf_id2_field(f: IntfId2Fields) -> (u32, u32, u32) {
+    let (offset, len) = match f {
+        IntfId2Fields::Vid => (0, 16),
+        IntfId2Fields::Did => (16, 16),
+    };
+
+    (CRB_INTF_ID2, offset, len)
+}
+
 const CRB_CTRL_REQ: u32 = 0x40;
 const CRB_CTRL_REQ_CMD_READY: u32 = 1 << 0;
 const CRB_CTRL_REQ_GO_IDLE: u32 = 1 << 1;
 const CRB_CTRL_STS: u32 = 0x44;
-const CRB_CTRL_STS_FIELDS: phf::Map<&str, [u32; 2]> = phf_map! {
-    "tpmSts" => [0, 1],
-    "tpmIdle" => [1, 1]
-};
+const fn get_crb_ctrl_sts_field(f: CtrlStsFields) -> (u32, u32, u32) {
+    let (offset, len) = match f {
+        CtrlStsFields::TpmSts => (0, 1),
+        CtrlStsFields::TpmIdle => (1, 1),
+    };
+
+    (CRB_CTRL_STS, offset, len)
+}
 const CRB_CTRL_CANCEL: u32 = 0x48;
 const CRB_CANCEL_INVOKE: u32 = 1 << 0;
 const CRB_CTRL_START: u32 = 0x4C;
@@ -94,7 +165,7 @@ const TPM_CRB_NO_LOCALITY: u32 = 0xff;
 const TPM_CRB_ADDR_BASE: u32 = TPM_START.0 as u32;
 const TPM_CRB_ADDR_SIZE: usize = TPM_SIZE as usize;
 
-const TPM_CRB_R_MAX: u32 = CRB_DATA_BUFFER;
+const TPM_CRB_R_MAX: usize = CRB_DATA_BUFFER as usize;
 
 // CRB Protocol details
 const CRB_INTF_TYPE_CRB_ACTIVE: u32 = 0b1;
@@ -109,47 +180,29 @@ const PCI_VENDOR_ID_IBM: u32 = 0x1014;
 const CRB_CTRL_CMD_SIZE_REG: u32 = 0x58;
 const CRB_CTRL_CMD_SIZE: usize = TPM_CRB_ADDR_SIZE - CRB_DATA_BUFFER as usize;
 
-fn get_fields_map(reg: u32) -> phf::Map<&'static str, [u32; 2]> {
+// Returns (register base, offset, len)
+const fn get_field(reg: CrbRegister) -> (u32, u32, u32) {
     match reg {
-        CRB_LOC_STATE => CRB_LOC_STATE_FIELDS,
-        CRB_LOC_STS => CRB_LOC_STS_FIELDS,
-        CRB_INTF_ID => CRB_INTF_ID_FIELDS,
-        CRB_INTF_ID2 => CRB_INTF_ID2_FIELDS,
-        CRB_CTRL_STS => CRB_CTRL_STS_FIELDS,
-        _ => {
-            panic!("Fields in '{reg:?}' register were accessed which are Invalid");
-        }
+        CrbRegister::LocState(f) => get_crb_loc_state_field(f),
+        CrbRegister::LocSts(f) => get_crb_loc_sts_field(f),
+        CrbRegister::IntfId(f) => get_crb_intf_id_field(f),
+        CrbRegister::IntfId2(f) => get_crb_intf_id2_field(f),
+        CrbRegister::CtrlSts(f) => get_crb_ctrl_sts_field(f),
     }
 }
 
-/// Set a particular field in a Register
-fn set_reg_field(regs: &mut [u32; TPM_CRB_R_MAX as usize], reg: u32, field: &str, value: u32) {
-    let reg_fields = get_fields_map(reg);
-    if reg_fields.contains_key(field) {
-        let start = reg_fields.get(field).unwrap()[0];
-        let len = reg_fields.get(field).unwrap()[1];
-        let mask = (!(0_u32) >> (32 - len)) << start;
-        regs[reg as usize] = (regs[reg as usize] & !mask) | ((value << start) & mask);
-    } else {
-        error!(
-            "Failed to tpm Register. {:?} is not a valid field in Reg {:#X}",
-            field, reg
-        )
-    }
+// Set a particular field in a Register
+fn set_reg_field(regs: &mut [u32; TPM_CRB_R_MAX], reg: CrbRegister, value: u32) {
+    let (base, offset, len) = get_field(reg);
+    let mask = (!(0_u32) >> (32 - len)) << offset;
+    regs[base as usize] = (regs[base as usize] & !mask) | ((value << offset) & mask);
 }
 
-/// Get the value of a particular field in a Register
-fn get_reg_field(regs: &[u32; TPM_CRB_R_MAX as usize], reg: u32, field: &str) -> u32 {
-    let reg_fields = get_fields_map(reg);
-    if reg_fields.contains_key(field) {
-        let start = reg_fields.get(field).unwrap()[0];
-        let len = reg_fields.get(field).unwrap()[1];
-        let mask = (!(0_u32) >> (32 - len)) << start;
-        (regs[reg as usize] & mask) >> start
-    } else {
-        // TODO: Sensible return value if fields do not exist
-        0x0
-    }
+// Get the value of a particular field in a Register
+const fn get_reg_field(regs: &[u32; TPM_CRB_R_MAX], reg: CrbRegister) -> u32 {
+    let (base, offset, len) = get_field(reg);
+    let mask = (!(0_u32) >> (32 - len)) << offset;
+    (regs[base as usize] & mask) >> offset
 }
 
 fn locality_from_addr(addr: u32) -> u8 {
@@ -158,8 +211,7 @@ fn locality_from_addr(addr: u32) -> u8 {
 
 pub struct Tpm {
     emulator: Emulator,
-    cmd: Option<BackendCmd>,
-    regs: [u32; TPM_CRB_R_MAX as usize],
+    regs: [u32; TPM_CRB_R_MAX],
     backend_buff_size: usize,
     data_buff: [u8; TPM_CRB_BUFFER_MAX],
     data_buff_len: usize,
@@ -171,8 +223,7 @@ impl Tpm {
             .map_err(|e| Error::Init(anyhow!("Failed while initializing tpm Emulator: {:?}", e)))?;
         let mut tpm = Tpm {
             emulator,
-            cmd: None,
-            regs: [0; TPM_CRB_R_MAX as usize],
+            regs: [0; TPM_CRB_R_MAX],
             backend_buff_size: TPM_CRB_BUFFER_MAX,
             data_buff: [0; TPM_CRB_BUFFER_MAX],
             data_buff_len: 0,
@@ -182,74 +233,93 @@ impl Tpm {
     }
 
     fn get_active_locality(&mut self) -> u32 {
-        if get_reg_field(&self.regs, CRB_LOC_STATE, "locAssigned") == 0 {
+        if get_reg_field(
+            &self.regs,
+            CrbRegister::LocState(LocStateFields::LocAssigned),
+        ) == 0
+        {
             return TPM_CRB_NO_LOCALITY;
         }
-        get_reg_field(&self.regs, CRB_LOC_STATE, "activeLocality")
+        get_reg_field(
+            &self.regs,
+            CrbRegister::LocState(LocStateFields::ActiveLocality),
+        )
     }
 
-    fn request_completed(&mut self, result: isize) {
+    fn request_completed(&mut self, success: bool) {
         self.regs[CRB_CTRL_START as usize] = !CRB_START_INVOKE;
-        if result != 0 {
-            set_reg_field(&mut self.regs, CRB_CTRL_STS, "tpmSts", 1);
+        if !success {
+            set_reg_field(
+                &mut self.regs,
+                CrbRegister::CtrlSts(CtrlStsFields::TpmSts),
+                1,
+            );
         }
     }
 
     fn reset(&mut self) -> Result<()> {
-        let cur_buff_size = self.emulator.get_buffer_size().unwrap();
-        self.regs = [0; TPM_CRB_R_MAX as usize];
-        set_reg_field(&mut self.regs, CRB_LOC_STATE, "tpmRegValidSts", 1);
-        set_reg_field(&mut self.regs, CRB_CTRL_STS, "tpmIdle", 1);
+        let cur_buff_size = self.emulator.get_buffer_size();
+        self.regs = [0; TPM_CRB_R_MAX];
         set_reg_field(
             &mut self.regs,
-            CRB_INTF_ID,
-            "InterfaceType",
+            CrbRegister::LocState(LocStateFields::TpmRegValidSts),
+            1,
+        );
+        set_reg_field(
+            &mut self.regs,
+            CrbRegister::CtrlSts(CtrlStsFields::TpmIdle),
+            1,
+        );
+        set_reg_field(
+            &mut self.regs,
+            CrbRegister::IntfId(IntfIdFields::InterfaceType),
             CRB_INTF_TYPE_CRB_ACTIVE,
         );
         set_reg_field(
             &mut self.regs,
-            CRB_INTF_ID,
-            "InterfaceVersion",
+            CrbRegister::IntfId(IntfIdFields::InterfaceVersion),
             CRB_INTF_VERSION_CRB,
         );
         set_reg_field(
             &mut self.regs,
-            CRB_INTF_ID,
-            "CapLocality",
+            CrbRegister::IntfId(IntfIdFields::CapLocality),
             CRB_INTF_CAP_LOCALITY_0_ONLY,
         );
         set_reg_field(
             &mut self.regs,
-            CRB_INTF_ID,
-            "CapCRBIdleBypass",
+            CrbRegister::IntfId(IntfIdFields::CapCRBIdleBypass),
             CRB_INTF_CAP_IDLE_FAST,
         );
         set_reg_field(
             &mut self.regs,
-            CRB_INTF_ID,
-            "CapDataXferSizeSupport",
+            CrbRegister::IntfId(IntfIdFields::CapDataXferSizeSupport),
             CRB_INTF_CAP_XFER_SIZE_64,
         );
         set_reg_field(
             &mut self.regs,
-            CRB_INTF_ID,
-            "CapFIFO",
+            CrbRegister::IntfId(IntfIdFields::CapFIFO),
             CRB_INTF_CAP_FIFO_NOT_SUPPORTED,
         );
         set_reg_field(
             &mut self.regs,
-            CRB_INTF_ID,
-            "CapCRB",
+            CrbRegister::IntfId(IntfIdFields::CapCRB),
             CRB_INTF_CAP_CRB_SUPPORTED,
         );
         set_reg_field(
             &mut self.regs,
-            CRB_INTF_ID,
-            "InterfaceSelector",
+            CrbRegister::IntfId(IntfIdFields::InterfaceSelector),
             CRB_INTF_IF_SELECTOR_CRB,
         );
-        set_reg_field(&mut self.regs, CRB_INTF_ID, "RID", 0b0000);
-        set_reg_field(&mut self.regs, CRB_INTF_ID2, "VID", PCI_VENDOR_ID_IBM);
+        set_reg_field(
+            &mut self.regs,
+            CrbRegister::IntfId(IntfIdFields::Rid),
+            0b0000,
+        );
+        set_reg_field(
+            &mut self.regs,
+            CrbRegister::IntfId2(IntfId2Fields::Vid),
+            PCI_VENDOR_ID_IBM,
+        );
 
         self.regs[CRB_CTRL_CMD_SIZE_REG as usize] = CRB_CTRL_CMD_SIZE as u32;
         self.regs[CRB_CTRL_CMD_LADDR as usize] = TPM_CRB_ADDR_BASE + CRB_DATA_BUFFER;
@@ -268,7 +338,6 @@ impl Tpm {
     }
 }
 
-//impl BusDevice for TPM
 impl BusDevice for Tpm {
     fn read(&mut self, _base: u64, offset: u64, data: &mut [u8]) {
         let mut offset: u32 = offset as u32;
@@ -367,10 +436,18 @@ impl BusDevice for Tpm {
                 }
                 CRB_CTRL_REQ => match v {
                     CRB_CTRL_REQ_CMD_READY => {
-                        set_reg_field(&mut self.regs, CRB_CTRL_STS, "tpmIdle", 0);
+                        set_reg_field(
+                            &mut self.regs,
+                            CrbRegister::CtrlSts(CtrlStsFields::TpmIdle),
+                            0,
+                        );
                     }
                     CRB_CTRL_REQ_GO_IDLE => {
-                        set_reg_field(&mut self.regs, CRB_CTRL_STS, "tpmIdle", 1);
+                        set_reg_field(
+                            &mut self.regs,
+                            CrbRegister::CtrlSts(CtrlStsFields::TpmIdle),
+                            1,
+                        );
                     }
                     _ => {
                         error!("Invalid value passed to CRTL_REQ register");
@@ -393,27 +470,14 @@ impl BusDevice for Tpm {
                     {
                         self.regs[CRB_CTRL_START as usize] |= CRB_START_INVOKE;
 
-                        self.cmd = Some(BackendCmd {
-                            locality: locality as u8,
-                            input: self.data_buff[0..self.data_buff_len].to_vec(),
+                        let mut cmd = BackendCmd {
+                            buffer: &mut self.data_buff,
                             input_len: cmp::min(self.data_buff_len, TPM_CRB_BUFFER_MAX),
-                            output: self.data_buff.to_vec(),
-                            output_len: TPM_CRB_BUFFER_MAX,
-                            selftest_done: false,
-                        });
+                        };
 
-                        let mut cmd = self.cmd.as_ref().unwrap().clone();
-                        let output = self.emulator.deliver_request(&mut cmd).map_err(|e| {
-                            Error::DeliverRequest(anyhow!(
-                                "Failed to deliver tpm request. Error :{:?}",
-                                e
-                            ))
-                        });
-                        //TODO: drop the copy here
-                        self.data_buff.fill(0);
-                        self.data_buff.clone_from_slice(output.unwrap().as_slice());
+                        let status = self.emulator.deliver_request(&mut cmd).is_ok();
 
-                        self.request_completed(TPM_SUCCESS as isize);
+                        self.request_completed(status);
                     }
                 }
                 CRB_LOC_CTRL => {
@@ -424,13 +488,33 @@ impl BusDevice for Tpm {
                     match v {
                         CRB_LOC_CTRL_RESET_ESTABLISHMENT_BIT => {}
                         CRB_LOC_CTRL_RELINQUISH => {
-                            set_reg_field(&mut self.regs, CRB_LOC_STATE, "locAssigned", 0);
-                            set_reg_field(&mut self.regs, CRB_LOC_STS, "Granted", 0);
+                            set_reg_field(
+                                &mut self.regs,
+                                CrbRegister::LocState(LocStateFields::LocAssigned),
+                                0,
+                            );
+                            set_reg_field(
+                                &mut self.regs,
+                                CrbRegister::LocSts(LocStsFields::Granted),
+                                0,
+                            );
                         }
                         CRB_LOC_CTRL_REQUEST_ACCESS => {
-                            set_reg_field(&mut self.regs, CRB_LOC_STS, "Granted", 1);
-                            set_reg_field(&mut self.regs, CRB_LOC_STS, "beenSeized", 0);
-                            set_reg_field(&mut self.regs, CRB_LOC_STATE, "locAssigned", 1);
+                            set_reg_field(
+                                &mut self.regs,
+                                CrbRegister::LocSts(LocStsFields::Granted),
+                                1,
+                            );
+                            set_reg_field(
+                                &mut self.regs,
+                                CrbRegister::LocSts(LocStsFields::BeenSeized),
+                                0,
+                            );
+                            set_reg_field(
+                                &mut self.regs,
+                                CrbRegister::LocState(LocStateFields::LocAssigned),
+                                1,
+                            );
                         }
                         _ => {
                             error!("Invalid value to write in CRB_LOC_CTRL {:#X} ", v);
@@ -456,10 +540,10 @@ mod tests {
 
     #[test]
     fn test_set_get_reg_field() {
-        let mut regs: [u32; TPM_CRB_R_MAX as usize] = [0; TPM_CRB_R_MAX as usize];
-        set_reg_field(&mut regs, CRB_INTF_ID, "RID", 0xAC);
+        let mut regs: [u32; TPM_CRB_R_MAX] = [0; TPM_CRB_R_MAX];
+        set_reg_field(&mut regs, CrbRegister::IntfId(IntfIdFields::Rid), 0xAC);
         assert_eq!(
-            get_reg_field(&regs, CRB_INTF_ID, "RID"),
+            get_reg_field(&regs, CrbRegister::IntfId(IntfIdFields::Rid)),
             0xAC,
             concat!("Test: ", stringify!(set_get_reg_field))
         );

docs/api.md

@@ -78,33 +78,40 @@ The Cloud Hypervisor API exposes the following actions through its endpoints:
 
 #### Virtual Machine (VM) Actions
 
-| Action                             | Endpoint              | Request Body                | Response Body            | Prerequisites                    |
-| ---------------------------------- | --------------------- | --------------------------- | ------------------------ | -------------------------------- |
-| Create the VM                      | `/vm.create`          | `/schemas/VmConfig`         | N/A                      | The VM is not created yet        |
-| Delete the VM                      | `/vm.delete`          | N/A                         | N/A                      | N/A                              |
-| Boot the VM                        | `/vm.boot`            | N/A                         | N/A                      | The VM is created but not booted |
-| Shut the VM down                   | `/vm.shutdown`        | N/A                         | N/A                      | The VM is booted                 |
-| Reboot the VM                      | `/vm.reboot`          | N/A                         | N/A                      | The VM is booted                 |
-| Trigger power button of the VM     | `/vm.power-button`    | N/A                         | N/A                      | The VM is booted                 |
-| Pause the VM                       | `/vm.pause`           | N/A                         | N/A                      | The VM is booted                 |
-| Resume the VM                      | `/vm.resume`          | N/A                         | N/A                      | The VM is paused                 |
-| Task a snapshot of the VM          | `/vm.snapshot`        | `/schemas/VmSnapshotConfig` | N/A                      | The VM is paused                 |
-| Perform a coredump of the VM       | `/vm.coredump`        | `/schemas/VmCoredumpData`   | N/A                      | The VM is paused                 |
-| Restore the VM from a snapshot     | `/vm.restore`         | `/schemas/RestoreConfig`    | N/A                      | The VM is created but not booted |
-| Add/remove CPUs to/from the VM     | `/vm.resize`          | `/schemas/VmResize`         | N/A                      | The VM is booted                 |
-| Add/remove memory from the VM      | `/vm.resize`          | `/schemas/VmResize`         | N/A                      | The VM is booted                 |
-| Add/remove memory from a zone      | `/vm.resize-zone`     | `/schemas/VmResizeZone`     | N/A                      | The VM is booted                 |
-| Dump the VM information            | `/vm.info`            | N/A                         | `/schemas/VmInfo`        | The VM is created                |
-| Add VFIO PCI device to the VM      | `/vm.add-device`      | `/schemas/VmAddDevice`      | `/schemas/PciDeviceInfo` | The VM is booted                 |
-| Add disk device to the VM          | `/vm.add-disk`        | `/schemas/DiskConfig`       | `/schemas/PciDeviceInfo` | The VM is booted                 |
-| Add fs device to the VM            | `/vm.add-fs`          | `/schemas/FsConfig`         | `/schemas/PciDeviceInfo` | The VM is booted                 |
-| Add pmem device to the VM          | `/vm.add-pmem`        | `/schemas/PmemConfig`       | `/schemas/PciDeviceInfo` | The VM is booted                 |
-| Add network device to the VM       | `/vm.add-net`         | `/schemas/NetConfig`        | `/schemas/PciDeviceInfo` | The VM is booted                 |
-| Add userspace PCI device to the VM | `/vm.add-user-device` | `/schemas/VmAddUserDevice`  | `/schemas/PciDeviceInfo` | The VM is booted                 |
-| Add vdpa device to the VM          | `/vm.add-vdpa`        | `/schemas/VdpaConfig`       | `/schemas/PciDeviceInfo` | The VM is booted                 |
-| Add vsock device to the VM         | `/vm.add-vsock`       | `/schemas/VsockConfig`      | `/schemas/PciDeviceInfo` | The VM is booted                 |
-| Remove device from the VM          | `/vm.remove-device`   | `/schemas/VmRemoveDevice`   | N/A                      | The VM is booted                 |
-| Dump the VM counters               | `/vm.counters`        | N/A                         | `/schemas/VmCounters`    | The VM is booted                 |
+| Action                             | Endpoint                | Request Body                    | Response Body            | Prerequisites                                          |
+| ---------------------------------- | ----------------------- | ------------------------------- | ------------------------ | ------------------------------------------------------ |
+| Create the VM                      | `/vm.create`            | `/schemas/VmConfig`             | N/A                      | The VM is not created yet                              |
+| Delete the VM                      | `/vm.delete`            | N/A                             | N/A                      | N/A                                                    |
+| Boot the VM                        | `/vm.boot`              | N/A                             | N/A                      | The VM is created but not booted                       |
+| Shut the VM down                   | `/vm.shutdown`          | N/A                             | N/A                      | The VM is booted                                       |
+| Reboot the VM                      | `/vm.reboot`            | N/A                             | N/A                      | The VM is booted                                       |
+| Trigger power button of the VM     | `/vm.power-button`      | N/A                             | N/A                      | The VM is booted                                       |
+| Pause the VM                       | `/vm.pause`             | N/A                             | N/A                      | The VM is booted                                       |
+| Resume the VM                      | `/vm.resume`            | N/A                             | N/A                      | The VM is paused                                       |
+| Task a snapshot of the VM          | `/vm.snapshot`          | `/schemas/VmSnapshotConfig`     | N/A                      | The VM is paused                                       |
+| Perform a coredump of the VM*      | `/vm.coredump`          | `/schemas/VmCoredumpData`       | N/A                      | The VM is paused                                       |
+| Restore the VM from a snapshot     | `/vm.restore`           | `/schemas/RestoreConfig`        | N/A                      | The VM is created but not booted                       |
+| Add/remove CPUs to/from the VM     | `/vm.resize`            | `/schemas/VmResize`             | N/A                      | The VM is booted                                       |
+| Add/remove memory from the VM      | `/vm.resize`            | `/schemas/VmResize`             | N/A                      | The VM is booted                                       |
+| Add/remove memory from a zone      | `/vm.resize-zone`       | `/schemas/VmResizeZone`         | N/A                      | The VM is booted                                       |
+| Dump the VM information            | `/vm.info`              | N/A                             | `/schemas/VmInfo`        | The VM is created                                      |
+| Add VFIO PCI device to the VM      | `/vm.add-device`        | `/schemas/VmAddDevice`          | `/schemas/PciDeviceInfo` | The VM is booted                                       |
+| Add disk device to the VM          | `/vm.add-disk`          | `/schemas/DiskConfig`           | `/schemas/PciDeviceInfo` | The VM is booted                                       |
+| Add fs device to the VM            | `/vm.add-fs`            | `/schemas/FsConfig`             | `/schemas/PciDeviceInfo` | The VM is booted                                       |
+| Add pmem device to the VM          | `/vm.add-pmem`          | `/schemas/PmemConfig`           | `/schemas/PciDeviceInfo` | The VM is booted                                       |
+| Add network device to the VM       | `/vm.add-net`           | `/schemas/NetConfig`            | `/schemas/PciDeviceInfo` | The VM is booted                                       |
+| Add userspace PCI device to the VM | `/vm.add-user-device`   | `/schemas/VmAddUserDevice`      | `/schemas/PciDeviceInfo` | The VM is booted                                       |
+| Add vdpa device to the VM          | `/vm.add-vdpa`          | `/schemas/VdpaConfig`           | `/schemas/PciDeviceInfo` | The VM is booted                                       |
+| Add vsock device to the VM         | `/vm.add-vsock`         | `/schemas/VsockConfig`          | `/schemas/PciDeviceInfo` | The VM is booted                                       |
+| Remove device from the VM          | `/vm.remove-device`     | `/schemas/VmRemoveDevice`       | N/A                      | The VM is booted                                       |
+| Dump the VM counters               | `/vm.counters`          | N/A                             | `/schemas/VmCounters`    | The VM is booted                                       |
+| Prepare to receive a migration     | `/vm.receive-migration` | `/schemas/ReceiveMigrationData` | N/A                      | N/A                                                    |
+| Start to send migration to target  | `/vm.send-migration`    | `/schemas/SendMigrationData`    | N/A                      | The VM is booted and (shared mem or hugepages enabled) |
+
+* The `vmcoredump` action is available exclusively for the `x86_64`
+architecture and can be executed only when the `guest_debug` feature is
+enabled. Without this feature, the corresponding REST API endpoint is not
+available.
 
 ### REST API Examples
 
@@ -217,11 +224,11 @@ From the CLI, one can either:
 The REST API and the CLI both rely on a common, [internal API](#internal-api).
 
 The CLI options are parsed by the
-[clap crate](https://docs.rs/clap/2.33.0/clap/) and then translated into
+[argh crate](https://docs.rs/argh/latest/argh/) and then translated into
 [internal API](#internal-api) commands.
 
 The REST API is processed by an HTTP thread using the
-[Firecracker's `micro_http`](https://github.com/firecracker-microvm/firecracker/tree/master/src/micro_http)
+[Firecracker's `micro_http`](https://github.com/firecracker-microvm/micro-http)
 crate. As with the CLI, the HTTP requests eventually get translated into
 [internal API](#internal-api) commands.
 
@@ -245,7 +252,7 @@ As a summary, the REST API and the CLI are essentially frontends for the
                        |                                      |      +------------------------+
                        |            +----------+              |                 VMM
                        |     CLI    |          |              |
-                       +----------->+   clap   +--------------+
+                       +----------->+   argh   +--------------+
                                     |          |
                                     +----------+
 
@@ -379,5 +386,5 @@ APIs work together, let's look at a complete VM creation flow, from the
    ```
 1. The Cloud Hypervisor HTTP thread sends the formed HTTP response back to the
    user. This is abstracted by the
-   [micro_http](https://github.com/firecracker-microvm/firecracker/tree/master/src/micro_http)
+   [micro_http](https://github.com/firecracker-microvm/micro-http)
    crate.

docs/arm64.md

@@ -1,152 +0,0 @@
-# How to build and test Cloud Hypervisor on AArch64
-
-This document introduces how to build and test Cloud Hypervisor on AArch64.
-Currently, Cloud Hypervisor supports 2 methods of booting on AArch64: UEFI
-booting and direct-kernel booting. The document covers both methods.
-
-All the steps are based on Ubuntu. We use the Ubuntu cloud image for guest VM
-disk.
-
-## Hardware requirements
-
-- AArch64 servers (recommended) or development boards equipped with the GICv3
-interrupt controller.
-
-- On development boards that have constrained RAM resources, if the creation of
-a VM consumes a large portion of the free memory on the host, it may be required
-to enable swap. For example, this was required on a board with 3 GB of RAM
-booting a 2 GB VM at a point in time when 2.8 GB were free. Without enabling
-swap the `cloud-hypervisor` process was terminated by the OOM killer. In this
-situation memory was allocated for the virtual machine using memfd while the
-page cache was filled, leading to a situation where the kernel could not even
-drop caches. Making a small section of swap available (observably, 1 to 15 MB),
-this situation can be resolved and the resulting memory footprint of
-`cloud-hypervisor` is as expected.
-
-## Getting started
-
-We create a folder to build and run Cloud Hypervisor at `$HOME/cloud-hypervisor`
-
-```shell
-$ export CLOUDH=$HOME/cloud-hypervisor
-$ mkdir $CLOUDH
-```
-
-## Prerequisites
-
-You need to install some prerequisite packages to build and test Cloud Hypervisor.
-
-### Tools
-
-```bash
-# Install rust tool chain
-$ curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
-# Install the tools used for building guest kernel, EDK2 and converting guest disk
-$ sudo apt-get update
-$ sudo apt-get install git build-essential m4 bison flex uuid-dev qemu-utils
-```
-
-### Building Cloud Hypervisor
-
-```bash
-$ pushd $CLOUDH
-$ git clone https://github.com/cloud-hypervisor/cloud-hypervisor.git
-$ cd cloud-hypervisor
-$ cargo build
-$ popd
-```
-
-### Disk image
-
-Download the Ubuntu cloud image and convert the image type.
-
-```bash
-$ pushd $CLOUDH
-$ wget https://cloud-images.ubuntu.com/focal/current/focal-server-cloudimg-arm64.img
-$ qemu-img convert -p -f qcow2 -O raw focal-server-cloudimg-arm64.img focal-server-cloudimg-arm64.raw
-$ popd
-```
-
-## UEFI booting
-
-This part introduces how to build EDK2 firmware and boot Cloud Hypervisor with it.
-
-### Building EDK2
-
-```bash
-$ pushd $CLOUDH
-
-# Clone source code repos
-$ git clone --depth 1 https://github.com/tianocore/edk2.git -b master
-$ cd edk2
-$ git submodule update --init
-$ cd ..
-$ git clone --depth 1 https://github.com/tianocore/edk2-platforms.git -b master
-$ git clone --depth 1 https://github.com/acpica/acpica.git -b master
-
-# Build tools
-$ export PACKAGES_PATH="$PWD/edk2:$PWD/edk2-platforms"
-$ export IASL_PREFIX="$PWD/acpica/generate/unix/bin/"
-$ make -C acpica
-$ cd edk2/
-$ . edksetup.sh
-$ cd ..
-$ make -C edk2/BaseTools
-
-# Build EDK2
-$ build -a AARCH64 -t GCC5 -p ArmVirtPkg/ArmVirtCloudHv.dsc -b RELEASE
-
-$ popd
-```
-
-If the build goes well, the EDK2 binary is available at
-`edk2/Build/ArmVirtCloudHv-AARCH64/RELEASE_GCC5/FV/CLOUDHV_EFI.fd`.
-
-### Booting the guest VM
-
-```bash
-$ pushd $CLOUDH
-$ sudo RUST_BACKTRACE=1 $CLOUDH/cloud-hypervisor/target/debug/cloud-hypervisor \
-           --api-socket /tmp/cloud-hypervisor.sock \
-           --kernel $CLOUDH/edk2/Build/ArmVirtCloudHv-AARCH64/RELEASE_GCC5/FV/CLOUDHV_EFI.fd \
-           --disk path=$CLOUDH/focal-server-cloudimg-arm64.raw \
-           --cpus boot=4 \
-           --memory size=4096M \
-           --net tap=,mac=12:34:56:78:90:01,ip=192.168.1.1,mask=255.255.255.0 \
-           --serial tty \
-           --console off
-$ popd
-```
-
-## Direct-kernel booting
-
-Alternativelly, you can build your own kernel for guest VM. This way, UEFI is
-not involved and ACPI cannot be enabled.
-
-### Building kernel
-
-```bash
-$ pushd $CLOUDH
-$ git clone --depth 1 "https://github.com/cloud-hypervisor/linux.git" -b ch-5.12
-$ cd linux
-$ cp $CLOUDH/cloud-hypervisor/resources/linux-config-aarch64 .config
-$ make -j `nproc`
-$ popd
-```
-
-### Booting the guest VM
-
-```bash
-$ pushd $CLOUDH
-$ sudo $CLOUDH/cloud-hypervisor/target/debug/cloud-hypervisor \
-           --api-socket /tmp/cloud-hypervisor.sock \
-           --kernel $CLOUDH/linux/arch/arm64/boot/Image \
-           --disk path=focal-server-cloudimg-arm64.raw \
-           --cmdline "keep_bootcon console=ttyAMA0 reboot=k panic=1 root=/dev/vda1 rw" \
-           --cpus boot=4 \
-           --memory size=4096M \
-           --net tap=,mac=12:34:56:78:90:01,ip=192.168.1.1,mask=255.255.255.0 \
-           --serial tty \
-           --console off
-$ popd
-```

docs/building.md

@@ -24,12 +24,15 @@ Hypervisor. Here, all the steps are based on Ubuntu, for other Linux
 distributions please replace the package manager and package name.
 
 ```shell
-# Install build-essential, git, and qemu-utils
-$ sudo apt install git build-essential qemu-utils
+# Install basic packages needed. For a package list targeting for more
+# functionalities for example the test, please see resources/Dockerfile.
+$ sudo apt-get update
+$ sudo apt install git build-essential m4 bison flex uuid-dev qemu-utils musl-tools
 # Install rust tool chain
 $ curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
 # If you want to build statically linked binary please add musl target
-$ rustup target add x86_64-unknown-linux-musl
+$ rustup target add x86_64-unknown-linux-musl  # x86-64
+$ rustup target add aarch64-unknown-linux-musl # AArch64
 ```
 
 ## Clone and build
@@ -46,7 +49,8 @@ $ cargo build --release
 $ sudo setcap cap_net_admin+ep ./target/release/cloud-hypervisor
 
 # If you want to build statically linked binary
-$ cargo build --release --target=x86_64-unknown-linux-musl --all
+$ cargo build --release --target=x86_64-unknown-linux-musl --all  # x86-64
+$ cargo build --release --target=aarch64-unknown-linux-musl --all # AArch64
 $ popd
 ```
 

docs/custom-image.md

@@ -88,7 +88,7 @@ Ubuntu distributions.
 
 ```bash
 apt update
-apt install fio iperf iperf3 socat stress
+apt install fio iperf iperf3 socat stress cpuid tpm2-tools
 ```
 
 ### Remove counterproductive packages
@@ -158,3 +158,172 @@ as we might need to update the direct kernel boot command line, replacing
 `/dev/vda1` with the appropriate partition number.
 
 Update all references to the previous image name to the new one.
+
+## NVIDIA image for VFIO baremetal CI
+
+Here we are going to describe how to create a cloud image that contains the
+necessary NVIDIA drivers for our VFIO baremetal CI.
+
+### Download base image
+
+We usually start from one of the custom cloud image we have previously created
+but we can use a stock cloud image as well.
+
+```bash
+wget https://cloud-hypervisor.azureedge.net/jammy-server-cloudimg-amd64-custom-20230119-0.raw
+mv jammy-server-cloudimg-amd64-custom-20230119-0.raw jammy-server-cloudimg-amd64-nvidia.raw
+```
+
+### Extend the image size
+
+The NVIDIA drivers consume lots of space, which is why we must resize the image
+before we proceed any further.
+
+```bash
+qemu-img resize jammy-server-cloudimg-amd64-nvidia.raw 5G
+```
+
+### Resize the partition
+
+We use `parted` for fixing the GPT after the image was resized, as well as for
+resizing the `Linux` partition.
+
+```bash
+sudo parted jammy-server-cloudimg-amd64-nvidia.raw
+
+(parted) print
+Warning: Not all of the space available to jammy-server-cloudimg-amd64-nvidia.raw
+appears to be used, you can fix the GPT to use all of the space (an extra 5873664
+blocks) or continue with the current setting?
+Fix/Ignore? Fix
+Model:  (file)
+Disk jammy-server-cloudimg-amd64-nvidia.raw: 5369MB
+Sector size (logical/physical): 512B/512B
+Partition Table: gpt
+Disk Flags:
+
+Number  Start   End     Size    File system  Name  Flags
+14      1049kB  5243kB  4194kB                     bios_grub
+15      5243kB  116MB   111MB   fat32              boot, esp
+ 1      116MB   2361MB  2245MB  ext4
+
+(parted) resizepart 1 5369MB
+(parted) print
+Model:  (file)
+Disk jammy-server-cloudimg-amd64-nvidia.raw: 5369MB
+Sector size (logical/physical): 512B/512B
+Partition Table: gpt
+Disk Flags:
+
+Number  Start   End     Size    File system  Name  Flags
+14      1049kB  5243kB  4194kB                     bios_grub
+15      5243kB  116MB   111MB   fat32              boot, esp
+ 1      116MB   5369MB  5252MB  ext4
+
+(parted) quit
+```
+
+### Create a macvtap interface
+
+Rely on the following [documentation](docs/macvtap-bridge.md) to set up a
+macvtap interface to provide your VM with proper connectivity.
+
+### Boot the image
+
+It is particularly important to boot with a `cloud-init` disk attached to the
+VM as it will automatically resize the Linux `ext4` filesystem based on the
+partition that we have previously resized.
+
+```bash
+./cloud-hypervisor \
+	--kernel hypervisor-fw  \
+	--disk path=focal-server-cloudimg-amd64-nvidia.raw path=/tmp/ubuntu-cloudinit.img \
+	--cpus boot=4 \
+	--memory size=4G \
+	--net fd=3,mac=$mac 3<>$"$tapdevice"
+```
+	
+### Bring up connectivity
+
+If your network has a DHCP server, run the following from your VM
+
+```bash
+sudo dhclient
+```
+
+But if that's not the case, let's give it an IP manually (the IP addresses
+depend on your actual network) and set the DNS server IP address as well.
+
+```bash
+sudo ip addr add 192.168.2.10/24 dev ens4
+sudo ip link set up dev ens4
+sudo ip route add default via 192.168.2.1
+sudo resolvectl dns ens4 8.8.8.8
+```
+
+#### Check connectivity and update the image
+
+```bash
+sudo apt update
+sudo apt upgrade
+```
+
+### Install NVIDIA drivers
+
+The following steps and commands are referenced from the
+[NVIDIA official documentation](https://docs.nvidia.com/datacenter/tesla/tesla-installation-notes/index.html#ubuntu-lts)
+about Tesla compute cards.
+
+```bash
+distribution=$(. /etc/os-release;echo $ID$VERSION_ID | sed -e 's/\.//g')
+wget https://developer.download.nvidia.com/compute/cuda/repos/$distribution/x86_64/cuda-keyring_1.0-1_all.deb
+sudo dpkg -i cuda-keyring_1.0-1_all.deb
+sudo apt-key del 7fa2af80
+sudo apt update
+sudo apt -y install cuda-drivers
+```
+
+### Check the `nvidia-smi` tool
+
+Quickly validate that you can find and run the `nvidia-smi` command from your
+VM. At this point it should fail given no NVIDIA card has been passed through
+the VM, therefore no NVIDIA driver is loaded.
+
+### Workaround LA57 reboot issue
+
+Add `reboot=a` to `GRUB_CMDLINE_LINUX` in `etc/default/grub` so that the VM
+will be booted with the ACPI reboot type. This resolves a reboot issue when
+running on 5-level paging systems.
+
+```bash
+sudo vim /etc/default/grub
+sudo update-grub
+sudo reboot
+```
+
+### Remove previous logins
+
+Since our integration tests rely on past logins to count the number of reboots,
+we must ensure to clear the list.
+
+```bash
+>/var/log/lastlog
+>/var/log/wtmp
+>/var/log/btmp
+```
+
+### Clear history
+
+```
+history -c
+rm /home/cloud/.bash_history
+```
+
+### Reset cloud-init
+
+This is mandatory as we want `cloud-init` provisioning to work again when a new
+VM will be booted with this image.
+
+```
+sudo cloud-init clean
+```

docs/debug-port.md

@@ -36,7 +36,7 @@ Assuming parts of the guest software stack have been instrumented to use the
 `cloud-hypervisor` debug I/O port, we may want to gather the related logs.
 
 To do so we need to start `cloud-hypervisor` with the right debug level
-(`-vvv`). It is also recommended to have it log into a dedicated file in order
+(`-v -v -v`). It is also recommended to have it log into a dedicated file in order
 to easily grep for the tracing logs (e.g.
 `--log-file /tmp/cloud-hypervisor.log`):
 
@@ -48,7 +48,7 @@ to easily grep for the tracing logs (e.g.
     --memory size=1024M \
     --rng \
     --log-file /tmp/ch-fw.log \
-    -vvv
+    -v -v -v
 ```
 
 After booting the guest, we then have to grep for the debug I/O port traces in

docs/gdb.md

@@ -8,7 +8,7 @@ To enable debugging with GDB, build with the `guest_debug` feature enabled:
 cargo build --features guest_debug
 ```
 
-To use the `--gdb` option, specify the Unix Domain Socket with `--path` that Cloud Hypervisor will use to communicate with the host's GDB:
+To use the `--gdb` option, specify the Unix Domain Socket with `path` that Cloud Hypervisor will use to communicate with the host's GDB:
 
 ```bash
 ./cloud-hypervisor \

docs/heap-profiling.md

@@ -0,0 +1,41 @@
+# Heap profiling
+
+Cloud Hypervisor supports generating a profile using
+[dhat](https://docs.rs/dhat/latest/dhat/) of the heap allocations made during
+the runtime of the process.
+
+## Building a suitable binary
+
+This adds the symbol information to the release binary but does not otherwise
+affect the performance.
+
+```
+$ cargo build --profile profiling --features "dhat-heap"
+```
+
+## Generating output
+
+Cloud Hypervisor can then be run as usual. However it is necessary to run with `--seccomp false` as the profiling requires extra syscalls.
+
+```
+$ target/profiling/cloud-hypervisor \
+        --kernel ~/src/linux/vmlinux \
+        --pmem file=~/workloads/focal.raw \
+        --cpus boot=1 --memory size=1G \
+        --cmdline "root=/dev/pmem0p1 console=ttyS0" \
+        --serial tty --console off \
+        --api-socket /tmp/api1 \
+        --seccomp false
+```
+
+When the VMM exits a message like the following will be shown:
+
+```
+dhat: Total:     384,582 bytes in 3,512 blocks
+dhat: At t-gmax: 133,885 bytes in 379 blocks
+dhat: At t-end:  12,160 bytes in 20 blocks
+dhat: The data has been saved to dhat-heap.json, and is viewable with dhat/dh_view.html
+```
+
+The JSON output can then be uploaded to [the dh_view tool](https://nnethercote.github.io/dh_view/dh_view.html) for analysis.
+

docs/hotplug.md

@@ -4,7 +4,7 @@ Currently Cloud Hypervisor supports hot plugging of CPUs devices (x86 only), PCI
 
 ## Kernel support
 
-For hotplug on Cloud Hypervisor ACPI GED support is needed. This can either be achieved by turning on `CONFIG_ACPI_REDUCED_HARDWARE_ONLY` 
+For hotplug on Cloud Hypervisor ACPI GED support is needed. This can either be achieved by turning on `CONFIG_ACPI_REDUCED_HARDWARE_ONLY`
 or by using this kernel patch (available in 5.5-rc1 and later): https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/patch/drivers/acpi/Makefile?id=ac36d37e943635fc072e9d4f47e40a48fbcdb3f0
 
 ## CPU Hot Plug
@@ -27,16 +27,16 @@ $ ./cloud-hypervisor/target/release/cloud-hypervisor \
 	--memory size=1024M \
 	--net "tap=,mac=,ip=,mask=" \
 	--rng \
-	--api-socket=/tmp/ch-socket
+	--api-socket /tmp/ch-socket
 $ popd
 ```
 
-Notice the addition of `--api-socket=/tmp/ch-socket` and a `max` parameter on `--cpus boot=4,max=8`.
+Notice the addition of `--api-socket /tmp/ch-socket` and a `max` parameter on `--cpus boot=4,max=8`.
 
 To ask the VMM to add additional vCPUs then use the resize API:
 
 ```shell
-./ch-remote --api-socket=/tmp/ch-socket resize --cpus 8
+./ch-remote --api-socket /tmp/ch-socket resize --cpus 8
 ```
 
 The extra vCPU threads will be created and advertised to the running kernel. The kernel does not bring up the CPUs immediately and instead the user must "online" them from inside the VM:
@@ -56,7 +56,7 @@ After a reboot the added CPUs will remain.
 Removing CPUs works similarly by reducing the number in the "desired_vcpus" field of the reisze API. The CPUs will be automatically offlined inside the guest so there is no need to run any commands inside the guest:
 
 ```shell
-./ch-remote --api-socket=/tmp/ch-socket resize --cpus 2
+./ch-remote --api-socket /tmp/ch-socket resize --cpus 2
 ```
 
 As per adding CPUs to the guest, after a reboot the VM will be running with the reduced number of vCPUs.
@@ -85,7 +85,7 @@ $ ./cloud-hypervisor/target/release/cloud-hypervisor \
 	--memory size=1024M,hotplug_size=8192M \
 	--net "tap=,mac=,ip=,mask=" \
 	--rng \
-	--api-socket=/tmp/ch-socket
+	--api-socket /tmp/ch-socket
 $ popd
 ```
 
@@ -98,7 +98,7 @@ root@ch-guest ~ # echo online | sudo tee /sys/devices/system/memory/auto_online_
 To ask the VMM to expand the RAM for the VM:
 
 ```shell
-./ch-remote --api-socket=/tmp/ch-socket resize --memory 3G
+./ch-remote --api-socket /tmp/ch-socket resize --memory 3G
 ```
 
 The new memory is now available to use inside the VM:
@@ -134,14 +134,14 @@ $ ./cloud-hypervisor/target/release/cloud-hypervisor \
 	--disk path=focal-server-cloudimg-amd64.raw \
 	--memory size=1024M,hotplug_size=8192M,hotplug_method=virtio-mem \
 	--net "tap=,mac=,ip=,mask=" \
-	--api-socket=/tmp/ch-socket
+	--api-socket /tmp/ch-socket
 $ popd
 ```
 
 To ask the VMM to expand the RAM for the VM (request is in bytes):
 
 ```shell
-./ch-remote --api-socket=/tmp/ch-socket resize --memory 3G
+./ch-remote --api-socket /tmp/ch-socket resize --memory 3G
 ```
 
 The new memory is now available to use inside the VM:
@@ -159,7 +159,7 @@ The same API can also be used to reduce the desired RAM for a VM. It is importan
 
 Extra PCI devices can be added and removed from a running `cloud-hypervisor` instance. This is controlled by making a HTTP API request to the VMM to ask for the additional device to be added, or for the existing device to be removed.
 
-Note: On AArch64 platform, PCI device hotplug can only be achieved using ACPI. Please refer to the [documentation](arm64.md#uefi-booting) for more information.
+Note: On AArch64 platform, PCI device hotplug can only be achieved using ACPI. Please refer to the [documentation](uefi.md#building-uefi-firmware-for-aarch64) for more information.
 
 To use PCI device hotplug start the VM with the HTTP server.
 
@@ -172,17 +172,17 @@ $ ./cloud-hypervisor/target/release/cloud-hypervisor \
 	--cpus boot=4 \
 	--memory size=1024M \
 	--net "tap=,mac=,ip=,mask=" \
-	--api-socket=/tmp/ch-socket
+	--api-socket /tmp/ch-socket
 ```
 
-Notice the addition of `--api-socket=/tmp/ch-socket`.
+Notice the addition of `--api-socket /tmp/ch-socket`.
 
 ### Add VFIO Device
 
 To ask the VMM to add additional VFIO device then use the `add-device` API.
 
 ```shell
-./ch-remote --api-socket=/tmp/ch-socket add-device path=/sys/bus/pci/devices/0000:01:00.0/
+./ch-remote --api-socket /tmp/ch-socket add-device path=/sys/bus/pci/devices/0000:01:00.0/
 ```
 
 ### Add Disk Device
@@ -190,7 +190,7 @@ To ask the VMM to add additional VFIO device then use the `add-device` API.
 To ask the VMM to add additional disk device then use the `add-disk` API.
 
 ```shell
-./ch-remote --api-socket=/tmp/ch-socket add-disk path=/foo/bar/cloud.img
+./ch-remote --api-socket /tmp/ch-socket add-disk path=/foo/bar/cloud.img
 ```
 
 ### Add Fs Device
@@ -198,7 +198,7 @@ To ask the VMM to add additional disk device then use the `add-disk` API.
 To ask the VMM to add additional fs device then use the `add-fs` API.
 
 ```shell
-./ch-remote --api-socket=/tmp/ch-socket add-fs tag=myfs,socket=/foo/bar/virtiofs.sock
+./ch-remote --api-socket /tmp/ch-socket add-fs tag=myfs,socket=/foo/bar/virtiofs.sock
 ```
 
 ### Add Net Device
@@ -206,7 +206,7 @@ To ask the VMM to add additional fs device then use the `add-fs` API.
 To ask the VMM to add additional network device then use the `add-net` API.
 
 ```shell
-./ch-remote --api-socket=/tmp/ch-socket add-net tap=chtap0
+./ch-remote --api-socket /tmp/ch-socket add-net tap=chtap0
 ```
 
 ### Add Pmem Device
@@ -214,7 +214,7 @@ To ask the VMM to add additional network device then use the `add-net` API.
 To ask the VMM to add additional PMEM device then use the `add-pmem` API.
 
 ```shell
-./ch-remote --api-socket=/tmp/ch-socket add-pmem file=/foo/bar.cloud.img
+./ch-remote --api-socket /tmp/ch-socket add-pmem file=/foo/bar.cloud.img
 ```
 
 ### Add Vsock Device
@@ -222,7 +222,7 @@ To ask the VMM to add additional PMEM device then use the `add-pmem` API.
 To ask the VMM to add additional vsock device then use the `add-vsock` API.
 
 ```shell
-./ch-remote --api-socket=/tmp/ch-socket add-vsock cid=3,socket=/foo/bar/vsock.sock
+./ch-remote --api-socket /tmp/ch-socket add-vsock cid=3,socket=/foo/bar/vsock.sock
 ```
 
 ### Common Across All PCI Devices
@@ -244,7 +244,7 @@ After a reboot the added PCI device will remain.
 Removing a PCI device works the same way for all kind of PCI devices. The unique identifier related to the device must be provided. This identifier can be provided by the user when adding the new device, or by default Cloud Hypervisor will assign one.
 
 ```shell
-./ch-remote --api-socket=/tmp/ch-socket remove-device _disk0
+./ch-remote --api-socket /tmp/ch-socket remove-device _disk0
 ```
 
 As per adding a PCI device to the guest, after a reboot the VM will be running without the removed PCI device.

docs/intel_tdx.md

@@ -2,49 +2,64 @@
 
 Intel® Trust Domain Extensions (Intel® TDX) is an Intel technology designed to
 isolate virtual machines from the VMM, hypervisor and any other software on the
-host platform.
+host platform. Here are some useful links:
 
-For more information about TDX technical aspects, design and specification
-please refer to the
-[TDX Homepage](https://www.intel.com/content/www/us/en/developer/articles/technical/intel-trust-domain-extensions.html).
+* [TDX Homepage](https://www.intel.com/content/www/us/en/developer/articles/technical/intel-trust-domain-extensions.html):
+more information about TDX technical aspects, design and specification
 
-The required Linux changes for the host side can be found in the
-[KVM TDX tree](https://github.com/intel/tdx/tree/kvm) while the changes for
-the guest side can be found in the [Guest TDX tree](https://github.com/intel/tdx/tree/guest).
+* [KVM TDX tree](https://github.com/intel/tdx/tree/kvm): the required
+  Linux kernel changes for the host side
 
-The TDVF firmware can be found in the
-[EDK2 staging project](https://github.com/tianocore/edk2-staging/tree/TDVF).
+* [Guest TDX tree](https://github.com/intel/tdx/tree/guest): the Linux
+  kernel changes for the guest side
 
-The TDShim firmware can be found in the
-[Confidential Containers project](https://github.com/confidential-containers/td-shim).
+* [EDK2 project](https://github.com/tianocore/edk2): the TDVF firmware
+
+* [Confidential Containers project](https://github.com/confidential-containers/td-shim):
+  the TDShim firmware
+
+* [TDX Tools](https://github.com/intel/tdx-tools): a collection of tools
+  and scripts to setup TDX environment for testing purpose (such as
+  installing required packages on the host, creating guest images, and
+  building the custom Linux kernel for TDX host and guest)
 
 ## Cloud Hypervisor support
 
-First, you must be running on a machine with TDX enabled in hardware, and
+It is required to use a machine with TDX enabled in hardware and
 with the host OS compiled from the [KVM TDX tree](https://github.com/intel/tdx/tree/kvm).
+The host environment can also be setup with the [TDX Tools](https://github.com/intel/tdx-tools).
 
-Cloud Hypervisor can run TDX VM (Trust Domain) by loading a TD firmware,
+Cloud Hypervisor can run TDX VM (Trust Domain) by loading a TD firmware ([TDVF](https://github.com/tianocore/edk2)),
 which will then load the guest kernel from the image. The image must be custom
 as it must include a kernel built from the [Guest TDX tree](https://github.com/intel/tdx/tree/guest).
+Cloud Hypervisor can also boot a TDX VM with direct kernel boot using [TDshim](https://github.com/confidential-containers/td-shim).
+The custom Linux kernel for the guest can be built with the [TDX Tools](https://github.com/intel/tdx-tools).
+
+> **Note**
+> The latest version of custom host and guest kernel being tested is
+> from [TDX Tools - 2023ww01](https://github.com/intel/tdx-tools/commits/2023ww01).
 
 ### TDVF
 
+> **Note**
+> The latest version of TDVF being tested is [_13b9773_](https://github.com/tianocore/edk2/commit/13b97736c876919b9786055829caaa4fa46984b7).
+
 The firmware can be built as follows:
 
 ```bash
-git clone https://github.com/tianocore/edk2-staging.git
-cd edk2-staging
-git checkout origin/TDVF
+git clone https://github.com/tianocore/edk2.git
+cd edk2
+git checkout 13b97736c876919b9786055829caaa4fa46984b7
 git submodule update --init --recursive
 make -C BaseTools
 source ./edksetup.sh
-build -p OvmfPkg/OvmfCh.dsc -a X64 -t GCC5 -b RELEASE
+build -p OvmfPkg/IntelTdx/IntelTdxX64.dsc -a X64 -t GCC5 -b RELEASE
 ```
 
 If debug logs are needed, here is the alternative command:
 
 ```bash
-build -p OvmfPkg/OvmfCh.dsc -a X64 -t GCC5 -D DEBUG_ON_SERIAL_PORT=TRUE
+build -p OvmfPkg/IntelTdx/IntelTdxX64.dsc -a X64 -t GCC5 -D DEBUG_ON_SERIAL_PORT=TRUE
 ```
 
 On the Cloud Hypervisor side, all you need is to build the project with the
@@ -62,7 +77,7 @@ meaning it will be printing guest kernel logs to the `virtio-console` device.
 ```bash
 ./cloud-hypervisor \
     --platform tdx=on
-    --firmware edk2-staging/Build/OvmfCh/RELEASE_GCC5/FV/OVMF.fd \
+    --firmware edk2/Build/IntelTdx/RELEASE_GCC5/FV/OVMF.fd \
     --cpus boot=1 \
     --memory size=1G \
     --disk path=tdx_guest_img
@@ -74,7 +89,7 @@ firmware:
 ```bash
 ./cloud-hypervisor \
     --platform tdx=on
-    --firmware edk2-staging/Build/OvmfCh/DEBUG_GCC5/FV/OVMF.fd \
+    --firmware edk2/Build/IntelTdx/DEBUG_GCC5/FV/OVMF.fd \
     --cpus boot=1 \
     --memory size=1G \
     --disk path=tdx_guest_img \
@@ -84,21 +99,60 @@ firmware:
 
 ### TDShim
 
+> **Note**
+> The latest version of TDShim being tested is [_66bb334_](https://github.com/confidential-containers/td-shim/tree/66bb33451befbf1291abe3cfea7ee9e99d922b0d).
+
 This is a lightweight version of the TDVF, written in Rust and designed for
 direct kernel boot, which is useful for containers use cases.
 
-You can find the instructions for building the firmware directly from the
-project [documentation](https://github.com/confidential-containers/td-shim/tree/staging#how-to-build).
+To build TDShim from source, it is required to install `Rust`, `NASM`,
+and `LLVM` first. The TDshim can be build as follows:
+```bash
+git clone https://github.com/confidential-containers/td-shim
+cd td-shim
+git checkout 66bb33451befbf1291abe3cfea7ee9e99d922b0d
+cargo install cargo-xbuild
+export CC=clang
+export AR=llvm-ar
+export CC_x86_64_unknown_none=clang
+export AR_x86_64_unknown_none=llvm-ar
+git submodule update --init --recursive
+./sh_script/preparation.sh
+cargo xbuild -p td-shim --target x86_64-unknown-none --release --features=main,tdx
+cargo run -p td-shim-tools --bin td-shim-ld --features=linker -- target/x86_64-unknown-none/release/ResetVector.bin target/x86_64-unknown-none/release/td-shim -o target/release/final.bin
+```
+
+If debug logs from the TDShim is needed, here are the alternative
+commands:
+```bash
+cargo xbuild -p td-shim --target x86_64-unknown-none --features=main,tdx
+cargo run -p td-shim-tools --bin td-shim-ld --features=linker -- target/x86_64-unknown-none/debug/ResetVector.bin target/x86_64-unknown-none/debug/td-shim -o target/debug/final.bin
+```
 
 And run a TDX VM by providing the firmware previously built, along with a guest
-kernel built from the [Guest TDX tree](https://github.com/intel/tdx/tree/guest).
+kernel built from the [Guest TDX tree](https://github.com/intel/tdx/tree/guest)
+or the [TDX Tools](https://github.com/intel/tdx-tools).
 The appropriate kernel boot options must be provided through the `--cmdline`
 option as well.
 
 ```bash
 ./cloud-hypervisor \
     --platform tdx=on
-    --firmware tdshim \
+    --firmware td-shim/target/release/final.bin \
+    --kernel bzImage \
+    --cmdline "root=/dev/vda3 console=hvc0 rw"
+    --cpus boot=1 \
+    --memory size=1G \
+    --disk path=tdx_guest_img
+```
+
+And here is the alternative command when looking for debug logs from the
+TDShim:
+
+```bash
+./cloud-hypervisor \
+    --platform tdx=on
+    --firmware td-shim/target/debug/final.bin \
     --kernel bzImage \
     --cmdline "root=/dev/vda3 console=hvc0 rw"
     --cpus boot=1 \

docs/iommu.md

@@ -245,7 +245,7 @@ e.g.
 
 ```bash
 ./cloud-hypervisor \
-    --api-socket=/tmp/api \
+    --api-socket /tmp/api \
     --cpus boot=1 \
     --memory size=4G,hugepages=on \
     --disk path=focal-server-cloudimg-amd64.raw \
@@ -260,7 +260,7 @@ requiring the IOMMU then may be hotplugged:
 e.g.
 
 ```bash
-./ch-remote --api-socket=/tmp/api add-device path=/sys/bus/pci/devices/0000:00:04.0,iommu=on,pci_segment=1
+./ch-remote --api-socket /tmp/api add-device path=/sys/bus/pci/devices/0000:00:04.0,iommu=on,pci_segment=1
 ```
 
 Devices that cannot be placed behind an IOMMU (e.g. lacking an `iommu=` option)

docs/live_migration.md

@@ -16,22 +16,22 @@ $ target/release/cloud-hypervisor
     --disk path=~/workloads/focal.raw \
     --cpus boot=1 --memory size=1G,shared=on \
     --cmdline "root=/dev/vda1 console=ttyS0"  \
-    --serial tty --console off --api-socket=/tmp/api1
+    --serial tty --console off --api-socket /tmp/api1
 ```
 
 Launch the destination VM from the same directory (on the host machine):
 ```bash
-$ target/release/cloud-hypervisor --api-socket=/tmp/api2
+$ target/release/cloud-hypervisor --api-socket /tmp/api2
 ```
 
 Get ready for receiving migration for the destination VM (on the host machine):
 ```bash
-$ target/release/ch-remote --api-socket=/tmp/api2 receive-migration unix:/tmp/sock
+$ target/release/ch-remote --api-socket /tmp/api2 receive-migration unix:/tmp/sock
 ```
 
 Start to send migration for the source VM (on the host machine):
 ```bash
-$ target/release/ch-remote --api-socket=/tmp/api1 send-migration --local unix:/tmp/sock
+$ target/release/ch-remote --api-socket /tmp/api1 send-migration --local unix:/tmp/sock
 ```
 
 When the above commands completed, the source VM should be successfully
@@ -51,7 +51,7 @@ $ sudo /target/release/cloud-hypervisor \
         --cpus boot=1 --memory size=512M \
         --kernel vmlinux \
         --cmdline "root=/dev/vda1 console=ttyS0"  \
-        --disk path=focal-1.raw path=focal-nested.raw path=tmp.img\
+        --disk path=focal-1.raw path=focal-nested.raw --disk path=tmp.img\
         --net ip=192.168.101.1
 ```
 
@@ -63,7 +63,7 @@ $ sudo /target/release/cloud-hypervisor \
         --cpus boot=1 --memory size=512M \
         --kernel vmlinux \
         --cmdline "root=/dev/vda1 console=ttyS0"  \
-        --disk path=focal-2.raw path=focal-nested.raw path=tmp.img\
+        --disk path=focal-2.raw path=focal-nested.raw --disk path=tmp.img\
         --net ip=192.168.102.1
 ```
 
@@ -74,8 +74,8 @@ vm-1:~$ sudo ./cloud-hypervisor \
         --memory size=128M \
         --kernel vmlinux \
         --cmdline "console=ttyS0 root=/dev/vda1" \
-        --disk path=/dev/vdb path=/dev/vdc \
-        --api-socket=/tmp/api1 \
+        --disk path=/dev/vdb --disk path=/dev/vdc \
+        --api-socket /tmp/api1 \
         --net ip=192.168.100.1
 vm-1:~$ # setup the guest network if needed
 vm-1:~$ sudo ip addr add 192.168.101.2/24 dev ens4
@@ -108,7 +108,7 @@ echo "tmp = $tmp"
 
 Launch the nested destination VM (inside the guest OS of the VM 2):
 ```bash
-vm-2:~$ sudo ./cloud-hypervisor --api-socket=/tmp/api2
+vm-2:~$ sudo ./cloud-hypervisor --api-socket /tmp/api2
 vm-2:~$ # setup the guest network with the following commands if needed
 vm-2:~$ sudo ip addr add 192.168.102.2/24 dev ens4
 vm-2:~$ sudo ip link set up dev ens4
@@ -122,7 +122,7 @@ vm-2:~$ ping 192.168.101.2 # This should succeed
 Get ready for receiving migration for the nested destination VM (inside
 the guest OS of the VM 2):
 ```bash
-vm-2:~$ sudo ./ch-remote --api-socket=/tmp/api2 receive-migration unix:/tmp/sock2
+vm-2:~$ sudo ./ch-remote --api-socket /tmp/api2 receive-migration unix:/tmp/sock2
 vm-2:~$ sudo socat TCP-LISTEN:6000,reuseaddr UNIX-CLIENT:/tmp/sock2
 ```
 
@@ -130,7 +130,7 @@ Start to send migration for the nested source VM (inside the guest OS of
 the VM 1):
 ```bash
 vm-1:~$ sudo socat UNIX-LISTEN:/tmp/sock1,reuseaddr TCP:192.168.102.2:6000
-vm-1:~$ sudo ./ch-remote --api-socket=/tmp/api1 send-migration unix:/tmp/sock1
+vm-1:~$ sudo ./ch-remote --api-socket /tmp/api1 send-migration unix:/tmp/sock1
 ```
 
 When the above commands completed, the source VM should be successfully

docs/logging.md

@@ -38,6 +38,6 @@ This level is for the benefit of developers. It should be used for sporadic and
 
 ### `debug!()`
 
-Use `-vv` to enable.
+Use `-v -v` to enable.
 
-For the most verbose of logging messages. It is acceptable to "spam" the log with repeated invocations of the same message. This level of logging would be combined with `--log-file`.
+For the most verbose of logging messages. It is acceptable to "spam" the log with repeated invocations of the same message. This level of logging would be combined with `--log-file`.

docs/memory.md

@@ -266,11 +266,8 @@ _Example_
 
 ### `file`
 
-Path to the file backing the memory zone. This can be either a file or a
-directory. In case of a file, it will be opened and used as the backing file
-for the `mmap(2)` operation. In case of a directory, a temporary file with no
-hard link on the filesystem will be created. This file will be used as the
-backing file for the `mmap(2)` operation.
+Path to the file backing the memory zone. The file will be opened and used as
+the backing file for the `mmap(2)` operation.
 
 This option can be particularly useful when trying to back a part of the guest
 RAM with a well known file. In the context of the snapshot/restore feature, and
@@ -519,7 +516,7 @@ different distances, it can be described with the following example.
 _Example_
 
 ```
---numa guest_numa_id=0,distances=[1@15,2@25] guest_numa_id=1,distances=[0@15,2@20] guest_numa_id=2,distances=[0@25,1@20]
+--numa guest_numa_id=0,distances=[1@15,2@25] --numa guest_numa_id=1,distances=[0@15,2@20] guest_numa_id=2,distances=[0@25,1@20]
 ```
 
 ### `memory_zones`
@@ -543,14 +540,14 @@ demarcate the list.
 
 Note that a memory zone must belong to a single NUMA node. The following
 configuration is incorrect, therefore not allowed:
-`--numa guest_numa_id=0,memory_zones=mem0 guest_numa_id=1,memory_zones=mem0`
+`--numa guest_numa_id=0,memory_zones=mem0 --numa guest_numa_id=1,memory_zones=mem0`
 
 _Example_
 
 ```
 --memory size=0
---memory-zone id=mem0,size=1G id=mem1,size=1G id=mem2,size=1G
---numa guest_numa_id=0,memory_zones=[mem0,mem2] guest_numa_id=1,memory_zones=mem1
+--memory-zone id=mem0,size=1G id=mem1,size=1G --memory-zone id=mem2,size=1G
+--numa guest_numa_id=0,memory_zones=[mem0,mem2] --numa guest_numa_id=1,memory_zones=mem1
 ```
 
 ### `sgx_epc_sections`
@@ -570,7 +567,7 @@ _Example_
 
 ```
 --sgx-epc id=epc0,size=32M id=epc1,size=64M id=epc2,size=32M
---numa guest_numa_id=0,sgx_epc_sections=epc1 guest_numa_id=1,sgx_epc_sections=[epc0,epc2]
+--numa guest_numa_id=0,sgx_epc_sections=epc1 --numa guest_numa_id=1,sgx_epc_sections=[epc0,epc2]
 ```
 
 ### PCI bus

docs/profiling.md

@@ -4,20 +4,12 @@
 
 ## Building a suitable binary
 
-Modify the `Cargo.toml` file to add `debug = 1` to the `[profile.release]` block. It should look like this:
-
-```
-[profile.release]
-lto = true
-debug = 1
-```
-
 This adds the symbol information to the release binary but does not otherwise affect the performance.
 
 The binary must also be built with frame pointers included so that the call graph can be captured by the profiler.
 
 ```
-$ cargo clean && RUSTFLAGS='-C force-frame-pointers=y' cargo build --release
+$ cargo clean && RUSTFLAGS='-C force-frame-pointers=y' cargo build --profile profiling
 ```
 
 ## Profiling
@@ -27,13 +19,13 @@ $ cargo clean && RUSTFLAGS='-C force-frame-pointers=y' cargo build --release
 e.g.
 
 ```
-$ perf record -g target/release/cloud-hypervisor \
+$ perf record -g target/profiling/cloud-hypervisor \
         --kernel ~/src/linux/vmlinux \
         --pmem file=~/workloads/focal.raw \
         --cpus boot=1 --memory size=1G \
         --cmdline "root=/dev/pmem0p1 console=ttyS0" \
         --serial tty --console off \
-        --api-socket=/tmp/api1
+        --api-socket /tmp/api1
 ```
 
 For analysing the samples:
@@ -60,5 +52,5 @@ $ perf record --call-graph lbr --all-user --user-callchains -g target/release/cl
         --cpus boot=1 --memory size=1G \
         --cmdline "root=/dev/pmem0p1 console=ttyS0" \
         --serial tty --console off \
-        --api-socket=/tmp/api1
+        --api-socket /tmp/api1
 ```

docs/snapshot_restore.md

@@ -25,14 +25,14 @@ First thing, we must run a Cloud Hypervisor VM:
 At any point in time when the VM is running, one might choose to pause it:
 
 ```bash
-./ch-remote --api-socket=/tmp/cloud-hypervisor.sock pause
+./ch-remote --api-socket /tmp/cloud-hypervisor.sock pause
 ```
 
 Once paused, the VM can be safely snapshot into the specified directory and
 using the following command:
 
 ```bash
-./ch-remote --api-socket=/tmp/cloud-hypervisor.sock snapshot file:///home/foo/snapshot
+./ch-remote --api-socket /tmp/cloud-hypervisor.sock snapshot file:///home/foo/snapshot
 ```
 
 Given the directory was present on the system, the snapshot will succeed and
@@ -79,7 +79,7 @@ Or using two different commands from two terminals:
 ./cloud-hypervisor --api-socket /tmp/cloud-hypervisor.sock
 
 # Second terminal
-./ch-remote --api-socket=/tmp/cloud-hypervisor.sock restore source_url=file:///home/foo/snapshot
+./ch-remote --api-socket /tmp/cloud-hypervisor.sock restore source_url=file:///home/foo/snapshot
 ```
 
 Remember the VM is restored in a `paused` state, which was the VM's state when
@@ -87,7 +87,7 @@ it was snapshot. For this reason, one must explicitly `resume` the VM before to
 start using it.
 
 ```bash
-./ch-remote --api-socket=/tmp/cloud-hypervisor.sock resume
+./ch-remote --api-socket /tmp/cloud-hypervisor.sock resume
 ```
 
 At this point, the VM is fully restored and is identical to the VM which was

docs/uefi.md

@@ -2,17 +2,19 @@
 
 Cloud Hypervisor supports UEFI boot through the utilization of the EDK II based UEFI firmware. 
 
-## Building UEFI Firmware
+## Building UEFI Firmware for x86-64
 
 To avoid any unnecessary issues, it is recommended to use Ubuntu 18.04 and its default toolset. Any other compatible Linux distribution is otherwise suitable, however it is suggested to use a temporary Docker container with Ubuntu 18.04 for a quick build on an existing Linux machine.
 
+Please note that nasm-2.15 is required for the build to succeed.
+
 The commands below will compile an OVMF firmware suitable for Cloud Hypervisor.
 
 ```shell
 sudo apt-get update
 sudo apt-get install uuid-dev nasm iasl build-essential python3-distutils git
 
-git clone https://github.com/cloud-hypervisor/edk2 -b ch
+git clone https://github.com/tianocore/edk2
 cd edk2
 . edksetup.sh
 git submodule update --init
@@ -27,11 +29,40 @@ build
 
 After the successful build, the resulting firmware binaries are available under `Build/CloudHvX64/DEBUG_GCC5/FV` underneath the edk2 checkout.
 
+## Building UEFI Firmware for AArch64
+
+```shell
+# On an AArch64 machine:
+$ sudo apt-get update
+$ sudo apt-get install uuid-dev nasm iasl build-essential python3-distutils git
+$ git clone --depth 1 https://github.com/tianocore/edk2.git -b master
+$ cd edk2
+$ git submodule update --init
+$ cd ..
+$ git clone --depth 1 https://github.com/tianocore/edk2-platforms.git -b master
+$ git clone --depth 1 https://github.com/acpica/acpica.git -b master
+
+# Build tools
+$ export PACKAGES_PATH="$PWD/edk2:$PWD/edk2-platforms"
+$ export IASL_PREFIX="$PWD/acpica/generate/unix/bin/"
+$ make -C acpica
+$ cd edk2/
+$ . edksetup.sh
+$ cd ..
+$ make -C edk2/BaseTools
+
+# Build EDK2
+$ build -a AARCH64 -t GCC5 -p ArmVirtPkg/ArmVirtCloudHv.dsc -b RELEASE
+```
+
+If the build goes well, the EDK2 binary is available at
+`edk2/Build/ArmVirtCloudHv-AARCH64/RELEASE_GCC5/FV/CLOUDHV_EFI.fd`.
+
 ## Using OVMF Binaries
 
 Any UEFI capable image can be booted using the Cloud Hypervisor specific firmware. Windows guests under Cloud Hypervisor only support UEFI boot, therefore OVMF is mandatory there.
 
-To make Cloud Hypervisor use UEFI boot, pass the `CLOUDHV.fd` file path as an argument to the `--kernel` option. The firmware file will be opened in read only mode.
+To make Cloud Hypervisor use UEFI boot, pass the `CLOUDHV.fd` (for x86-64) / `CLOUDHV_EFI.fd` (for AArch64) file path as an argument to the `--kernel` option. The firmware file will be opened in read only mode.
 
 # Links
 

docs/vhost-user-blk-testing.md

@@ -94,7 +94,7 @@ VMs run in client mode. They connect to the socket created by the `dpdkvhostuser
         --memory size=1024M,hugepages=on,shared=true \
         --kernel linux/arch/x86/boot/compressed/vmlinux.bin \
         --cmdline "console=ttyS0 root=/dev/vda1 rw iommu=off" \
-        --disk path=images/focal-server-cloudimg-amd64.raw vhost_user=true,socket=/var/tmp/vhost.1,num_queues=4,queue_size=128 \
+        --disk path=images/focal-server-cloudimg-amd64.raw --disk vhost_user=true,socket=/var/tmp/vhost.1,num_queues=4,queue_size=128 \
         --console off \
         --serial tty \
         --rng

event_monitor/Cargo.toml

@@ -5,7 +5,6 @@ authors = ["The Cloud Hypervisor Authors"]
 edition = "2021"
 
 [dependencies]
-libc = "0.2.138"
-once_cell = "1.16.0"
-serde = { version = "1.0.150", features = ["rc", "derive"] }
-serde_json = "1.0.89"
+libc = "0.2.139"
+serde = { version = "1.0.156", features = ["rc", "derive"] }
+serde_json = "1.0.96"

event_monitor/src/lib.rs

@@ -3,7 +3,6 @@
 // SPDX-License-Identifier: Apache-2.0
 //
 
-use once_cell::sync::OnceCell;
 use serde::Serialize;
 use std::borrow::Cow;
 use std::collections::HashMap;
@@ -12,14 +11,15 @@ use std::io::Write;
 use std::os::unix::io::AsRawFd;
 use std::time::{Duration, Instant};
 
-static MONITOR: OnceCell<(File, Instant)> = OnceCell::new();
+static mut MONITOR: Option<(File, Instant)> = None;
 
 /// This function must only be called once from the main process before any threads
 /// are created to avoid race conditions
 pub fn set_monitor(file: File) -> Result<(), std::io::Error> {
-    // There is only one caller of this function, so MONITOR is written to only once
-    assert!(MONITOR.get().is_none());
+    // SAFETY: there is only one caller of this function, so MONITOR is written to only once
+    assert!(unsafe { MONITOR.is_none() });
     let fd = file.as_raw_fd();
+    // SAFETY: FFI call to configure the fd
     let ret = unsafe {
         let mut flags = libc::fcntl(fd, libc::F_GETFL);
         flags |= libc::O_NONBLOCK;
@@ -28,9 +28,10 @@ pub fn set_monitor(file: File) -> Result<(), std::io::Error> {
     if ret < 0 {
         return Err(std::io::Error::last_os_error());
     }
-
-    MONITOR.get_or_init(|| (file, Instant::now()));
-
+    // SAFETY: MONITOR is None. Nobody else can hold a reference to it.
+    unsafe {
+        MONITOR = Some((file, Instant::now()));
+    };
     Ok(())
 }
 
@@ -43,7 +44,8 @@ struct Event<'a> {
 }
 
 pub fn event_log(source: &str, event: &str, properties: Option<&HashMap<Cow<str>, Cow<str>>>) {
-    if let Some((file, start)) = MONITOR.get().as_ref() {
+    // SAFETY: MONITOR is always in a valid state (None or Some).
+    if let Some((file, start)) = unsafe { MONITOR.as_ref() } {
         let e = Event {
             timestamp: start.elapsed(),
             source,

fuzz/Cargo.lock

@@ -5,15 +5,16 @@ version = 3
 [[package]]
 name = "acpi_tables"
 version = "0.1.0"
+source = "git+https://github.com/rust-vmm/acpi_tables?branch=main#98dcb0309d362dd83f6ffcac4f66914a2fbd5a73"
 dependencies = [
- "vm-memory",
+ "zerocopy",
 ]
 
 [[package]]
 name = "anyhow"
-version = "1.0.66"
+version = "1.0.71"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "216261ddc8289130e551ddcd5ce8a064710c0d064a4d2895c67151c92b5443f6"
+checksum = "9c7d0618f0e0b7e8ff11427422b64564d5fb0be1940354bfe2e0529b18a9d9b8"
 
 [[package]]
 name = "api_client"
@@ -24,15 +25,15 @@ dependencies = [
 
 [[package]]
 name = "arbitrary"
-version = "1.2.0"
+version = "1.3.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "29d47fbf90d5149a107494b15a7dc8d69b351be2db3bb9691740e88ec17fd880"
+checksum = "e2d098ff73c1ca148721f37baad5ea6a465a13f9573aba8641fbbbae8164a54e"
 
 [[package]]
 name = "arc-swap"
-version = "1.5.1"
+version = "1.6.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "983cd8b9d4b02a6dc6ffa557262eb5858a27a0038ffffe21a0f133eaa819a164"
+checksum = "bddcadddf5e9015d310179a59bb28c4d4b9920ad0f11e8e14dbadf654890c9a6"
 
 [[package]]
 name = "arch"
@@ -56,6 +57,34 @@ dependencies = [
  "vmm-sys-util",
 ]
 
+[[package]]
+name = "argh"
+version = "0.1.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ab257697eb9496bf75526f0217b5ed64636a9cfafa78b8365c71bd283fcef93e"
+dependencies = [
+ "argh_derive",
+ "argh_shared",
+]
+
+[[package]]
+name = "argh_derive"
+version = "0.1.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b382dbd3288e053331f03399e1db106c9fb0d8562ad62cb04859ae926f324fa6"
+dependencies = [
+ "argh_shared",
+ "proc-macro2",
+ "quote",
+ "syn 1.0.109",
+]
+
+[[package]]
+name = "argh_shared"
+version = "0.1.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "64cb94155d965e3d37ffbbe7cc5b82c3dd79dd33bd48e536f73d2cfb8d85506f"
+
 [[package]]
 name = "bincode"
 version = "1.3.3"
@@ -71,12 +100,6 @@ version = "1.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "bef38d45163c2f1dde094a7dfd33ccf595c92905c8f8f4fdc18d06fb1037718a"
 
-[[package]]
-name = "bitflags"
-version = "2.4.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "327762f6e5a765692301e5bb513e0d9fef63be86bbc14528052b1cd3e6f03e07"
-
 [[package]]
 name = "block_util"
 version = "0.1.0"
@@ -85,11 +108,12 @@ dependencies = [
  "libc",
  "log",
  "qcow",
+ "smallvec",
  "thiserror",
  "versionize",
  "versionize_derive",
  "vhdx",
- "virtio-bindings",
+ "virtio-bindings 0.2.0",
  "virtio-queue",
  "vm-memory",
  "vm-virtio",
@@ -104,9 +128,9 @@ checksum = "14c189c53d098945499cdfa7ecc63567cf3886b3332b312a5b4585d8d3a6a610"
 
 [[package]]
 name = "cc"
-version = "1.0.77"
+version = "1.0.79"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e9f73505338f7d905b19d18738976aae232eb46b8efc15554ffc56deb5d9ebe4"
+checksum = "50d30906286121d95be3d479533b458f87493b30a4b5f79a607db8f5d11aa91f"
 dependencies = [
  "jobserver",
 ]
@@ -117,37 +141,13 @@ version = "1.0.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "baf1de4339761588bc0619e3cbc0120ee582ebb74b53b4efbf79117bd2da40fd"
 
-[[package]]
-name = "clap"
-version = "4.0.29"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4d63b9e9c07271b9957ad22c173bae2a4d9a81127680962039296abcd2f8251d"
-dependencies = [
- "bitflags 1.3.2",
- "clap_lex",
- "is-terminal",
- "once_cell",
- "strsim",
- "termcolor",
- "terminal_size",
-]
-
-[[package]]
-name = "clap_lex"
-version = "0.3.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0d4198f73e42b4936b35b5bb248d81d2b595ecb170da0bac7655c54eedfa8da8"
-dependencies = [
- "os_str_bytes",
-]
-
 [[package]]
 name = "cloud-hypervisor"
-version = "28.3.0"
+version = "31.0.0"
 dependencies = [
  "anyhow",
  "api_client",
- "clap",
+ "argh",
  "epoll",
  "event_monitor",
  "hypervisor",
@@ -175,7 +175,9 @@ dependencies = [
  "epoll",
  "libc",
  "libfuzzer-sys",
+ "linux-loader",
  "micro_http",
+ "net_util",
  "once_cell",
  "qcow",
  "seccompiler",
@@ -206,9 +208,9 @@ checksum = "55626594feae15d266d52440b26ff77de0e22230cf0c113abe619084c1ddc910"
 
 [[package]]
 name = "darling"
-version = "0.14.2"
+version = "0.20.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b0dd3cd20dc6b5a876612a6e5accfe7f3dd883db6d07acfbf14c128f61550dfa"
+checksum = "0558d22a7b463ed0241e993f76f09f30b126687447751a8638587b864e4b3944"
 dependencies = [
  "darling_core",
  "darling_macro",
@@ -216,27 +218,27 @@ dependencies = [
 
 [[package]]
 name = "darling_core"
-version = "0.14.2"
+version = "0.20.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a784d2ccaf7c98501746bf0be29b2022ba41fd62a2e622af997a03e9f972859f"
+checksum = "ab8bfa2e259f8ee1ce5e97824a3c55ec4404a0d772ca7fa96bf19f0752a046eb"
 dependencies = [
  "fnv",
  "ident_case",
  "proc-macro2",
  "quote",
  "strsim",
- "syn",
+ "syn 2.0.15",
 ]
 
 [[package]]
 name = "darling_macro"
-version = "0.14.2"
+version = "0.20.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7618812407e9402654622dd402b0a89dff9ba93badd6540781526117b92aab7e"
+checksum = "29a358ff9f12ec09c3e61fef9b5a9902623a695a46a917b07f269bff1445611a"
 dependencies = [
  "darling_core",
  "quote",
- "syn",
+ "syn 2.0.15",
 ]
 
 [[package]]
@@ -246,12 +248,11 @@ dependencies = [
  "acpi_tables",
  "anyhow",
  "arch",
- "bitflags 2.4.1",
+ "bitflags",
  "byteorder",
  "hypervisor",
  "libc",
  "log",
- "phf",
  "thiserror",
  "tpm",
  "versionize",
@@ -268,28 +269,7 @@ version = "4.3.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "20df693c700404f7e19d4d6fae6b15215d2913c27955d2b9d6f2c0f537511cd0"
 dependencies = [
- "bitflags 1.3.2",
- "libc",
-]
-
-[[package]]
-name = "errno"
-version = "0.2.8"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f639046355ee4f37944e44f60642c6f3a7efa3cf6b78c78a0d989a8ce6c396a1"
-dependencies = [
- "errno-dragonfly",
- "libc",
- "winapi",
-]
-
-[[package]]
-name = "errno-dragonfly"
-version = "0.1.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "aa68f1b12764fab894d2755d2518754e71b4fd80ecfb822714a1206c2aab39bf"
-dependencies = [
- "cc",
+ "bitflags",
  "libc",
 ]
 
@@ -298,16 +278,15 @@ name = "event_monitor"
 version = "0.1.0"
 dependencies = [
  "libc",
- "once_cell",
  "serde",
  "serde_json",
 ]
 
 [[package]]
 name = "fdt"
-version = "0.1.4"
+version = "0.1.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "964f5becd44d069dca0beea2b4bc05639ae7bf3b3f5369c295aff360bb57cca2"
+checksum = "784a4df722dc6267a04af36895398f59d21d07dce47232adf31ec0ff2fa45e67"
 
 [[package]]
 name = "fnv"
@@ -317,24 +296,15 @@ checksum = "3f9eec918d3f24069decb9af1554cad7c880e2da24a9afd88aca000531ab82c1"
 
 [[package]]
 name = "getrandom"
-version = "0.2.8"
+version = "0.2.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c05aeb6a22b8f62540c194aac980f2115af067bfe15a0734d7277a768d396b31"
+checksum = "c85e1d9ab2eadba7e5040d4e09cbd6d072b76a557ad64e797c2cb9d4da21d7e4"
 dependencies = [
  "cfg-if",
  "libc",
  "wasi",
 ]
 
-[[package]]
-name = "hermit-abi"
-version = "0.2.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ee512640fe35acbfb4bb779db6f0d80704c2cacfa2e39b601ef3e3f47d1ae4c7"
-dependencies = [
- "libc",
-]
-
 [[package]]
 name = "hypervisor"
 version = "0.1.0"
@@ -356,12 +326,11 @@ dependencies = [
 
 [[package]]
 name = "iced-x86"
-version = "1.17.0"
+version = "1.18.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "158f5204401d08f91d19176112146d75e99b3cf745092e268fa7be33e09adcec"
+checksum = "1dd04b950d75b3498320253b17fb92745b2cc79ead8814aede2f7c1bab858bec"
 dependencies = [
  "lazy_static",
- "static_assertions",
 ]
 
 [[package]]
@@ -370,49 +339,27 @@ version = "1.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b9e0384b61958566e926dc50660321d12159025e767c18e043daf26b70104c39"
 
-[[package]]
-name = "io-lifetimes"
-version = "1.0.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "46112a93252b123d31a119a8d1a1ac19deac4fac6e0e8b0df58f0d4e5870e63c"
-dependencies = [
- "libc",
- "windows-sys",
-]
-
 [[package]]
 name = "io-uring"
-version = "0.5.9"
+version = "0.5.13"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7ba34abb5175052fc1a2227a10d2275b7386c9990167de9786c0b88d8b062330"
+checksum = "dd1e1a01cfb924fd8c5c43b6827965db394f5a3a16c599ce03452266e1cf984c"
 dependencies = [
- "bitflags 1.3.2",
+ "bitflags",
  "libc",
 ]
 
-[[package]]
-name = "is-terminal"
-version = "0.4.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "927609f78c2913a6f6ac3c27a4fe87f43e2a35367c0c4b0f8265e8f49a104330"
-dependencies = [
- "hermit-abi",
- "io-lifetimes",
- "rustix",
- "windows-sys",
-]
-
 [[package]]
 name = "itoa"
-version = "1.0.4"
+version = "1.0.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4217ad341ebadf8d8e724e264f13e593e0648f5b3e94b3896a5df283be015ecc"
+checksum = "453ad9f582a441959e5f0d088b02ce04cfe8d51a8eaf077f12ac6d3e94164ca6"
 
 [[package]]
 name = "jobserver"
-version = "0.1.25"
+version = "0.1.26"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "068b1ee6743e4d11fb9c6a1e6064b3693a1b600e7f5f5988047d98b3dc9fb90b"
+checksum = "936cfd212a0155903bcbc060e316fb6cc7cbf2e1907329391ebadc1fe0ce77c2"
 dependencies = [
  "libc",
 ]
@@ -446,15 +393,15 @@ checksum = "e2abad23fbc42b3700f2f279844dc832adb2b2eb069b2df918f455c4e18cc646"
 
 [[package]]
 name = "libc"
-version = "0.2.138"
+version = "0.2.144"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "db6d7e329c562c5dfab7a46a2afabc8b987ab9a4834c9d1ca04dc54c1546cef8"
+checksum = "2b00cc1c228a6782d0f076e7b232802e0c5689d41bb5df366f2a6b6621cfdfe1"
 
 [[package]]
 name = "libfuzzer-sys"
-version = "0.4.5"
+version = "0.4.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c8fff891139ee62800da71b7fd5b508d570b9ad95e614a53c6f453ca08366038"
+checksum = "beb09950ae85a0a94b27676cccf37da5ff13f27076aa1adbc6545dd0d0e1bd4e"
 dependencies = [
  "arbitrary",
  "cc",
@@ -470,12 +417,6 @@ dependencies = [
  "vm-memory",
 ]
 
-[[package]]
-name = "linux-raw-sys"
-version = "0.1.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8f9f08d8963a6c613f4b1a78f4f4a4dbfadf8e6545b2d72861731e4858b8b47f"
-
 [[package]]
 name = "log"
 version = "0.4.17"
@@ -488,7 +429,7 @@ dependencies = [
 [[package]]
 name = "micro_http"
 version = "0.1.0"
-source = "git+https://github.com/firecracker-microvm/micro-http?branch=main#4b18a043e997da5b5f679e3defc279fec908753e"
+source = "git+https://github.com/firecracker-microvm/micro-http?branch=main#b538bf89e50be83b6fa9ab1896727ff61e02fa13"
 dependencies = [
  "libc",
  "vmm-sys-util",
@@ -515,7 +456,7 @@ dependencies = [
  "thiserror",
  "versionize",
  "versionize_derive",
- "virtio-bindings",
+ "virtio-bindings 0.2.0",
  "virtio-queue",
  "vm-memory",
  "vm-virtio",
@@ -524,20 +465,14 @@ dependencies = [
 
 [[package]]
 name = "once_cell"
-version = "1.16.0"
+version = "1.17.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "86f0b0d4bf799edbc74508c1e8bf170ff5f41238e5f8225603ca7caaae2b7860"
+checksum = "b7e5500299e16ebb147ae15a00a942af264cf3688f47923b8fc2cd5858f23ad3"
 
 [[package]]
 name = "option_parser"
 version = "0.1.0"
 
-[[package]]
-name = "os_str_bytes"
-version = "6.4.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9b7820b9daea5457c9f21c69448905d723fbd21136ccf521748f23fd49e723ee"
-
 [[package]]
 name = "pci"
 version = "0.1.0"
@@ -561,53 +496,11 @@ dependencies = [
  "vmm-sys-util",
 ]
 
-[[package]]
-name = "phf"
-version = "0.11.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "928c6535de93548188ef63bb7c4036bd415cd8f36ad25af44b9789b2ee72a48c"
-dependencies = [
- "phf_macros",
- "phf_shared",
-]
-
-[[package]]
-name = "phf_generator"
-version = "0.11.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b1181c94580fa345f50f19d738aaa39c0ed30a600d95cb2d3e23f94266f14fbf"
-dependencies = [
- "phf_shared",
- "rand",
-]
-
-[[package]]
-name = "phf_macros"
-version = "0.11.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "92aacdc5f16768709a569e913f7451034034178b05bdc8acda226659a3dccc66"
-dependencies = [
- "phf_generator",
- "phf_shared",
- "proc-macro2",
- "quote",
- "syn",
-]
-
-[[package]]
-name = "phf_shared"
-version = "0.11.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e1fb5f6f826b772a8d4c0394209441e7d37cbbb967ae9c7e0e8134365c9ee676"
-dependencies = [
- "siphasher",
-]
-
 [[package]]
 name = "proc-macro2"
-version = "1.0.76"
+version = "1.0.57"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "95fc56cda0b5c3325f5fbbd7ff9fda9e02bb00bb3dac51252d2f1bfa1cb8cc8c"
+checksum = "c4ec6d5fe0b140acb27c9a0444118cf55bfbb4e0b259739429abb4521dd67c16"
 dependencies = [
  "unicode-ident",
 ]
@@ -625,28 +518,13 @@ dependencies = [
 
 [[package]]
 name = "quote"
-version = "1.0.21"
+version = "1.0.27"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "bbe448f377a7d6961e30f5955f9b8d106c3f5e449d493ee1b125c1d43c2b5179"
+checksum = "8f4f29d145265ec1c483c7c654450edde0bfe043d3938d6972630663356d9500"
 dependencies = [
  "proc-macro2",
 ]
 
-[[package]]
-name = "rand"
-version = "0.8.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "34af8d1a0e25924bc5b7c43c079c942339d8f0a8b57c39049bef581b46327404"
-dependencies = [
- "rand_core",
-]
-
-[[package]]
-name = "rand_core"
-version = "0.6.4"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ec0be4795e2f6a28069bec0b5ff3e2ac9bafc99e6a9a7dc3547996c5c816922c"
-
 [[package]]
 name = "rate_limiter"
 version = "0.1.0"
@@ -658,13 +536,13 @@ dependencies = [
 
 [[package]]
 name = "remain"
-version = "0.2.5"
+version = "0.2.8"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1a81bc6a3aaeb180f767bd2bda8c03bac5b93b3b69a783c66f35d16a4df276cf"
+checksum = "e13cca257d068dd3a390d04b2c3009a3fad2ee5048dfa8f239d048372810470c"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn",
+ "syn 2.0.15",
 ]
 
 [[package]]
@@ -676,25 +554,11 @@ dependencies = [
  "semver",
 ]
 
-[[package]]
-name = "rustix"
-version = "0.36.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a3807b5d10909833d3e9acd1eb5fb988f79376ff10fce42937de71a449c4c588"
-dependencies = [
- "bitflags 1.3.2",
- "errno",
- "io-lifetimes",
- "libc",
- "linux-raw-sys",
- "windows-sys",
-]
-
 [[package]]
 name = "ryu"
-version = "1.0.11"
+version = "1.0.13"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4501abdff3ae82a1c1b477a17252eb69cee9e66eb915c1abaa4f44d873df9f09"
+checksum = "f91339c0467de62360649f8d3e185ca8de4224ff281f66000de5eb2a77a79041"
 
 [[package]]
 name = "seccompiler"
@@ -707,35 +571,35 @@ dependencies = [
 
 [[package]]
 name = "semver"
-version = "1.0.14"
+version = "1.0.17"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e25dfac463d778e353db5be2449d1cce89bd6fd23c9f1ea21310ce6e5a1b29c4"
+checksum = "bebd363326d05ec3e2f532ab7660680f3b02130d780c299bca73469d521bc0ed"
 
 [[package]]
 name = "serde"
-version = "1.0.150"
+version = "1.0.163"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e326c9ec8042f1b5da33252c8a37e9ffbd2c9bef0155215b6e6c80c790e05f91"
+checksum = "2113ab51b87a539ae008b5c6c02dc020ffa39afd2d83cffcb3f4eb2722cebec2"
 dependencies = [
  "serde_derive",
 ]
 
 [[package]]
 name = "serde_derive"
-version = "1.0.150"
+version = "1.0.163"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "42a3df25b0713732468deadad63ab9da1f1fd75a48a15024b50363f128db627e"
+checksum = "8c805777e3930c8883389c602315a24224bcc738b63905ef87cd1420353ea93e"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn",
+ "syn 2.0.15",
 ]
 
 [[package]]
 name = "serde_json"
-version = "1.0.89"
+version = "1.0.96"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "020ff22c755c2ed3f8cf162dbb41a7268d934702f3ed3631656ea597e08fc3db"
+checksum = "057d394a50403bcac12672b2b18fb387ab6d289d957dab67dd201875391e52f1"
 dependencies = [
  "itoa",
  "ryu",
@@ -744,9 +608,9 @@ dependencies = [
 
 [[package]]
 name = "serde_with"
-version = "2.1.0"
+version = "2.3.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "25bf4a5a814902cd1014dbccfa4d4560fb8432c779471e96e035602519f82eef"
+checksum = "07ff71d2c147a7b57362cead5e22f772cd52f6ab31cfcd9edcd7f6aeb2a0afbe"
 dependencies = [
  "serde",
  "serde_with_macros",
@@ -754,14 +618,14 @@ dependencies = [
 
 [[package]]
 name = "serde_with_macros"
-version = "2.1.0"
+version = "2.3.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e3452b4c0f6c1e357f73fdb87cd1efabaa12acf328c7a528e252893baeb3f4aa"
+checksum = "881b6f881b17d13214e5d494c939ebab463d01264ce1811e9d4ac3a882e7695f"
 dependencies = [
  "darling",
  "proc-macro2",
  "quote",
- "syn",
+ "syn 2.0.15",
 ]
 
 [[package]]
@@ -770,9 +634,9 @@ version = "0.1.0"
 
 [[package]]
 name = "signal-hook"
-version = "0.3.14"
+version = "0.3.15"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a253b5e89e2698464fc26b545c9edceb338e18a89effeeecfea192c3025be29d"
+checksum = "732768f1176d21d09e076c23a93123d40bba92d50c4058da34d45c8de8e682b9"
 dependencies = [
  "libc",
  "signal-hook-registry",
@@ -780,24 +644,18 @@ dependencies = [
 
 [[package]]
 name = "signal-hook-registry"
-version = "1.4.0"
+version = "1.4.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e51e73328dc4ac0c7ccbda3a494dfa03df1de2f46018127f60c693f2648455b0"
+checksum = "d8229b473baa5980ac72ef434c4415e70c4b5e71b423043adb4ba059f89c99a1"
 dependencies = [
  "libc",
 ]
 
 [[package]]
-name = "siphasher"
-version = "0.3.10"
+name = "smallvec"
+version = "1.10.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7bd3e3206899af3f8b12af284fafc038cc1dc2b41d1b89dd17297221c5d225de"
-
-[[package]]
-name = "static_assertions"
-version = "1.1.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a2eb9349b6444b326872e140eb1cf5e7c522154d69e7a0ffb0fb81c06b37543f"
+checksum = "a507befe795404456341dfab10cef66ead4c041f62b8b11bbb92bffe5d0953e0"
 
 [[package]]
 name = "strsim"
@@ -807,9 +665,9 @@ checksum = "73473c0e59e6d5812c5dfe2a064a6444949f089e20eec9a2e5506596494e4623"
 
 [[package]]
 name = "syn"
-version = "1.0.105"
+version = "1.0.109"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "60b9b43d45702de4c839cb9b51d9f529c5dd26a4aff255b42b1ebc03e88ee908"
+checksum = "72b64191b275b66ffe2469e8af2c1cfe3bafa67b529ead792a6d0160888b4237"
 dependencies = [
  "proc-macro2",
  "quote",
@@ -817,42 +675,34 @@ dependencies = [
 ]
 
 [[package]]
-name = "termcolor"
-version = "1.1.3"
+name = "syn"
+version = "2.0.15"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "bab24d30b911b2376f3a13cc2cd443142f0c81dda04c118693e35b3835757755"
+checksum = "a34fcf3e8b60f57e6a14301a2e916d323af98b0ea63c599441eec8558660c822"
 dependencies = [
- "winapi-util",
-]
-
-[[package]]
-name = "terminal_size"
-version = "0.2.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cb20089a8ba2b69debd491f8d2d023761cbf196e999218c591fa1e7e15a21907"
-dependencies = [
- "rustix",
- "windows-sys",
+ "proc-macro2",
+ "quote",
+ "unicode-ident",
 ]
 
 [[package]]
 name = "thiserror"
-version = "1.0.37"
+version = "1.0.40"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "10deb33631e3c9018b9baf9dcbbc4f737320d2b576bac10f6aefa048fa407e3e"
+checksum = "978c9a314bd8dc99be594bc3c175faaa9794be04a5a5e153caba6915336cebac"
 dependencies = [
  "thiserror-impl",
 ]
 
 [[package]]
 name = "thiserror-impl"
-version = "1.0.37"
+version = "1.0.40"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "982d17546b47146b28f7c22e3d08465f6b8903d0ea13c1660d9d84a6e7adcdbb"
+checksum = "f9456a42c5b0d803c8cd86e73dd7cc9edd429499f37a3550d286d5e86720569f"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn",
+ "syn 2.0.15",
 ]
 
 [[package]]
@@ -863,6 +713,7 @@ dependencies = [
  "byteorder",
  "libc",
  "log",
+ "net_gen",
  "thiserror",
  "vmm-sys-util",
 ]
@@ -880,24 +731,24 @@ dependencies = [
 
 [[package]]
 name = "unicode-ident"
-version = "1.0.5"
+version = "1.0.8"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6ceab39d59e4c9499d4e5a8ee0e2735b891bb7308ac83dfb4e80cad195c9f6f3"
+checksum = "e5464a87b239f13a63a501f2701565754bae92d243d4bb7eb12f6d57d2269bf4"
 
 [[package]]
 name = "uuid"
-version = "1.2.2"
+version = "1.3.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "422ee0de9031b5b948b97a8fc04e3aa35230001a722ddd27943e0be31564ce4c"
+checksum = "345444e32442451b267fc254ae85a209c64be56d2890e601a0c37ff0c3c5ecd2"
 dependencies = [
  "getrandom",
 ]
 
 [[package]]
 name = "versionize"
-version = "0.1.9"
+version = "0.1.10"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d6e2495726cf917e7ba7ec8bf0f0fceab543dd38d0a4195ed6bef331e38a290f"
+checksum = "dca4b7062e7e6d685901e815c35f9671e059de97c1c0905eeff8592f3fff442f"
 dependencies = [
  "bincode",
  "crc64",
@@ -905,7 +756,7 @@ dependencies = [
  "quote",
  "serde",
  "serde_derive",
- "syn",
+ "syn 1.0.109",
  "versionize_derive",
  "vmm-sys-util",
 ]
@@ -913,11 +764,11 @@ dependencies = [
 [[package]]
 name = "versionize_derive"
 version = "0.1.4"
-source = "git+https://github.com/cloud-hypervisor/versionize_derive?branch=ch#ae35ef7a3ddabd3371ab8ac0193a383aff6e4b1b"
+source = "git+https://github.com/cloud-hypervisor/versionize_derive?branch=ch#e502b1d4aabab342386f0c53780d49f21a6a1df6"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn",
+ "syn 1.0.109",
 ]
 
 [[package]]
@@ -947,7 +798,9 @@ dependencies = [
 [[package]]
 name = "vfio_user"
 version = "0.1.0"
+source = "git+https://github.com/rust-vmm/vfio-user?branch=main#afbbd5722885e961ce12baea12efe01d52ce14b0"
 dependencies = [
+ "bitflags",
  "libc",
  "log",
  "serde",
@@ -978,7 +831,7 @@ version = "0.6.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c9b791c5b0717a0558888a4cf7240cea836f39a99cb342e12ce633dcaa078072"
 dependencies = [
- "bitflags 1.3.2",
+ "bitflags",
  "libc",
  "vm-memory",
  "vmm-sys-util",
@@ -990,6 +843,12 @@ version = "0.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3ff512178285488516ed85f15b5d0113a7cdb89e9e8a760b269ae4f02b84bd6b"
 
+[[package]]
+name = "virtio-bindings"
+version = "0.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0b9084faf91b9aa9676ae2cac8f1432df2839d9566e6f19f29dbc13a8b831dff"
+
 [[package]]
 name = "virtio-devices"
 version = "0.1.0"
@@ -1015,7 +874,7 @@ dependencies = [
  "versionize",
  "versionize_derive",
  "vhost",
- "virtio-bindings",
+ "virtio-bindings 0.2.0",
  "virtio-queue",
  "vm-allocator",
  "vm-device",
@@ -1027,12 +886,12 @@ dependencies = [
 
 [[package]]
 name = "virtio-queue"
-version = "0.7.0"
+version = "0.7.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "19e927d93d54c365034fd7f31a5f458a1f540de4a37c52e892670dad9692173c"
+checksum = "3ba81e2bcc21c0d2fc5e6683e79367e26ad219197423a498df801d79d5ba77bd"
 dependencies = [
  "log",
- "virtio-bindings",
+ "virtio-bindings 0.1.0",
  "vm-memory",
  "vmm-sys-util",
 ]
@@ -1105,9 +964,8 @@ dependencies = [
  "anyhow",
  "arc-swap",
  "arch",
- "bitflags 2.4.1",
+ "bitflags",
  "block_util",
- "clap",
  "devices",
  "epoll",
  "event_monitor",
@@ -1142,15 +1000,16 @@ dependencies = [
  "vm-migration",
  "vm-virtio",
  "vmm-sys-util",
+ "zerocopy",
 ]
 
 [[package]]
 name = "vmm-sys-util"
-version = "0.11.0"
+version = "0.11.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cc06a16ee8ebf0d9269aed304030b0d20a866b8b3dd3d4ce532596ac567a0d24"
+checksum = "dd64fe09d8e880e600c324e7d664760a17f56e9672b7495a86381b49e4f72f46"
 dependencies = [
- "bitflags 1.3.2",
+ "bitflags",
  "libc",
  "serde",
  "serde_derive",
@@ -1178,15 +1037,6 @@ version = "0.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ac3b87c63620426dd9b991e5ce0329eff545bccbbb34f3be09ff6fb6ab51b7b6"
 
-[[package]]
-name = "winapi-util"
-version = "0.1.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "70ec6ce85bb158151cae5e5c87f95a8e97d2c0c4b001223f33a334e3ce5de178"
-dependencies = [
- "winapi",
-]
-
 [[package]]
 name = "winapi-x86_64-pc-windows-gnu"
 version = "0.4.0"
@@ -1194,58 +1044,22 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "712e227841d057c1ee1cd2fb22fa7e5a5461ae8e48fa2ca79ec42cfc1931183f"
 
 [[package]]
-name = "windows-sys"
-version = "0.42.0"
+name = "zerocopy"
+version = "0.6.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5a3e1820f08b8513f676f7ab6c1f99ff312fb97b553d30ff4dd86f9f15728aa7"
+checksum = "332f188cc1bcf1fe1064b8c58d150f497e697f49774aa846f2dc949d9a25f236"
 dependencies = [
- "windows_aarch64_gnullvm",
- "windows_aarch64_msvc",
- "windows_i686_gnu",
- "windows_i686_msvc",
- "windows_x86_64_gnu",
- "windows_x86_64_gnullvm",
- "windows_x86_64_msvc",
+ "byteorder",
+ "zerocopy-derive",
 ]
 
 [[package]]
-name = "windows_aarch64_gnullvm"
-version = "0.42.0"
+name = "zerocopy-derive"
+version = "0.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "41d2aa71f6f0cbe00ae5167d90ef3cfe66527d6f613ca78ac8024c3ccab9a19e"
-
-[[package]]
-name = "windows_aarch64_msvc"
-version = "0.42.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "dd0f252f5a35cac83d6311b2e795981f5ee6e67eb1f9a7f64eb4500fbc4dcdb4"
-
-[[package]]
-name = "windows_i686_gnu"
-version = "0.42.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "fbeae19f6716841636c28d695375df17562ca208b2b7d0dc47635a50ae6c5de7"
-
-[[package]]
-name = "windows_i686_msvc"
-version = "0.42.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "84c12f65daa39dd2babe6e442988fc329d6243fdce47d7d2d155b8d874862246"
-
-[[package]]
-name = "windows_x86_64_gnu"
-version = "0.42.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "bf7b1b21b5362cbc318f686150e5bcea75ecedc74dd157d874d754a2ca44b0ed"
-
-[[package]]
-name = "windows_x86_64_gnullvm"
-version = "0.42.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "09d525d2ba30eeb3297665bd434a54297e4170c7f1a44cad4ef58095b4cd2028"
-
-[[package]]
-name = "windows_x86_64_msvc"
-version = "0.42.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f40009d85759725a34da6d89a94e63d7bdc50a862acf0dbc7c8e488f1edcb6f5"
+checksum = "6505e6815af7de1746a08f69c69606bb45695a17149517680f3b2149713b19a3"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 1.0.109",
+]

fuzz/Cargo.toml

@@ -12,17 +12,19 @@ cargo-fuzz = true
 block_util = { path = "../block_util" }
 devices = { path = "../devices" }
 epoll = "4.3.1"
-libc = "0.2.135"
-libfuzzer-sys = "0.4.5"
+libc = "0.2.144"
+libfuzzer-sys = "0.4.6"
+linux-loader = { version = "0.8.1", features = ["elf", "bzimage", "pe"] }
 micro_http = { git = "https://github.com/firecracker-microvm/micro-http", branch = "main" }
-once_cell = "1.16.0"
+net_util = { path = "../net_util" }
+once_cell = "1.17.1"
 qcow = { path = "../qcow" }
 seccompiler = "0.3.0"
 vhdx = { path = "../vhdx" }
 virtio-devices = { path = "../virtio-devices" }
-virtio-queue = "0.7.0"
+virtio-queue = "0.7.1"
 vmm = { path = "../vmm" }
-vmm-sys-util = "0.11.0"
+vmm-sys-util = "0.11.1"
 vm-memory = "0.10.0"
 vm-device = { path = "../vm-device" }
 vm-virtio = { path = "../vm-virtio" }
@@ -74,12 +76,30 @@ path = "fuzz_targets/iommu.rs"
 test = false
 doc = false
 
+[[bin]]
+name = "linux_loader"
+path = "fuzz_targets/linux_loader.rs"
+test = false
+doc = false
+
+[[bin]]
+name = "linux_loader_cmdline"
+path = "fuzz_targets/linux_loader_cmdline.rs"
+test = false
+doc = false
+
 [[bin]]
 name = "mem"
 path = "fuzz_targets/mem.rs"
 test = false
 doc = false
 
+[[bin]]
+name = "net"
+path = "fuzz_targets/net.rs"
+test = false
+doc = false
+
 [[bin]]
 name = "pmem"
 path = "fuzz_targets/pmem.rs"

fuzz/fuzz_targets/linux_loader.rs

@@ -0,0 +1,50 @@
+// Copyright 2018 The Chromium OS Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+//
+// Copyright © 2022 Intel Corporation
+//
+// SPDX-License-Identifier: Apache-2.0 AND BSD-3-Clause
+
+#![no_main]
+
+use libfuzzer_sys::fuzz_target;
+use linux_loader::loader::KernelLoader;
+use std::ffi;
+use std::fs::File;
+use std::io;
+use std::io::{Seek, SeekFrom, Write};
+use std::os::unix::io::{FromRawFd, RawFd};
+use vm_memory::{bitmap::AtomicBitmap, GuestAddress};
+
+type GuestMemoryMmap = vm_memory::GuestMemoryMmap<AtomicBitmap>;
+
+const MEM_SIZE: usize = 256 * 1024 * 1024;
+// From 'arch::x86_64::layout::HIGH_RAM_START'
+const HIGH_RAM_START: GuestAddress = GuestAddress(0x100000);
+
+fuzz_target!(|bytes| {
+    let shm = memfd_create(&ffi::CString::new("fuzz_load_kernel").unwrap(), 0).unwrap();
+    let mut kernel_file: File = unsafe { File::from_raw_fd(shm) };
+    kernel_file.write_all(&bytes).unwrap();
+    kernel_file.seek(SeekFrom::Start(0)).unwrap();
+
+    let guest_memory = GuestMemoryMmap::from_ranges(&[(GuestAddress(0), MEM_SIZE)]).unwrap();
+    linux_loader::loader::elf::Elf::load(
+        &guest_memory,
+        None,
+        &mut kernel_file,
+        Some(HIGH_RAM_START),
+    )
+    .ok();
+});
+
+fn memfd_create(name: &ffi::CStr, flags: u32) -> Result<RawFd, io::Error> {
+    let res = unsafe { libc::syscall(libc::SYS_memfd_create, name.as_ptr(), flags) };
+
+    if res < 0 {
+        Err(io::Error::last_os_error())
+    } else {
+        Ok(res as RawFd)
+    }
+}

fuzz/fuzz_targets/linux_loader_cmdline.rs

@@ -0,0 +1,34 @@
+// Copyright 2018 The Chromium OS Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+//
+// Copyright © 2022 Intel Corporation
+//
+// SPDX-License-Identifier: Apache-2.0 AND BSD-3-Clause
+
+#![no_main]
+
+use libfuzzer_sys::fuzz_target;
+use vm_memory::{bitmap::AtomicBitmap, GuestAddress};
+
+type GuestMemoryMmap = vm_memory::GuestMemoryMmap<AtomicBitmap>;
+
+const MEM_SIZE: usize = 256 * 1024 * 1024;
+// From 'arch::x86_64::layout::CMDLINE_START'
+const CMDLINE_START: GuestAddress = GuestAddress(0x20000);
+
+fuzz_target!(|bytes| {
+    let payload_config = vmm::config::PayloadConfig {
+        firmware: None,
+        kernel: None,
+        cmdline: Some(String::from_utf8_lossy(&bytes).to_string()),
+        initramfs: None,
+    };
+    let kernel_cmdline = match vmm::vm::Vm::generate_cmdline(&payload_config) {
+        Ok(cmdline) => cmdline,
+        _ => return,
+    };
+    let guest_memory = GuestMemoryMmap::from_ranges(&[(GuestAddress(0), MEM_SIZE)]).unwrap();
+
+    linux_loader::loader::load_cmdline(&guest_memory, CMDLINE_START, &kernel_cmdline).ok();
+});

fuzz/fuzz_targets/net.rs

@@ -0,0 +1,285 @@
+// Copyright © 2022 Intel Corporation
+//
+// SPDX-License-Identifier: Apache-2.0
+
+#![no_main]
+
+use libfuzzer_sys::fuzz_target;
+use seccompiler::SeccompAction;
+use std::fs::File;
+use std::io::{Read, Write};
+use std::os::unix::io::{AsRawFd, FromRawFd};
+use std::sync::Arc;
+use virtio_devices::{VirtioDevice, VirtioInterrupt, VirtioInterruptType};
+use virtio_queue::{Queue, QueueT};
+use vm_memory::{bitmap::AtomicBitmap, Bytes, GuestAddress, GuestMemoryAtomic};
+use vmm::EpollContext;
+use vmm_sys_util::eventfd::{EventFd, EFD_NONBLOCK};
+
+type GuestMemoryMmap = vm_memory::GuestMemoryMmap<AtomicBitmap>;
+
+macro_rules! align {
+    ($n:expr, $align:expr) => {{
+        (($n + $align - 1) / $align) * $align
+    }};
+}
+
+const TAP_INPUT_SIZE: usize = 128;
+const QUEUE_DATA_SIZE: usize = 4;
+const MEM_SIZE: usize = 32 * 1024 * 1024;
+// Guest memory gap
+const GUEST_MEM_GAP: u64 = 1 * 1024 * 1024;
+// Guest physical address for the first virt queue
+const BASE_VIRT_QUEUE_ADDR: u64 = MEM_SIZE as u64 + GUEST_MEM_GAP;
+// Number of queues
+const QUEUE_NUM: usize = 2;
+// Max entries in the queue.
+const QUEUE_SIZE: u16 = 256;
+// Descriptor table alignment
+const DESC_TABLE_ALIGN_SIZE: u64 = 16;
+// Used ring alignment
+const USED_RING_ALIGN_SIZE: u64 = 4;
+// Descriptor table size
+const DESC_TABLE_SIZE: u64 = 16_u64 * QUEUE_SIZE as u64;
+// Available ring size
+const AVAIL_RING_SIZE: u64 = 6_u64 + 2 * QUEUE_SIZE as u64;
+// Padding size before used ring
+const PADDING_SIZE: u64 = align!(AVAIL_RING_SIZE, USED_RING_ALIGN_SIZE) - AVAIL_RING_SIZE;
+// Used ring size
+const USED_RING_SIZE: u64 = 6_u64 + 8 * QUEUE_SIZE as u64;
+// Virtio-queue size in bytes
+const QUEUE_BYTES_SIZE: usize = align!(
+    DESC_TABLE_SIZE + AVAIL_RING_SIZE + PADDING_SIZE + USED_RING_SIZE,
+    DESC_TABLE_ALIGN_SIZE
+) as usize;
+
+fuzz_target!(|bytes| {
+    if bytes.len() < TAP_INPUT_SIZE + (QUEUE_DATA_SIZE + QUEUE_BYTES_SIZE) * QUEUE_NUM
+        || bytes.len()
+            > TAP_INPUT_SIZE + (QUEUE_DATA_SIZE + QUEUE_BYTES_SIZE) * QUEUE_NUM + MEM_SIZE
+    {
+        return;
+    }
+
+    let (dummy_tap_frontend, dummy_tap_backend) = create_socketpair().unwrap();
+    let if_name = "fuzzer_tap_name".as_bytes().to_vec();
+    let tap = net_util::Tap::new_for_fuzzing(dummy_tap_frontend, if_name);
+
+    let mut net = virtio_devices::Net::new_with_tap(
+        "fuzzer_net".to_owned(),
+        vec![tap],
+        None,  // guest_mac
+        false, // iommu
+        QUEUE_NUM,
+        QUEUE_SIZE,
+        SeccompAction::Allow,
+        None,
+        EventFd::new(EFD_NONBLOCK).unwrap(),
+        None,
+        true,
+        true,
+        true,
+    )
+    .unwrap();
+
+    let tap_input_bytes = &bytes[..TAP_INPUT_SIZE];
+    let queue_data = &bytes[TAP_INPUT_SIZE..TAP_INPUT_SIZE + QUEUE_DATA_SIZE * QUEUE_NUM];
+    let queue_bytes = &bytes[TAP_INPUT_SIZE + QUEUE_DATA_SIZE * QUEUE_NUM
+        ..TAP_INPUT_SIZE + (QUEUE_DATA_SIZE + QUEUE_BYTES_SIZE) * QUEUE_NUM];
+    let mem_bytes = &bytes[TAP_INPUT_SIZE + (QUEUE_DATA_SIZE + QUEUE_BYTES_SIZE) * QUEUE_NUM..];
+
+    // Setup the virt queues with the input bytes
+    let mut queues = setup_virt_queues(
+        &[
+            &queue_data[..QUEUE_DATA_SIZE].try_into().unwrap(),
+            &queue_data[QUEUE_DATA_SIZE..QUEUE_DATA_SIZE * 2]
+                .try_into()
+                .unwrap(),
+        ],
+        BASE_VIRT_QUEUE_ADDR,
+    );
+
+    // Setup the guest memory with the input bytes
+    let mem = GuestMemoryMmap::from_ranges(&[
+        (GuestAddress(0), MEM_SIZE),
+        (GuestAddress(BASE_VIRT_QUEUE_ADDR), queue_bytes.len()),
+    ])
+    .unwrap();
+    if mem
+        .write_slice(queue_bytes, GuestAddress(BASE_VIRT_QUEUE_ADDR))
+        .is_err()
+    {
+        return;
+    }
+    if mem.write_slice(mem_bytes, GuestAddress(0 as u64)).is_err() {
+        return;
+    }
+    let guest_memory = GuestMemoryAtomic::new(mem);
+
+    let input_queue = queues.remove(0);
+    let input_evt = EventFd::new(0).unwrap();
+    let input_queue_evt = unsafe { EventFd::from_raw_fd(libc::dup(input_evt.as_raw_fd())) };
+    let output_queue = queues.remove(0);
+    let output_evt = EventFd::new(0).unwrap();
+    let output_queue_evt = unsafe { EventFd::from_raw_fd(libc::dup(output_evt.as_raw_fd())) };
+
+    // Start the thread of dummy tap backend to handle the rx and tx from the virtio-net
+    let exit_evt = EventFd::new(libc::EFD_NONBLOCK).unwrap();
+    let tap_backend_thread = {
+        let dummy_tap_backend = dummy_tap_backend.try_clone().unwrap();
+        let tap_input_bytes: [u8; TAP_INPUT_SIZE] = tap_input_bytes[..].try_into().unwrap();
+        let exit_evt = exit_evt.try_clone().unwrap();
+        std::thread::Builder::new()
+            .name("dummy_tap_backend".to_string())
+            .spawn(move || {
+                tap_backend_stub(dummy_tap_backend, &tap_input_bytes, exit_evt);
+            })
+            .unwrap()
+    };
+
+    // Kick the 'queue' events and endpoint event before activate the net device
+    input_queue_evt.write(1).unwrap();
+    output_queue_evt.write(1).unwrap();
+
+    net.activate(
+        guest_memory,
+        Arc::new(NoopVirtioInterrupt {}),
+        vec![(0, input_queue, input_evt), (1, output_queue, output_evt)],
+    )
+    .unwrap();
+
+    // Wait for the events to finish and net device worker thread to return
+    net.wait_for_epoll_threads();
+    // Terminate the thread for the dummy tap backend
+    exit_evt.write(1).ok();
+    tap_backend_thread.join().unwrap();
+});
+
+pub struct NoopVirtioInterrupt {}
+
+impl VirtioInterrupt for NoopVirtioInterrupt {
+    fn trigger(&self, _int_type: VirtioInterruptType) -> std::result::Result<(), std::io::Error> {
+        Ok(())
+    }
+}
+
+fn setup_virt_queues(bytes: &[&[u8; QUEUE_DATA_SIZE]], base_addr: u64) -> Vec<Queue> {
+    let mut queues = Vec::new();
+    for (i, b) in bytes.iter().enumerate() {
+        let mut q = Queue::new(QUEUE_SIZE).unwrap();
+
+        let desc_table_addr = base_addr + (QUEUE_BYTES_SIZE * i) as u64;
+        let avail_ring_addr = desc_table_addr + DESC_TABLE_SIZE;
+        let used_ring_addr = avail_ring_addr + PADDING_SIZE + AVAIL_RING_SIZE;
+        q.try_set_desc_table_address(GuestAddress(desc_table_addr))
+            .unwrap();
+        q.try_set_avail_ring_address(GuestAddress(avail_ring_addr))
+            .unwrap();
+        q.try_set_used_ring_address(GuestAddress(used_ring_addr))
+            .unwrap();
+
+        q.set_next_avail(b[0] as u16); // 'u8' is enough given the 'QUEUE_SIZE' is small
+        q.set_next_used(b[1] as u16);
+        q.set_event_idx(b[2] % 2 != 0);
+        q.set_size(b[3] as u16 % QUEUE_SIZE);
+
+        q.set_ready(true);
+        queues.push(q);
+    }
+
+    queues
+}
+
+fn create_socketpair() -> Result<(File, File), std::io::Error> {
+    let mut fds = [-1, -1];
+    unsafe {
+        let ret = libc::socketpair(
+            libc::AF_UNIX,
+            libc::SOCK_STREAM | libc::SOCK_NONBLOCK,
+            0,
+            fds.as_mut_ptr(),
+        );
+        if ret == -1 {
+            return Err(std::io::Error::last_os_error());
+        }
+    }
+
+    let socket1 = unsafe { File::from_raw_fd(fds[0]) };
+    let socket2 = unsafe { File::from_raw_fd(fds[1]) };
+    Ok((socket1, socket2))
+}
+
+enum EpollEvent {
+    Exit = 0,
+    Rx = 1,
+    Tx = 2,
+    Unknown,
+}
+
+impl From<u64> for EpollEvent {
+    fn from(v: u64) -> Self {
+        use EpollEvent::*;
+        match v {
+            0 => Exit,
+            1 => Rx,
+            2 => Tx,
+            _ => Unknown,
+        }
+    }
+}
+
+// Handle the rx and tx requests from the virtio-net device
+fn tap_backend_stub(
+    mut dummy_tap: File,
+    tap_input_bytes: &[u8; TAP_INPUT_SIZE],
+    exit_evt: EventFd,
+) {
+    let mut epoll = EpollContext::new().unwrap();
+    epoll
+        .add_event_custom(&exit_evt, EpollEvent::Exit as u64, epoll::Events::EPOLLIN)
+        .unwrap();
+    let dummy_tap_write = dummy_tap.try_clone().unwrap();
+    epoll
+        .add_event_custom(
+            &dummy_tap_write,
+            EpollEvent::Rx as u64,
+            epoll::Events::EPOLLOUT,
+        )
+        .unwrap();
+    epoll
+        .add_event_custom(&dummy_tap, EpollEvent::Tx as u64, epoll::Events::EPOLLIN)
+        .unwrap();
+
+    let epoll_fd = epoll.as_raw_fd();
+    let mut events = vec![epoll::Event::new(epoll::Events::empty(), 0); 3];
+    loop {
+        let num_events = match epoll::wait(epoll_fd, -1, &mut events[..]) {
+            Ok(num_events) => num_events,
+            Err(e) => match e.raw_os_error() {
+                Some(libc::EAGAIN) | Some(libc::EINTR) => continue,
+                _ => panic!("Unexpected epoll::wait error!"),
+            },
+        };
+
+        for event in events.iter().take(num_events) {
+            let dispatch_event: EpollEvent = event.data.into();
+            match dispatch_event {
+                EpollEvent::Exit => {
+                    return;
+                }
+                EpollEvent::Rx => {
+                    dummy_tap.write_all(tap_input_bytes).unwrap();
+                    break;
+                }
+                EpollEvent::Tx => {
+                    let mut buffer = Vec::new();
+                    dummy_tap.read_to_end(&mut buffer).ok();
+                    break;
+                }
+                _ => {
+                    panic!("Unexpected Epoll event");
+                }
+            }
+        }
+    }
+}

fuzz/fuzz_targets/serial.rs

@@ -15,6 +15,7 @@ fuzz_target!(|bytes| {
     let mut serial = Serial::new_sink(
         "serial".into(),
         Arc::new(TestInterrupt::new(EventFd::new(EFD_NONBLOCK).unwrap())),
+        None,
     );
 
     let mut i = 0;

hypervisor/Cargo.toml

@@ -6,28 +6,28 @@ edition = "2021"
 license = "Apache-2.0 OR BSD-3-Clause"
 
 [features]
-kvm = ["kvm-ioctls", "kvm-bindings"]
-mshv = ["mshv-ioctls", "mshv-bindings"]
+kvm = ["kvm-ioctls", "kvm-bindings", "vfio-ioctls/kvm"]
+mshv = ["mshv-ioctls", "mshv-bindings", "vfio-ioctls/mshv"]
 tdx = []
 
 [dependencies]
-anyhow = "1.0.66"
+anyhow = "1.0.70"
 byteorder = "1.4.3"
-thiserror = "1.0.37"
-libc = "0.2.138"
+thiserror = "1.0.39"
+libc = "0.2.139"
 log = "0.4.17"
 kvm-ioctls = { version = "0.13.0", optional = true }
 kvm-bindings = { git = "https://github.com/cloud-hypervisor/kvm-bindings", branch = "ch-v0.6.0-tdx", features = ["with-serde", "fam-wrappers"], optional  = true }
 mshv-bindings = { git = "https://github.com/rust-vmm/mshv", branch = "main", features = ["with-serde", "fam-wrappers"], optional  = true }
 mshv-ioctls = { git = "https://github.com/rust-vmm/mshv", branch = "main", optional  = true}
-serde = { version = "1.0.150", features = ["rc", "derive"] }
-serde_with = { version = "2.1.0", default-features = false, features = ["macros"] }
+serde = { version = "1.0.156", features = ["rc", "derive"] }
+serde_with = { version = "2.3.2", default-features = false, features = ["macros"] }
 vfio-ioctls = { git = "https://github.com/rust-vmm/vfio", branch = "main", default-features = false }
 vm-memory = { version = "0.10.0", features = ["backend-mmap", "backend-atomic"] }
 vmm-sys-util = { version = "0.11.0", features = ["with-serde"] }
 
 [target.'cfg(target_arch = "x86_64")'.dependencies.iced-x86]
-version = "1.17.0"
+version = "1.18.0"
 default-features = false
 features = ["std", "decoder", "op_code_info", "instr_info", "fast_fmt"]
 

hypervisor/src/arch/x86/emulator/instructions/mod.rs

@@ -18,8 +18,8 @@ fn get_op<T: CpuStateManager>(
     insn: &Instruction,
     op_index: u32,
     op_size: usize,
-    state: &T,
-    platform: &dyn PlatformEmulator<CpuState = T>,
+    state: &mut T,
+    platform: &mut dyn PlatformEmulator<CpuState = T>,
 ) -> Result<u64, PlatformError> {
     if insn.op_count() < op_index + 1 {
         return Err(PlatformError::InvalidOperand(anyhow!(

hypervisor/src/arch/x86/emulator/instructions/movs.rs

@@ -15,73 +15,89 @@ use crate::arch::x86::emulator::instructions::*;
 use crate::arch::x86::regs::DF;
 use crate::arch::x86::Exception;
 
+macro_rules! movs {
+    ($bound:ty) => {
+        fn emulate(
+            &self,
+            insn: &Instruction,
+            state: &mut T,
+            platform: &mut dyn PlatformEmulator<CpuState = T>,
+        ) -> Result<(), EmulationError<Exception>> {
+            let mut count: u64 = if insn.has_rep_prefix() {
+                state
+                    .read_reg(Register::ECX)
+                    .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?
+            } else {
+                1
+            };
+
+            let mut rsi = state
+                .read_reg(Register::RSI)
+                .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
+            let mut rdi = state
+                .read_reg(Register::RDI)
+                .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
+
+            let df = (state.flags() & DF) != 0;
+            let len = std::mem::size_of::<$bound>();
+
+            while count > 0 {
+                let mut memory: [u8; 4] = [0; 4];
+
+                let src = state
+                    .linearize(Register::DS, rsi, false)
+                    .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
+                let dst = state
+                    .linearize(Register::ES, rdi, true)
+                    .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
+
+                platform
+                    .read_memory(src, &mut memory[0..len])
+                    .map_err(EmulationError::PlatformEmulationError)?;
+                platform
+                    .write_memory(dst, &memory[0..len])
+                    .map_err(EmulationError::PlatformEmulationError)?;
+
+                if df {
+                    rsi = rsi.wrapping_sub(len as u64);
+                    rdi = rdi.wrapping_sub(len as u64);
+                } else {
+                    rsi = rsi.wrapping_add(len as u64);
+                    rdi = rdi.wrapping_add(len as u64);
+                }
+                count -= 1;
+            }
+
+            state
+                .write_reg(Register::RSI, rsi)
+                .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
+            state
+                .write_reg(Register::RDI, rdi)
+                .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
+            if insn.has_rep_prefix() {
+                state
+                    .write_reg(Register::ECX, 0)
+                    .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
+            }
+
+            Ok(())
+        }
+    };
+}
+
 pub struct Movsd_m32_m32;
 impl<T: CpuStateManager> InstructionHandler<T> for Movsd_m32_m32 {
-    fn emulate(
-        &self,
-        insn: &Instruction,
-        state: &mut T,
-        platform: &mut dyn PlatformEmulator<CpuState = T>,
-    ) -> Result<(), EmulationError<Exception>> {
-        let mut count: u64 = if insn.has_rep_prefix() {
-            state
-                .read_reg(Register::ECX)
-                .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?
-        } else {
-            1
-        };
+    movs!(u32);
+}
 
-        let mut rsi = state
-            .read_reg(Register::RSI)
-            .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
-        let mut rdi = state
-            .read_reg(Register::RDI)
-            .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
+pub struct Movsw_m16_m16;
+impl<T: CpuStateManager> InstructionHandler<T> for Movsw_m16_m16 {
+    movs!(u16);
+}
 
-        let df = (state.flags() & DF) != 0;
-        let len = std::mem::size_of::<u32>();
-
-        while count > 0 {
-            let mut memory: [u8; 4] = [0; 4];
-
-            let src = state
-                .linearize(Register::DS, rsi, false)
-                .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
-            let dst = state
-                .linearize(Register::ES, rdi, true)
-                .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
-
-            platform
-                .read_memory(src, &mut memory[0..len])
-                .map_err(EmulationError::PlatformEmulationError)?;
-            platform
-                .write_memory(dst, &memory[0..len])
-                .map_err(EmulationError::PlatformEmulationError)?;
-
-            if df {
-                rsi = rsi.wrapping_sub(len as u64);
-                rdi = rdi.wrapping_sub(len as u64);
-            } else {
-                rsi = rsi.wrapping_add(len as u64);
-                rdi = rdi.wrapping_add(len as u64);
-            }
-            count -= 1;
-        }
-
-        state
-            .write_reg(Register::RSI, rsi)
-            .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
-        state
-            .write_reg(Register::RDI, rdi)
-            .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
-        if insn.has_rep_prefix() {
-            state
-                .write_reg(Register::ECX, 0)
-                .map_err(|e| EmulationError::InvalidOperand(anyhow!(e)))?;
-        }
-
-        Ok(())
-    }
+pub struct Movsb_m8_m8;
+impl<T: CpuStateManager> InstructionHandler<T> for Movsb_m8_m8 {
+    movs!(u8);
 }
 
 #[cfg(test)]
@@ -142,4 +158,127 @@ mod tests {
         vmm.read_memory(0x8 + 8, &mut data).unwrap();
         assert_eq!(0x0, <u32>::from_le_bytes(data));
     }
+
+    #[test]
+    fn test_rep_movsw_m16_m16() {
+        let ip: u64 = 0x1000;
+        let memory: [u8; 24] = [
+            0x78, 0x56, 0x34, 0x12, // 0x12345678
+            0xdd, 0xcc, 0xbb, 0xaa, // 0xaabbccdd
+            0xa5, 0x5a, 0xa5, 0x5a, // 0x5aa55aa5
+            0x00, 0x00, 0x00, 0x00, // 0x00000000
+            0x00, 0x00, 0x00, 0x00, // 0x00000000
+            0x00, 0x00, 0x00, 0x00, // 0x00000000
+        ];
+        let insn = [0x66, 0xf3, 0xa5]; // rep movsw
+        let regs = vec![(Register::ECX, 6), (Register::ESI, 0), (Register::EDI, 0xc)];
+        let mut data = [0u8; 2];
+
+        let mut vmm = MockVmm::new(ip, regs, Some((0, &memory)));
+
+        assert!(vmm.emulate_first_insn(0, &insn).is_ok());
+
+        vmm.read_memory(0xc, &mut data).unwrap();
+        assert_eq!(0x5678, <u16>::from_le_bytes(data));
+        vmm.read_memory(0xc + 2, &mut data).unwrap();
+        assert_eq!(0x1234, <u16>::from_le_bytes(data));
+        vmm.read_memory(0xc + 4, &mut data).unwrap();
+        assert_eq!(0xccdd, <u16>::from_le_bytes(data));
+        vmm.read_memory(0xc + 6, &mut data).unwrap();
+        assert_eq!(0xaabb, <u16>::from_le_bytes(data));
+        vmm.read_memory(0xc + 8, &mut data).unwrap();
+        assert_eq!(0x5aa5, <u16>::from_le_bytes(data));
+        vmm.read_memory(0xc + 10, &mut data).unwrap();
+        assert_eq!(0x5aa5, <u16>::from_le_bytes(data));
+        // The rest should be default value 0 from MockVmm
+        vmm.read_memory(0xc + 12, &mut data).unwrap();
+        assert_eq!(0x0, <u16>::from_le_bytes(data));
+    }
+
+    #[test]
+    fn test_movsw_m16_m16() {
+        let ip: u64 = 0x1000;
+        let memory: [u8; 4] = [
+            0x78, 0x56, 0x34, 0x12, // 0x12345678
+        ];
+        let insn = [0x66, 0xa5]; // movsw
+        let regs = vec![(Register::ESI, 0), (Register::EDI, 0x8)];
+        let mut data = [0u8; 2];
+
+        let mut vmm = MockVmm::new(ip, regs, Some((0, &memory)));
+
+        assert!(vmm.emulate_first_insn(0, &insn).is_ok());
+
+        vmm.read_memory(0x8, &mut data).unwrap();
+        assert_eq!(0x5678, <u16>::from_le_bytes(data));
+        // Only two bytes were copied, so the value at 0xa should be zero
+        vmm.read_memory(0xa, &mut data).unwrap();
+        assert_eq!(0x0, <u16>::from_le_bytes(data));
+        // The rest should be default value 0 from MockVmm
+        vmm.read_memory(0x4, &mut data).unwrap();
+        assert_eq!(0x0, <u16>::from_le_bytes(data));
+        vmm.read_memory(0x8 + 8, &mut data).unwrap();
+        assert_eq!(0x0, <u16>::from_le_bytes(data));
+    }
+
+    #[test]
+    fn test_movsb_m8_m8() {
+        let ip: u64 = 0x1000;
+        let memory: [u8; 4] = [
+            0x78, 0x56, 0x34, 0x12, // 0x12345678
+        ];
+        let insn = [0x66, 0xa4]; // movsb
+        let regs = vec![(Register::ESI, 0), (Register::EDI, 0x8)];
+        let mut data = [0u8; 1];
+
+        let mut vmm = MockVmm::new(ip, regs, Some((0, &memory)));
+
+        assert!(vmm.emulate_first_insn(0, &insn).is_ok());
+
+        vmm.read_memory(0x8, &mut data).unwrap();
+        assert_eq!(0x78, data[0]);
+        // Only one byte was copied, so the value at 0x9 should be zero
+        vmm.read_memory(0x9, &mut data).unwrap();
+        assert_eq!(0x0, data[0]);
+        // The rest should be default value 0 from MockVmm
+        vmm.read_memory(0x4, &mut data).unwrap();
+        assert_eq!(0x0, data[0]);
+        // the src value is left as is after movb
+        vmm.read_memory(0x0, &mut data).unwrap();
+        assert_eq!(0x78, data[0]);
+    }
+
+    #[test]
+    fn test_rep_movsb_m8_m8() {
+        let ip: u64 = 0x1000;
+        let memory: [u8; 16] = [
+            0x78, 0x56, 0x34, 0x12, // 0x12345678
+            0xbb, 0xaa, 0x00, 0x00, // 0x0000aabb
+            0x00, 0x00, 0x00, 0x00, // 0x00000000
+            0x00, 0x00, 0x00, 0x00, // 0x00000000
+        ];
+        let insn = [0x66, 0xf3, 0xa4]; // rep movsw
+        let regs = vec![(Register::ECX, 6), (Register::ESI, 0), (Register::EDI, 0x8)];
+        let mut data = [0u8; 1];
+
+        let mut vmm = MockVmm::new(ip, regs, Some((0, &memory)));
+
+        assert!(vmm.emulate_first_insn(0, &insn).is_ok());
+
+        vmm.read_memory(0x8, &mut data).unwrap();
+        assert_eq!(0x78, data[0]);
+        vmm.read_memory(0x8 + 1, &mut data).unwrap();
+        assert_eq!(0x56, data[0]);
+        vmm.read_memory(0x8 + 2, &mut data).unwrap();
+        assert_eq!(0x34, data[0]);
+        vmm.read_memory(0x8 + 3, &mut data).unwrap();
+        assert_eq!(0x12, data[0]);
+        vmm.read_memory(0x8 + 4, &mut data).unwrap();
+        assert_eq!(0xbb, data[0]);
+        vmm.read_memory(0x8 + 5, &mut data).unwrap();
+        assert_eq!(0xaa, data[0]);
+        // The rest should be default value 0 from MockVmm
+        vmm.read_memory(0x8 + 6, &mut data).unwrap();
+        assert_eq!(0x0, data[0]);
+    }
 }

hypervisor/src/arch/x86/emulator/mod.rs

@@ -526,6 +526,8 @@ impl<'a, T: CpuStateManager> Emulator<'a, T> {
             (mov, Movzx_r64_rm16),
             // MOVS
             (movs, Movsd_m32_m32),
+            (movs, Movsw_m16_m16),
+            (movs, Movsb_m8_m8),
             // OR
             (or, Or_rm8_r8)
         );

hypervisor/src/arch/x86/mod.rs

@@ -26,7 +26,6 @@ pub const MTRR_MEM_TYPE_WB: u64 = 0x6;
 pub const NUM_IOAPIC_PINS: usize = 24;
 
 // X86 Exceptions
-#[allow(clippy::upper_case_acronyms)]
 #[derive(Clone, Debug)]
 pub enum Exception {
     DE = 0,  // Divide Error
@@ -272,6 +271,7 @@ impl LapicState {
         use std::io::Cursor;
         use std::mem;
 
+        // SAFETY: plain old data type
         let sliceu8 = unsafe {
             // This array is only accessed as parts of a u32 word, so interpret it as a u8 array.
             // Cursors are only readable on arrays of u8, not i8(c_char).
@@ -290,6 +290,7 @@ impl LapicState {
         use std::io::Cursor;
         use std::mem;
 
+        // SAFETY: plain old data type
         let sliceu8 = unsafe {
             // This array is only accessed as parts of a u32 word, so interpret it as a u8 array.
             // Cursors are only readable on arrays of u8, not i8(c_char).

hypervisor/src/hypervisor.rs

@@ -108,7 +108,7 @@ pub trait Hypervisor: Send + Sync {
     ///
     /// Get the supported CpuID
     ///
-    fn get_cpuid(&self) -> Result<Vec<CpuIdEntry>>;
+    fn get_supported_cpuid(&self) -> Result<Vec<CpuIdEntry>>;
     ///
     /// Check particular extensions if any
     ///
@@ -133,4 +133,7 @@ pub trait Hypervisor: Send + Sync {
     fn get_guest_debug_hw_bps(&self) -> usize {
         unimplemented!()
     }
+
+    /// Get maximum number of vCPUs
+    fn get_max_vcpus(&self) -> u32;
 }

hypervisor/src/kvm/aarch64/gic/mod.rs

@@ -243,7 +243,6 @@ impl KvmGicV3Its {
     }
 
     /// Method to initialize the GIC device
-    #[allow(clippy::new_ret_no_self)]
     pub fn new(vm: &dyn Vm, config: VgicConfig) -> Result<KvmGicV3Its> {
         // This is inside KVM module
         let vm = vm.as_any().downcast_ref::<KvmVm>().expect("Wrong VM type?");

hypervisor/src/kvm/aarch64/gic/redist_regs.rs

@@ -39,8 +39,8 @@ const GICR_ICFGR0: u32 = GICR_SGI_OFFSET + 0x0C00;
 const KVM_DEV_ARM_VGIC_V3_MPIDR_SHIFT: u32 = 32;
 const KVM_DEV_ARM_VGIC_V3_MPIDR_MASK: u64 = 0xffffffff << KVM_DEV_ARM_VGIC_V3_MPIDR_SHIFT as u64;
 
-const KVM_ARM64_SYSREG_MPIDR_EL1: u64 = KVM_REG_ARM64 as u64
-    | KVM_REG_SIZE_U64 as u64
+const KVM_ARM64_SYSREG_MPIDR_EL1: u64 = KVM_REG_ARM64
+    | KVM_REG_SIZE_U64
     | KVM_REG_ARM64_SYSREG as u64
     | (((3_u64) << KVM_REG_ARM64_SYSREG_OP0_SHIFT) & KVM_REG_ARM64_SYSREG_OP0_MASK as u64)
     | (((5_u64) << KVM_REG_ARM64_SYSREG_OP2_SHIFT) & KVM_REG_ARM64_SYSREG_OP2_MASK as u64);

hypervisor/src/kvm/aarch64/mod.rs

@@ -24,8 +24,8 @@ pub use {kvm_ioctls::Cap, kvm_ioctls::Kvm};
 // This macro gets the offset of a structure (i.e `str`) member (i.e `field`) without having
 // an instance of that structure.
 #[macro_export]
-macro_rules! offset__of {
-    ($str:ty, $($field:ident)+) => ({
+macro_rules! offset_of {
+    ($str:ty, $field:ident) => {{
         let tmp: std::mem::MaybeUninit<$str> = std::mem::MaybeUninit::uninit();
         let base = tmp.as_ptr();
 
@@ -34,14 +34,16 @@ macro_rules! offset__of {
         // SAFETY: The pointer is valid and aligned, just not initialised. Using `addr_of` ensures
         // that we don't actually read from `base` (which would be UB) nor create an intermediate
         // reference.
-        let member = unsafe { core::ptr::addr_of!((*base).$($field)*) } as *const u8;
+        let member = unsafe { core::ptr::addr_of!((*base).$field) } as *const u8;
 
         // Avoid warnings when nesting `unsafe` blocks.
         #[allow(unused_unsafe)]
         // SAFETY: The two pointers are within the same allocated object `tmp`. All requirements
         // from offset_from are upheld.
-        unsafe { member.offset_from(base as *const u8) as usize }
-    });
+        unsafe {
+            member.offset_from(base as *const u8) as usize
+        }
+    }};
 }
 
 // Following are macros that help with getting the ID of a aarch64 core register.

hypervisor/src/kvm/mod.rs

@@ -23,7 +23,7 @@ use crate::vec_with_array_field;
 use crate::vm::{self, InterruptSourceConfig, VmOps};
 use crate::HypervisorType;
 #[cfg(target_arch = "aarch64")]
-use crate::{arm64_core_reg_id, offset__of};
+use crate::{arm64_core_reg_id, offset_of};
 use kvm_ioctls::{NoDatamatch, VcpuFd, VmFd};
 use std::any::Any;
 use std::collections::HashMap;
@@ -106,7 +106,7 @@ pub use {
 const KVM_CAP_SGX_ATTRIBUTE: u32 = 196;
 
 #[cfg(feature = "tdx")]
-const KVM_EXIT_TDX: u32 = 35;
+const KVM_EXIT_TDX: u32 = 50;
 #[cfg(feature = "tdx")]
 const TDG_VP_VMCALL_GET_QUOTE: u64 = 0x10002;
 #[cfg(feature = "tdx")]
@@ -335,16 +335,17 @@ impl KvmVm {
     }
 }
 
-///
 /// Implementation of Vm trait for KVM
-/// Example:
-/// #[cfg(feature = "kvm")]
-/// extern crate hypervisor
-/// let kvm = hypervisor::kvm::KvmHypervisor::new().unwrap();
-/// let hypervisor: Arc<dyn hypervisor::Hypervisor> = Arc::new(kvm);
-/// let vm = hypervisor.create_vm().expect("new VM fd creation failed");
-/// vm.set/get().unwrap()
 ///
+/// # Examples
+///
+/// ```
+/// # use hypervisor::kvm::KvmHypervisor;
+/// # use std::sync::Arc;
+/// let kvm = KvmHypervisor::new().unwrap();
+/// let hypervisor = Arc::new(kvm);
+/// let vm = hypervisor.create_vm().expect("new VM fd creation failed");
+/// ```
 impl vm::Vm for KvmVm {
     #[cfg(target_arch = "x86_64")]
     ///
@@ -746,36 +747,34 @@ impl vm::Vm for KvmVm {
     ///
     #[cfg(feature = "tdx")]
     fn tdx_init(&self, cpuid: &[CpuIdEntry], max_vcpus: u32) -> vm::Result<()> {
-        use std::io::{Error, ErrorKind};
-        let cpuid: Vec<kvm_bindings::kvm_cpuid_entry2> =
+        const TDX_ATTR_SEPT_VE_DISABLE: usize = 28;
+
+        let mut cpuid: Vec<kvm_bindings::kvm_cpuid_entry2> =
             cpuid.iter().map(|e| (*e).into()).collect();
-        let kvm_cpuid = kvm_bindings::CpuId::from_entries(&cpuid).map_err(|_| {
-            vm::HypervisorVmError::InitializeTdx(Error::new(
-                ErrorKind::Other,
-                "failed to allocate CpuId",
-            ))
-        })?;
+        cpuid.resize(256, kvm_bindings::kvm_cpuid_entry2::default());
 
         #[repr(C)]
         struct TdxInitVm {
-            max_vcpus: u32,
-            tsc_khz: u32,
             attributes: u64,
-            cpuid: u64,
+            max_vcpus: u32,
+            padding: u32,
             mrconfigid: [u64; 6],
             mrowner: [u64; 6],
             mrownerconfig: [u64; 6],
-            reserved: [u64; 43],
+            cpuid_nent: u32,
+            cpuid_padding: u32,
+            cpuid_entries: [kvm_bindings::kvm_cpuid_entry2; 256],
         }
         let data = TdxInitVm {
+            attributes: 1 << TDX_ATTR_SEPT_VE_DISABLE,
             max_vcpus,
-            tsc_khz: 0,
-            attributes: 0,
-            cpuid: kvm_cpuid.as_fam_struct_ptr() as u64,
+            padding: 0,
             mrconfigid: [0; 6],
             mrowner: [0; 6],
             mrownerconfig: [0; 6],
-            reserved: [0; 43],
+            cpuid_nent: cpuid.len() as u32,
+            cpuid_padding: 0,
+            cpuid_entries: cpuid.as_slice().try_into().unwrap(),
         };
 
         tdx_command(
@@ -837,19 +836,23 @@ impl vm::Vm for KvmVm {
 fn tdx_command(
     fd: &RawFd,
     command: TdxCommand,
-    metadata: u32,
+    flags: u32,
     data: u64,
 ) -> std::result::Result<(), std::io::Error> {
     #[repr(C)]
     struct TdxIoctlCmd {
         command: TdxCommand,
-        metadata: u32,
+        flags: u32,
         data: u64,
+        error: u64,
+        unused: u64,
     }
     let cmd = TdxIoctlCmd {
         command,
-        metadata,
+        flags,
         data,
+        error: 0,
+        unused: 0,
     };
     // SAFETY: FFI call. All input parameters are valid.
     let ret = unsafe {
@@ -915,13 +918,16 @@ impl KvmHypervisor {
     }
 }
 /// Implementation of Hypervisor trait for KVM
-/// Example:
-/// #[cfg(feature = "kvm")]
-/// extern crate hypervisor
-/// let kvm = hypervisor::kvm::KvmHypervisor::new().unwrap();
-/// let hypervisor: Arc<dyn hypervisor::Hypervisor> = Arc::new(kvm);
-/// let vm = hypervisor.create_vm().expect("new VM fd creation failed");
 ///
+/// # Examples
+///
+/// ```
+/// # use hypervisor::kvm::KvmHypervisor;
+/// # use std::sync::Arc;
+/// let kvm = KvmHypervisor::new().unwrap();
+/// let hypervisor = Arc::new(kvm);
+/// let vm = hypervisor.create_vm().expect("new VM fd creation failed");
+/// ```
 impl hypervisor::Hypervisor for KvmHypervisor {
     ///
     /// Returns the type of the hypervisor
@@ -930,13 +936,15 @@ impl hypervisor::Hypervisor for KvmHypervisor {
         HypervisorType::Kvm
     }
     /// Create a KVM vm object of a specific VM type and return the object as Vm trait object
-    /// Example
-    /// # extern crate hypervisor;
-    /// # use hypervisor::KvmHypervisor;
-    /// use hypervisor::KvmVm;
-    /// let hypervisor = KvmHypervisor::new().unwrap();
-    /// let vm = hypervisor.create_vm_with_type(KvmVmType::LegacyVm).unwrap()
     ///
+    /// # Examples
+    ///
+    /// ```
+    /// # use hypervisor::kvm::KvmHypervisor;
+    /// use hypervisor::kvm::KvmVm;
+    /// let hypervisor = KvmHypervisor::new().unwrap();
+    /// let vm = hypervisor.create_vm_with_type(0).unwrap();
+    /// ```
     fn create_vm_with_type(&self, vm_type: u64) -> hypervisor::Result<Arc<dyn vm::Vm>> {
         let fd: VmFd;
         loop {
@@ -990,13 +998,15 @@ impl hypervisor::Hypervisor for KvmHypervisor {
     }
 
     /// Create a KVM vm object and return the object as Vm trait object
-    /// Example
-    /// # extern crate hypervisor;
-    /// # use hypervisor::KvmHypervisor;
-    /// use hypervisor::KvmVm;
-    /// let hypervisor = KvmHypervisor::new().unwrap();
-    /// let vm = hypervisor.create_vm().unwrap()
     ///
+    /// # Examples
+    ///
+    /// ```
+    /// # use hypervisor::kvm::KvmHypervisor;
+    /// use hypervisor::kvm::KvmVm;
+    /// let hypervisor = KvmHypervisor::new().unwrap();
+    /// let vm = hypervisor.create_vm().unwrap();
+    /// ```
     fn create_vm(&self) -> hypervisor::Result<Arc<dyn vm::Vm>> {
         #[allow(unused_mut)]
         let mut vm_type: u64 = 0; // Create with default platform type
@@ -1021,7 +1031,7 @@ impl hypervisor::Hypervisor for KvmHypervisor {
     ///
     /// X86 specific call to get the system supported CPUID values.
     ///
-    fn get_cpuid(&self) -> hypervisor::Result<Vec<CpuIdEntry>> {
+    fn get_supported_cpuid(&self) -> hypervisor::Result<Vec<CpuIdEntry>> {
         let kvm_cpuid = self
             .kvm
             .get_supported_cpuid(kvm_bindings::KVM_MAX_CPUID_ENTRIES)
@@ -1074,6 +1084,11 @@ impl hypervisor::Hypervisor for KvmHypervisor {
             self.kvm.get_guest_debug_hw_bps() as usize
         }
     }
+
+    /// Get maximum number of vCPUs
+    fn get_max_vcpus(&self) -> u32 {
+        self.kvm.get_max_vcpus().min(u32::MAX as usize) as u32
+    }
 }
 /// Vcpu struct for KVM
 pub struct KvmVcpu {
@@ -1085,15 +1100,17 @@ pub struct KvmVcpu {
     hyperv_synic: AtomicBool,
 }
 /// Implementation of Vcpu trait for KVM
-/// Example:
-/// #[cfg(feature = "kvm")]
-/// extern crate hypervisor
-/// let kvm = hypervisor::kvm::KvmHypervisor::new().unwrap();
-/// let hypervisor: Arc<dyn hypervisor::Hypervisor> = Arc::new(kvm);
+///
+/// # Examples
+///
+/// ```
+/// # use hypervisor::kvm::KvmHypervisor;
+/// # use std::sync::Arc;
+/// let kvm = KvmHypervisor::new().unwrap();
+/// let hypervisor = Arc::new(kvm);
 /// let vm = hypervisor.create_vm().expect("new VM fd creation failed");
 /// let vcpu = vm.create_vcpu(0, None).unwrap();
-/// vcpu.get/set().unwrap()
-///
+/// ```
 impl cpu::Vcpu for KvmVcpu {
     #[cfg(target_arch = "x86_64")]
     ///
@@ -1114,7 +1131,7 @@ impl cpu::Vcpu for KvmVcpu {
     #[cfg(target_arch = "aarch64")]
     fn get_regs(&self) -> cpu::Result<StandardRegisters> {
         let mut state: StandardRegisters = kvm_regs::default();
-        let mut off = offset__of!(user_pt_regs, regs);
+        let mut off = offset_of!(user_pt_regs, regs);
         // There are 31 user_pt_regs:
         // https://elixir.free-electrons.com/linux/v4.14.174/source/arch/arm64/include/uapi/asm/ptrace.h#L72
         // These actually are the general-purpose registers of the Armv8-a
@@ -1131,7 +1148,7 @@ impl cpu::Vcpu for KvmVcpu {
 
         // We are now entering the "Other register" section of the ARMv8-a architecture.
         // First one, stack pointer.
-        let off = offset__of!(user_pt_regs, sp);
+        let off = offset_of!(user_pt_regs, sp);
         state.regs.sp = self
             .fd
             .get_one_reg(arm64_core_reg_id!(KVM_REG_SIZE_U64, off))
@@ -1140,7 +1157,7 @@ impl cpu::Vcpu for KvmVcpu {
             .unwrap();
 
         // Second one, the program counter.
-        let off = offset__of!(user_pt_regs, pc);
+        let off = offset_of!(user_pt_regs, pc);
         state.regs.pc = self
             .fd
             .get_one_reg(arm64_core_reg_id!(KVM_REG_SIZE_U64, off))
@@ -1149,7 +1166,7 @@ impl cpu::Vcpu for KvmVcpu {
             .unwrap();
 
         // Next is the processor state.
-        let off = offset__of!(user_pt_regs, pstate);
+        let off = offset_of!(user_pt_regs, pstate);
         state.regs.pstate = self
             .fd
             .get_one_reg(arm64_core_reg_id!(KVM_REG_SIZE_U64, off))
@@ -1158,7 +1175,7 @@ impl cpu::Vcpu for KvmVcpu {
             .unwrap();
 
         // The stack pointer associated with EL1
-        let off = offset__of!(kvm_regs, sp_el1);
+        let off = offset_of!(kvm_regs, sp_el1);
         state.sp_el1 = self
             .fd
             .get_one_reg(arm64_core_reg_id!(KVM_REG_SIZE_U64, off))
@@ -1168,7 +1185,7 @@ impl cpu::Vcpu for KvmVcpu {
 
         // Exception Link Register for EL1, when taking an exception to EL1, this register
         // holds the address to which to return afterwards.
-        let off = offset__of!(kvm_regs, elr_el1);
+        let off = offset_of!(kvm_regs, elr_el1);
         state.elr_el1 = self
             .fd
             .get_one_reg(arm64_core_reg_id!(KVM_REG_SIZE_U64, off))
@@ -1177,7 +1194,7 @@ impl cpu::Vcpu for KvmVcpu {
             .unwrap();
 
         // Saved Program Status Registers, there are 5 of them used in the kernel.
-        let mut off = offset__of!(kvm_regs, spsr);
+        let mut off = offset_of!(kvm_regs, spsr);
         for i in 0..KVM_NR_SPSR as usize {
             state.spsr[i] = self
                 .fd
@@ -1190,7 +1207,7 @@ impl cpu::Vcpu for KvmVcpu {
 
         // Now moving on to floting point registers which are stored in the user_fpsimd_state in the kernel:
         // https://elixir.free-electrons.com/linux/v4.9.62/source/arch/arm64/include/uapi/asm/kvm.h#L53
-        let mut off = offset__of!(kvm_regs, fp_regs) + offset__of!(user_fpsimd_state, vregs);
+        let mut off = offset_of!(kvm_regs, fp_regs) + offset_of!(user_fpsimd_state, vregs);
         for i in 0..32 {
             state.fp_regs.vregs[i] = self
                 .fd
@@ -1200,7 +1217,7 @@ impl cpu::Vcpu for KvmVcpu {
         }
 
         // Floating-point Status Register
-        let off = offset__of!(kvm_regs, fp_regs) + offset__of!(user_fpsimd_state, fpsr);
+        let off = offset_of!(kvm_regs, fp_regs) + offset_of!(user_fpsimd_state, fpsr);
         state.fp_regs.fpsr = self
             .fd
             .get_one_reg(arm64_core_reg_id!(KVM_REG_SIZE_U32, off))
@@ -1209,7 +1226,7 @@ impl cpu::Vcpu for KvmVcpu {
             .unwrap();
 
         // Floating-point Control Register
-        let off = offset__of!(kvm_regs, fp_regs) + offset__of!(user_fpsimd_state, fpcr);
+        let off = offset_of!(kvm_regs, fp_regs) + offset_of!(user_fpsimd_state, fpcr);
         state.fp_regs.fpcr = self
             .fd
             .get_one_reg(arm64_core_reg_id!(KVM_REG_SIZE_U32, off))
@@ -1238,7 +1255,7 @@ impl cpu::Vcpu for KvmVcpu {
     fn set_regs(&self, state: &StandardRegisters) -> cpu::Result<()> {
         // The function follows the exact identical order from `state`. Look there
         // for some additional info on registers.
-        let mut off = offset__of!(user_pt_regs, regs);
+        let mut off = offset_of!(user_pt_regs, regs);
         for i in 0..31 {
             self.fd
                 .set_one_reg(
@@ -1249,7 +1266,7 @@ impl cpu::Vcpu for KvmVcpu {
             off += std::mem::size_of::<u64>();
         }
 
-        let off = offset__of!(user_pt_regs, sp);
+        let off = offset_of!(user_pt_regs, sp);
         self.fd
             .set_one_reg(
                 arm64_core_reg_id!(KVM_REG_SIZE_U64, off),
@@ -1257,7 +1274,7 @@ impl cpu::Vcpu for KvmVcpu {
             )
             .map_err(|e| cpu::HypervisorCpuError::SetCoreRegister(e.into()))?;
 
-        let off = offset__of!(user_pt_regs, pc);
+        let off = offset_of!(user_pt_regs, pc);
         self.fd
             .set_one_reg(
                 arm64_core_reg_id!(KVM_REG_SIZE_U64, off),
@@ -1265,7 +1282,7 @@ impl cpu::Vcpu for KvmVcpu {
             )
             .map_err(|e| cpu::HypervisorCpuError::SetCoreRegister(e.into()))?;
 
-        let off = offset__of!(user_pt_regs, pstate);
+        let off = offset_of!(user_pt_regs, pstate);
         self.fd
             .set_one_reg(
                 arm64_core_reg_id!(KVM_REG_SIZE_U64, off),
@@ -1273,7 +1290,7 @@ impl cpu::Vcpu for KvmVcpu {
             )
             .map_err(|e| cpu::HypervisorCpuError::SetCoreRegister(e.into()))?;
 
-        let off = offset__of!(kvm_regs, sp_el1);
+        let off = offset_of!(kvm_regs, sp_el1);
         self.fd
             .set_one_reg(
                 arm64_core_reg_id!(KVM_REG_SIZE_U64, off),
@@ -1281,7 +1298,7 @@ impl cpu::Vcpu for KvmVcpu {
             )
             .map_err(|e| cpu::HypervisorCpuError::SetCoreRegister(e.into()))?;
 
-        let off = offset__of!(kvm_regs, elr_el1);
+        let off = offset_of!(kvm_regs, elr_el1);
         self.fd
             .set_one_reg(
                 arm64_core_reg_id!(KVM_REG_SIZE_U64, off),
@@ -1289,7 +1306,7 @@ impl cpu::Vcpu for KvmVcpu {
             )
             .map_err(|e| cpu::HypervisorCpuError::SetCoreRegister(e.into()))?;
 
-        let mut off = offset__of!(kvm_regs, spsr);
+        let mut off = offset_of!(kvm_regs, spsr);
         for i in 0..KVM_NR_SPSR as usize {
             self.fd
                 .set_one_reg(
@@ -1300,7 +1317,7 @@ impl cpu::Vcpu for KvmVcpu {
             off += std::mem::size_of::<u64>();
         }
 
-        let mut off = offset__of!(kvm_regs, fp_regs) + offset__of!(user_fpsimd_state, vregs);
+        let mut off = offset_of!(kvm_regs, fp_regs) + offset_of!(user_fpsimd_state, vregs);
         for i in 0..32 {
             self.fd
                 .set_one_reg(
@@ -1311,7 +1328,7 @@ impl cpu::Vcpu for KvmVcpu {
             off += mem::size_of::<u128>();
         }
 
-        let off = offset__of!(kvm_regs, fp_regs) + offset__of!(user_fpsimd_state, fpsr);
+        let off = offset_of!(kvm_regs, fp_regs) + offset_of!(user_fpsimd_state, fpsr);
         self.fd
             .set_one_reg(
                 arm64_core_reg_id!(KVM_REG_SIZE_U32, off),
@@ -1319,7 +1336,7 @@ impl cpu::Vcpu for KvmVcpu {
             )
             .map_err(|e| cpu::HypervisorCpuError::SetCoreRegister(e.into()))?;
 
-        let off = offset__of!(kvm_regs, fp_regs) + offset__of!(user_fpsimd_state, fpcr);
+        let off = offset_of!(kvm_regs, fp_regs) + offset_of!(user_fpsimd_state, fpcr);
         self.fd
             .set_one_reg(
                 arm64_core_reg_id!(KVM_REG_SIZE_U32, off),
@@ -1700,8 +1717,8 @@ impl cpu::Vcpu for KvmVcpu {
         // it to the corresponding KVM ID, and call `KVM_GET_ONE_REG` API to
         // get the value of the system parameter.
         //
-        let id: u64 = KVM_REG_ARM64 as u64
-            | KVM_REG_SIZE_U64 as u64
+        let id: u64 = KVM_REG_ARM64
+            | KVM_REG_SIZE_U64
             | KVM_REG_ARM64_SYSREG as u64
             | ((((sys_reg) >> 5)
                 & (KVM_REG_ARM64_SYSREG_OP0_MASK
@@ -1733,10 +1750,10 @@ impl cpu::Vcpu for KvmVcpu {
         const PSTATE_FAULT_BITS_64: u64 =
             PSR_MODE_EL1h | PSR_A_BIT | PSR_F_BIT | PSR_I_BIT | PSR_D_BIT;
 
-        let kreg_off = offset__of!(kvm_regs, regs);
+        let kreg_off = offset_of!(kvm_regs, regs);
 
         // Get the register index of the PSTATE (Processor State) register.
-        let pstate = offset__of!(user_pt_regs, pstate) + kreg_off;
+        let pstate = offset_of!(user_pt_regs, pstate) + kreg_off;
         self.fd
             .set_one_reg(
                 arm64_core_reg_id!(KVM_REG_SIZE_U64, pstate),
@@ -1747,7 +1764,7 @@ impl cpu::Vcpu for KvmVcpu {
         // Other vCPUs are powered off initially awaiting PSCI wakeup.
         if cpu_id == 0 {
             // Setting the PC (Processor Counter) to the current program address (kernel address).
-            let pc = offset__of!(user_pt_regs, pc) + kreg_off;
+            let pc = offset_of!(user_pt_regs, pc) + kreg_off;
             self.fd
                 .set_one_reg(arm64_core_reg_id!(KVM_REG_SIZE_U64, pc), boot_ip.into())
                 .map_err(|e| cpu::HypervisorCpuError::SetCoreRegister(e.into()))?;
@@ -1756,7 +1773,7 @@ impl cpu::Vcpu for KvmVcpu {
             // "The device tree blob (dtb) must be placed on an 8-byte boundary and must
             // not exceed 2 megabytes in size." -> https://www.kernel.org/doc/Documentation/arm64/booting.txt.
             // We are choosing to place it the end of DRAM. See `get_fdt_addr`.
-            let regs0 = offset__of!(user_pt_regs, regs) + kreg_off;
+            let regs0 = offset_of!(user_pt_regs, regs) + kreg_off;
             self.fd
                 .set_one_reg(
                     arm64_core_reg_id!(KVM_REG_SIZE_U64, regs0),
@@ -1795,11 +1812,10 @@ impl cpu::Vcpu for KvmVcpu {
     /// # Example
     ///
     /// ```rust
-    /// # extern crate hypervisor;
-    /// # use hypervisor::KvmHypervisor;
+    /// # use hypervisor::kvm::KvmHypervisor;
     /// # use std::sync::Arc;
-    /// let kvm = hypervisor::kvm::KvmHypervisor::new().unwrap();
-    /// let hv: Arc<dyn hypervisor::Hypervisor> = Arc::new(kvm);
+    /// let kvm = KvmHypervisor::new().unwrap();
+    /// let hv = Arc::new(kvm);
     /// let vm = hv.create_vm().expect("new VM fd creation failed");
     /// vm.enable_split_irq().unwrap();
     /// let vcpu = vm.create_vcpu(0, None).unwrap();
@@ -1967,11 +1983,10 @@ impl cpu::Vcpu for KvmVcpu {
     /// # Example
     ///
     /// ```rust
-    /// # extern crate hypervisor;
-    /// # use hypervisor::KvmHypervisor;
+    /// # use hypervisor::kvm::KvmHypervisor;
     /// # use std::sync::Arc;
-    /// let kvm = hypervisor::kvm::KvmHypervisor::new().unwrap();
-    /// let hv: Arc<dyn hypervisor::Hypervisor> = Arc::new(kvm);
+    /// let kvm = KvmHypervisor::new().unwrap();
+    /// let hv = Arc::new(kvm);
     /// let vm = hv.create_vm().expect("new VM fd creation failed");
     /// vm.enable_split_irq().unwrap();
     /// let vcpu = vm.create_vcpu(0, None).unwrap();
@@ -2066,6 +2081,7 @@ impl cpu::Vcpu for KvmVcpu {
     #[cfg(feature = "tdx")]
     fn get_tdx_exit_details(&mut self) -> cpu::Result<TdxExitDetails> {
         let kvm_run = self.fd.get_kvm_run();
+        // SAFETY: accessing a union field in a valid structure
         let tdx_vmcall = unsafe { &mut kvm_run.__bindgen_anon_1.tdx.u.vmcall };
 
         tdx_vmcall.status_code = TDG_VP_VMCALL_INVALID_OPERAND;
@@ -2089,6 +2105,7 @@ impl cpu::Vcpu for KvmVcpu {
     #[cfg(feature = "tdx")]
     fn set_tdx_status(&mut self, status: TdxExitStatus) {
         let kvm_run = self.fd.get_kvm_run();
+        // SAFETY: accessing a union field in a valid structure
         let tdx_vmcall = unsafe { &mut kvm_run.__bindgen_anon_1.tdx.u.vmcall };
 
         tdx_vmcall.status_code = match status {

hypervisor/src/lib.rs

@@ -18,8 +18,6 @@
 //! - arm64
 //!
 
-#![allow(clippy::significant_drop_in_scrutinee)]
-
 #[macro_use]
 extern crate anyhow;
 #[cfg(target_arch = "x86_64")]
@@ -50,9 +48,9 @@ mod cpu;
 /// Device related module
 mod device;
 
+pub use crate::hypervisor::{Hypervisor, HypervisorError};
 pub use cpu::{HypervisorCpuError, Vcpu, VmExit};
 pub use device::HypervisorDeviceError;
-pub use hypervisor::{Hypervisor, HypervisorError};
 #[cfg(all(feature = "kvm", target_arch = "aarch64"))]
 pub use kvm::{aarch64, GicState};
 use std::sync::Arc;
@@ -61,9 +59,11 @@ pub use vm::{
     Vm, VmOps,
 };
 
-#[derive(Debug, Copy, Clone, Eq, PartialEq)]
+#[derive(Debug, Copy, Clone)]
 pub enum HypervisorType {
+    #[cfg(feature = "kvm")]
     Kvm,
+    #[cfg(feature = "mshv")]
     Mshv,
 }
 

hypervisor/src/mshv/mod.rs

@@ -38,7 +38,9 @@ use std::fs::File;
 use std::os::unix::io::AsRawFd;
 
 #[cfg(target_arch = "x86_64")]
-use crate::arch::x86::{CpuIdEntry, FpuState, MsrEntry};
+use crate::arch::x86::{
+    CpuIdEntry, FpuState, LapicState, MsrEntry, SpecialRegisters, StandardRegisters,
+};
 
 const DIRTY_BITMAP_CLEAR_DIRTY: u64 = 0x4;
 const DIRTY_BITMAP_SET_DIRTY: u64 = 0x8;
@@ -193,13 +195,16 @@ impl MshvHypervisor {
     }
 }
 /// Implementation of Hypervisor trait for Mshv
-/// Example:
-/// #[cfg(feature = "mshv")]
-/// extern crate hypervisor
-/// let mshv = hypervisor::mshv::MshvHypervisor::new().unwrap();
-/// let hypervisor: Arc<dyn hypervisor::Hypervisor> = Arc::new(mshv);
-/// let vm = hypervisor.create_vm().expect("new VM fd creation failed");
 ///
+/// # Examples
+///
+/// ```
+/// # use hypervisor::mshv::MshvHypervisor;
+/// # use std::sync::Arc;
+/// let mshv = MshvHypervisor::new().unwrap();
+/// let hypervisor = Arc::new(mshv);
+/// let vm = hypervisor.create_vm().expect("new VM fd creation failed");
+/// ```
 impl hypervisor::Hypervisor for MshvHypervisor {
     ///
     /// Returns the type of the hypervisor
@@ -208,13 +213,16 @@ impl hypervisor::Hypervisor for MshvHypervisor {
         HypervisorType::Mshv
     }
     /// Create a mshv vm object and return the object as Vm trait object
-    /// Example
-    /// # extern crate hypervisor;
-    /// # use hypervisor::MshvHypervisor;
-    /// use hypervisor::MshvVm;
-    /// let hypervisor = MshvHypervisor::new().unwrap();
-    /// let vm = hypervisor.create_vm().unwrap()
     ///
+    /// # Examples
+    ///
+    /// ```
+    /// # extern crate hypervisor;
+    /// # use hypervisor::mshv::MshvHypervisor;
+    /// use hypervisor::mshv::MshvVm;
+    /// let hypervisor = MshvHypervisor::new().unwrap();
+    /// let vm = hypervisor.create_vm().unwrap();
+    /// ```
     fn create_vm(&self) -> hypervisor::Result<Arc<dyn vm::Vm>> {
         let fd: VmFd;
         loop {
@@ -268,9 +276,16 @@ impl hypervisor::Hypervisor for MshvHypervisor {
     ///
     /// Get the supported CpuID
     ///
-    fn get_cpuid(&self) -> hypervisor::Result<Vec<CpuIdEntry>> {
+    fn get_supported_cpuid(&self) -> hypervisor::Result<Vec<CpuIdEntry>> {
         Ok(Vec::new())
     }
+
+    /// Get maximum number of vCPUs
+    fn get_max_vcpus(&self) -> u32 {
+        // TODO: Using HV_MAXIMUM_PROCESSORS would be better
+        // but the ioctl API is limited to u8
+        256
+    }
 }
 
 /// Vcpu struct for Microsoft Hypervisor
@@ -283,21 +298,23 @@ pub struct MshvVcpu {
 }
 
 /// Implementation of Vcpu trait for Microsoft Hypervisor
-/// Example:
-/// #[cfg(feature = "mshv")]
-/// extern crate hypervisor
-/// let mshv = hypervisor::mshv::MshvHypervisor::new().unwrap();
-/// let hypervisor: Arc<dyn hypervisor::Hypervisor> = Arc::new(mshv);
-/// let vm = hypervisor.create_vm().expect("new VM fd creation failed");
-/// let vcpu = vm.create_vcpu(0).unwrap();
-/// vcpu.get/set().unwrap()
 ///
+/// # Examples
+///
+/// ```
+/// # use hypervisor::mshv::MshvHypervisor;
+/// # use std::sync::Arc;
+/// let mshv = MshvHypervisor::new().unwrap();
+/// let hypervisor = Arc::new(mshv);
+/// let vm = hypervisor.create_vm().expect("new VM fd creation failed");
+/// let vcpu = vm.create_vcpu(0, None).unwrap();
+/// ```
 impl cpu::Vcpu for MshvVcpu {
     #[cfg(target_arch = "x86_64")]
     ///
     /// Returns the vCPU general purpose registers.
     ///
-    fn get_regs(&self) -> cpu::Result<crate::arch::x86::StandardRegisters> {
+    fn get_regs(&self) -> cpu::Result<StandardRegisters> {
         Ok(self
             .fd
             .get_regs()
@@ -308,7 +325,7 @@ impl cpu::Vcpu for MshvVcpu {
     ///
     /// Sets the vCPU general purpose registers.
     ///
-    fn set_regs(&self, regs: &crate::arch::x86::StandardRegisters) -> cpu::Result<()> {
+    fn set_regs(&self, regs: &StandardRegisters) -> cpu::Result<()> {
         let regs = (*regs).into();
         self.fd
             .set_regs(&regs)
@@ -318,7 +335,7 @@ impl cpu::Vcpu for MshvVcpu {
     ///
     /// Returns the vCPU special registers.
     ///
-    fn get_sregs(&self) -> cpu::Result<crate::arch::x86::SpecialRegisters> {
+    fn get_sregs(&self) -> cpu::Result<SpecialRegisters> {
         Ok(self
             .fd
             .get_sregs()
@@ -329,7 +346,7 @@ impl cpu::Vcpu for MshvVcpu {
     ///
     /// Sets the vCPU special registers.
     ///
-    fn set_sregs(&self, sregs: &crate::arch::x86::SpecialRegisters) -> cpu::Result<()> {
+    fn set_sregs(&self, sregs: &SpecialRegisters) -> cpu::Result<()> {
         let sregs = (*sregs).into();
         self.fd
             .set_sregs(&sregs)
@@ -444,10 +461,10 @@ impl cpu::Vcpu for MshvVcpu {
                             /* Advance RIP and update RAX */
                             let arr_reg_name_value = [
                                 (
-                                    hv_register_name::HV_X64_REGISTER_RIP,
+                                    hv_register_name_HV_X64_REGISTER_RIP,
                                     info.header.rip + insn_len,
                                 ),
-                                (hv_register_name::HV_X64_REGISTER_RAX, ret_rax),
+                                (hv_register_name_HV_X64_REGISTER_RAX, ret_rax),
                             ];
                             set_registers_64!(self.fd, arr_reg_name_value)
                                 .map_err(|e| cpu::HypervisorCpuError::SetRegister(e.into()))?;
@@ -456,12 +473,13 @@ impl cpu::Vcpu for MshvVcpu {
                         _ => {}
                     }
 
-                    // SAFETY: access_info is valid, otherwise we won't be here
                     assert!(
+                        // SAFETY: access_info is valid, otherwise we won't be here
                         (unsafe { access_info.__bindgen_anon_1.string_op() } != 1),
                         "String IN/OUT not supported"
                     );
                     assert!(
+                        // SAFETY: access_info is valid, otherwise we won't be here
                         (unsafe { access_info.__bindgen_anon_1.rep_prefix() } != 1),
                         "Rep IN/OUT not supported"
                     );
@@ -492,10 +510,10 @@ impl cpu::Vcpu for MshvVcpu {
                     /* Advance RIP and update RAX */
                     let arr_reg_name_value = [
                         (
-                            hv_register_name::HV_X64_REGISTER_RIP,
+                            hv_register_name_HV_X64_REGISTER_RIP,
                             info.header.rip + insn_len,
                         ),
-                        (hv_register_name::HV_X64_REGISTER_RAX, ret_rax),
+                        (hv_register_name_HV_X64_REGISTER_RAX, ret_rax),
                     ];
                     set_registers_64!(self.fd, arr_reg_name_value)
                         .map_err(|e| cpu::HypervisorCpuError::SetRegister(e.into()))?;
@@ -565,8 +583,14 @@ impl cpu::Vcpu for MshvVcpu {
     ///
     /// X86 specific call to setup the CPUID registers.
     ///
-    fn set_cpuid2(&self, _cpuid: &[CpuIdEntry]) -> cpu::Result<()> {
-        Ok(())
+    fn set_cpuid2(&self, cpuid: &[CpuIdEntry]) -> cpu::Result<()> {
+        let cpuid: Vec<mshv_bindings::hv_cpuid_entry> = cpuid.iter().map(|e| (*e).into()).collect();
+        let mshv_cpuid = <CpuId>::from_entries(&cpuid)
+            .map_err(|_| cpu::HypervisorCpuError::SetCpuid(anyhow!("failed to create CpuId")))?;
+
+        self.fd
+            .register_intercept_result_cpuid(&mshv_cpuid)
+            .map_err(|e| cpu::HypervisorCpuError::SetCpuid(e.into()))
     }
     #[cfg(target_arch = "x86_64")]
     ///
@@ -579,7 +603,7 @@ impl cpu::Vcpu for MshvVcpu {
     ///
     /// Returns the state of the LAPIC (Local Advanced Programmable Interrupt Controller).
     ///
-    fn get_lapic(&self) -> cpu::Result<crate::arch::x86::LapicState> {
+    fn get_lapic(&self) -> cpu::Result<LapicState> {
         Ok(self
             .fd
             .get_lapic()
@@ -590,7 +614,7 @@ impl cpu::Vcpu for MshvVcpu {
     ///
     /// Sets the state of the LAPIC (Local Advanced Programmable Interrupt Controller).
     ///
-    fn set_lapic(&self, lapic: &crate::arch::x86::LapicState) -> cpu::Result<()> {
+    fn set_lapic(&self, lapic: &LapicState) -> cpu::Result<()> {
         let lapic: mshv_bindings::LapicState = (*lapic).clone().into();
         self.fd
             .set_lapic(&lapic)
@@ -914,15 +938,17 @@ impl MshvVm {
 
 ///
 /// Implementation of Vm trait for Mshv
-/// Example:
-/// #[cfg(feature = "mshv")]
-/// # extern crate hypervisor;
-/// # use hypervisor::MshvHypervisor;
-/// let mshv = MshvHypervisor::new().unwrap();
-/// let hypervisor: Arc<dyn hypervisor::Hypervisor> = Arc::new(mshv);
-/// let vm = hypervisor.create_vm().expect("new VM fd creation failed");
-/// vm.set/get().unwrap()
 ///
+/// # Examples
+///
+/// ```
+/// # extern crate hypervisor;
+/// # use hypervisor::mshv::MshvHypervisor;
+/// # use std::sync::Arc;
+/// let mshv = MshvHypervisor::new().unwrap();
+/// let hypervisor = Arc::new(mshv);
+/// let vm = hypervisor.create_vm().expect("new VM fd creation failed");
+/// ```
 impl vm::Vm for MshvVm {
     #[cfg(target_arch = "x86_64")]
     ///

net_gen/src/iff.rs

@@ -1187,6 +1187,7 @@ pub struct ifreq {
 
 impl Default for ifreq {
     fn default() -> Self {
+        // SAFETY: all zeros is a valid pattern for this data type
         unsafe { std::mem::zeroed() }
     }
 }

net_util/Cargo.toml

@@ -7,22 +7,22 @@ edition = "2021"
 [dependencies]
 epoll = "4.3.1"
 getrandom = "0.2.8"
-libc = "0.2.138"
+libc = "0.2.139"
 log = "0.4.17"
 net_gen = { path = "../net_gen" }
 rate_limiter = { path = "../rate_limiter" }
-serde = "1.0.150"
-thiserror = "1.0.37"
-versionize = "0.1.9"
+serde = "1.0.156"
+thiserror = "1.0.39"
+versionize = "0.1.10"
 versionize_derive = "0.1.4"
-virtio-bindings = "0.1.0"
-virtio-queue = "0.7.0"
+virtio-bindings = "0.2.0"
+virtio-queue = "0.7.1"
 vm-memory = { version = "0.10.0", features = ["backend-mmap", "backend-atomic", "backend-bitmap"] }
 vm-virtio = { path = "../vm-virtio" }
 vmm-sys-util = "0.11.0"
 
 [dev-dependencies]
-once_cell = "1.16.0"
-pnet = "0.34.0"
-pnet_datalink = "0.34.0"
-serde_json = "1.0.89"
+once_cell = "1.17.1"
+pnet = "0.33.0"
+pnet_datalink = "0.33.0"
+serde_json = "1.0.96"

net_util/src/ctrl_queue.rs

@@ -6,7 +6,7 @@ use crate::GuestMemoryMmap;
 use crate::Tap;
 use libc::c_uint;
 use std::sync::Arc;
-use virtio_bindings::bindings::virtio_net::{
+use virtio_bindings::virtio_net::{
     VIRTIO_NET_CTRL_GUEST_OFFLOADS, VIRTIO_NET_CTRL_GUEST_OFFLOADS_SET, VIRTIO_NET_CTRL_MQ,
     VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MAX, VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MIN,
     VIRTIO_NET_CTRL_MQ_VQ_PAIRS_SET, VIRTIO_NET_ERR, VIRTIO_NET_F_GUEST_CSUM,

net_util/src/lib.rs

@@ -21,7 +21,7 @@ use std::{io, mem, net};
 use thiserror::Error;
 use versionize::{VersionMap, Versionize, VersionizeResult};
 use versionize_derive::Versionize;
-use virtio_bindings::bindings::virtio_net::{
+use virtio_bindings::virtio_net::{
     virtio_net_hdr_v1, VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MAX, VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MIN,
     VIRTIO_NET_F_GUEST_CSUM, VIRTIO_NET_F_GUEST_ECN, VIRTIO_NET_F_GUEST_TSO4,
     VIRTIO_NET_F_GUEST_TSO6, VIRTIO_NET_F_GUEST_UFO, VIRTIO_NET_F_MAC, VIRTIO_NET_F_MQ,
@@ -66,32 +66,34 @@ fn create_sockaddr(ip_addr: net::Ipv4Addr) -> net_gen::sockaddr {
     let addr_in = net_gen::sockaddr_in {
         sin_family: net_gen::AF_INET as u16,
         sin_port: 0,
+        // SAFETY: ip_addr can be safely transmute to in_addr
         sin_addr: unsafe { mem::transmute(ip_addr.octets()) },
         __pad: [0; 8usize],
     };
 
+    // SAFETY: addr_in can be safely transmute to sockaddr
     unsafe { mem::transmute(addr_in) }
 }
 
 fn create_inet_socket() -> Result<net::UdpSocket> {
-    // This is safe since we check the return value.
+    // SAFETY: we check the return value.
     let sock = unsafe { libc::socket(libc::AF_INET, libc::SOCK_DGRAM, 0) };
     if sock < 0 {
         return Err(Error::CreateSocket(IoError::last_os_error()));
     }
 
-    // This is safe; nothing else will use or hold onto the raw sock fd.
+    // SAFETY: nothing else will use or hold onto the raw sock fd.
     Ok(unsafe { net::UdpSocket::from_raw_fd(sock) })
 }
 
 fn create_unix_socket() -> Result<net::UdpSocket> {
-    // This is safe since we check the return value.
+    // SAFETY: we check the return value.
     let sock = unsafe { libc::socket(libc::AF_UNIX, libc::SOCK_DGRAM, 0) };
     if sock < 0 {
         return Err(Error::CreateSocket(IoError::last_os_error()));
     }
 
-    // This is safe; nothing else will use or hold onto the raw sock fd.
+    // SAFETY: nothing else will use or hold onto the raw sock fd.
     Ok(unsafe { net::UdpSocket::from_raw_fd(sock) })
 }
 

net_util/src/queue_pair.rs

@@ -83,10 +83,11 @@ impl TxVirtio {
             }
 
             let len = if !iovecs.is_empty() {
+                // SAFETY: FFI call with correct arguments
                 let result = unsafe {
                     libc::writev(
                         tap.as_raw_fd() as libc::c_int,
-                        iovecs.as_ptr(),
+                        iovecs.as_ptr() as *const libc::iovec,
                         iovecs.len() as libc::c_int,
                     )
                 };
@@ -221,10 +222,11 @@ impl RxVirtio {
             }
 
             let len = if !iovecs.is_empty() {
+                // SAFETY: FFI call with correct arguments
                 let result = unsafe {
                     libc::readv(
                         tap.as_raw_fd() as libc::c_int,
-                        iovecs.as_ptr(),
+                        iovecs.as_ptr() as *const libc::iovec,
                         iovecs.len() as libc::c_int,
                     )
                 };

net_util/src/tap.rs

@@ -28,8 +28,8 @@ pub enum Error {
     GetFeatures(IoError),
     #[error("Missing multiqueue support in the kernel")]
     MultiQueueKernelSupport,
-    #[error("ioctl failed: {0}")]
-    IoctlError(IoError),
+    #[error("ioctl ({0}) failed: {1}")]
+    IoctlError(c_ulong, IoError),
     #[error("Failed to create a socket: {0}")]
     NetUtil(NetUtilError),
     #[error("Invalid interface name")]
@@ -87,9 +87,37 @@ fn build_terminated_if_name(if_name: &str) -> Result<Vec<u8>> {
 }
 
 impl Tap {
+    unsafe fn ioctl_with_mut_ref<F: AsRawFd, T>(fd: &F, req: c_ulong, arg: &mut T) -> Result<()> {
+        let ret = ioctl_with_mut_ref(fd, req, arg);
+        if ret < 0 {
+            return Err(Error::IoctlError(req, IoError::last_os_error()));
+        }
+
+        Ok(())
+    }
+
+    unsafe fn ioctl_with_ref<F: AsRawFd, T>(fd: &F, req: c_ulong, arg: &T) -> Result<()> {
+        let ret = ioctl_with_ref(fd, req, arg);
+        if ret < 0 {
+            return Err(Error::IoctlError(req, IoError::last_os_error()));
+        }
+
+        Ok(())
+    }
+
+    unsafe fn ioctl_with_val<F: AsRawFd>(fd: &F, req: c_ulong, arg: c_ulong) -> Result<()> {
+        let ret = ioctl_with_val(fd, req, arg);
+        if ret < 0 {
+            return Err(Error::IoctlError(req, IoError::last_os_error()));
+        }
+
+        Ok(())
+    }
+
     pub fn open_named(if_name: &str, num_queue_pairs: usize, flags: Option<i32>) -> Result<Tap> {
         let terminated_if_name = build_terminated_if_name(if_name)?;
 
+        // SAFETY: FFI call
         let fd = unsafe {
             // Open calls are safe because we give a constant null-terminated
             // string and verify the result.
@@ -102,13 +130,14 @@ impl Tap {
             return Err(Error::OpenTun(IoError::last_os_error()));
         }
 
-        // We just checked that the fd is valid.
+        // SAFETY: We just checked that the fd is valid.
         let tuntap = unsafe { File::from_raw_fd(fd) };
 
         // Let's validate some features before going any further.
         // ioctl is safe since we call it with a valid tap fd and check the return
         // value.
         let mut features = 0;
+        // SAFETY: IOCTL with correct arguments
         let ret = unsafe { ioctl_with_mut_ref(&tuntap, net_gen::TUNGETFEATURES(), &mut features) };
         if ret < 0 {
             return Err(Error::GetFeatures(IoError::last_os_error()));
@@ -123,6 +152,7 @@ impl Tap {
         // don't call as_mut on the same union field more than once, this block
         // is safe.
         let mut ifreq: net_gen::ifreq = Default::default();
+        // SAFETY: see the comment above.
         unsafe {
             let ifrn_name = ifreq.ifr_ifrn.ifrn_name.as_mut();
             let name_slice = &mut ifrn_name[..terminated_if_name.len()];
@@ -134,16 +164,16 @@ impl Tap {
             }
         }
 
-        // ioctl is safe since we call it with a valid tap fd and check the return
+        // SAFETY: ioctl is safe since we call it with a valid tap fd and check the return
         // value.
         let ret = unsafe { ioctl_with_mut_ref(&tuntap, net_gen::TUNSETIFF(), &mut ifreq) };
         if ret < 0 {
             return Err(Error::ConfigureTap(IoError::last_os_error()));
         }
 
+        // SAFETY: only the name is accessed, and it's cloned out.
         let mut if_name = unsafe { ifreq.ifr_ifrn.ifrn_name }.to_vec();
         if_name.truncate(terminated_if_name.len() - 1);
-        // Safe since only the name is accessed, and it's cloned out.
         Ok(Tap {
             tap_file: tuntap,
             if_name,
@@ -158,6 +188,7 @@ impl Tap {
     pub fn from_tap_fd(fd: RawFd, num_queue_pairs: usize) -> Result<Tap> {
         // Ensure that the file is opened non-blocking, this is particularly
         // needed when opened via the shell for macvtap.
+        // SAFETY: FFI call
         let ret = unsafe {
             let mut flags = libc::fcntl(fd, libc::F_GETFL);
             flags |= libc::O_NONBLOCK;
@@ -167,19 +198,20 @@ impl Tap {
             return Err(Error::ConfigureTap(IoError::last_os_error()));
         }
 
+        // SAFETY: fd is a tap fd
         let tap_file = unsafe { File::from_raw_fd(fd) };
         let mut ifreq: net_gen::ifreq = Default::default();
 
         // Get current config including name
-        let ret = unsafe { ioctl_with_mut_ref(&tap_file, net_gen::TUNGETIFF(), &mut ifreq) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
-        // We only access one field of the ifru union, hence this is safe.
+        // SAFETY: IOCTL with correct arugments
+        unsafe { Self::ioctl_with_mut_ref(&tap_file, net_gen::TUNGETIFF(), &mut ifreq)? };
+
+        // SAFETY: We only access one field of the ifru union
         let if_name = unsafe { ifreq.ifr_ifrn.ifrn_name }.to_vec();
 
         // Try and update flags. Depending on how the tap was created (macvtap
         // or via open_named()) this might return -EEXIST so we just ignore that.
+        // SAFETY: access union fields
         unsafe {
             ifreq.ifr_ifru.ifru_flags =
                 (net_gen::IFF_TAP | net_gen::IFF_NO_PI | net_gen::IFF_VNET_HDR) as c_short;
@@ -187,6 +219,7 @@ impl Tap {
                 ifreq.ifr_ifru.ifru_flags |= net_gen::IFF_MULTI_QUEUE as c_short;
             }
         }
+        // SAFETY: IOCTL with correct arguments
         let ret = unsafe { ioctl_with_mut_ref(&tap_file, net_gen::TUNSETIFF(), &mut ifreq) };
         if ret < 0 && IoError::last_os_error().raw_os_error().unwrap() != libc::EEXIST {
             return Err(Error::ConfigureTap(IoError::last_os_error()));
@@ -208,14 +241,8 @@ impl Tap {
 
         ifreq.ifr_ifru.ifru_addr = addr;
 
-        // ioctl is safe. Called with a valid sock fd, and we check the return.
-        let ret =
-            unsafe { ioctl_with_ref(&sock, net_gen::sockios::SIOCSIFADDR as c_ulong, &ifreq) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
-
-        Ok(())
+        // SAFETY: ioctl is safe. Called with a valid sock fd, and we check the return.
+        unsafe { Self::ioctl_with_ref(&sock, net_gen::sockios::SIOCSIFADDR as c_ulong, &ifreq) }
     }
 
     /// Set mac addr for tap interface.
@@ -233,13 +260,10 @@ impl Tap {
 
         let mut ifreq = self.get_ifreq();
 
-        // ioctl is safe. Called with a valid sock fd, and we check the return.
-        let ret =
-            unsafe { ioctl_with_ref(&sock, net_gen::sockios::SIOCGIFHWADDR as c_ulong, &ifreq) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
-        // We only access one field of the ifru union, hence this is safe.
+        // SAFETY: ioctl is safe. Called with a valid sock fd, and we check the return.
+        unsafe { Self::ioctl_with_ref(&sock, net_gen::sockios::SIOCGIFHWADDR as c_ulong, &ifreq)? };
+
+        // SAFETY: We only access one field of the ifru union
         unsafe {
             let ifru_hwaddr = &mut ifreq.ifr_ifru.ifru_hwaddr;
             for (i, v) in addr.get_bytes().iter().enumerate() {
@@ -247,14 +271,8 @@ impl Tap {
             }
         }
 
-        // ioctl is safe. Called with a valid sock fd, and we check the return.
-        let ret =
-            unsafe { ioctl_with_ref(&sock, net_gen::sockios::SIOCSIFHWADDR as c_ulong, &ifreq) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
-
-        Ok(())
+        // SAFETY: ioctl is safe. Called with a valid sock fd, and we check the return.
+        unsafe { Self::ioctl_with_ref(&sock, net_gen::sockios::SIOCSIFHWADDR as c_ulong, &ifreq) }
     }
 
     /// Get mac addr for tap interface.
@@ -263,14 +281,10 @@ impl Tap {
 
         let ifreq = self.get_ifreq();
 
-        // ioctl is safe. Called with a valid sock fd, and we check the return.
-        let ret =
-            unsafe { ioctl_with_ref(&sock, net_gen::sockios::SIOCGIFHWADDR as c_ulong, &ifreq) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
+        // SAFETY: ioctl is safe. Called with a valid sock fd, and we check the return.
+        unsafe { Self::ioctl_with_ref(&sock, net_gen::sockios::SIOCGIFHWADDR as c_ulong, &ifreq)? };
 
-        // We only access one field of the ifru union, hence this is safe.
+        // SAFETY: We only access one field of the ifru union
         let addr = unsafe {
             MacAddr::from_bytes(&ifreq.ifr_ifru.ifru_hwaddr.sa_data[0..MAC_ADDR_LEN])
                 .map_err(Error::MacParsing)?
@@ -287,57 +301,45 @@ impl Tap {
 
         ifreq.ifr_ifru.ifru_addr = addr;
 
-        // ioctl is safe. Called with a valid sock fd, and we check the return.
-        let ret =
-            unsafe { ioctl_with_ref(&sock, net_gen::sockios::SIOCSIFNETMASK as c_ulong, &ifreq) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
-
-        Ok(())
+        // SAFETY: ioctl is safe. Called with a valid sock fd, and we check the return.
+        unsafe { Self::ioctl_with_ref(&sock, net_gen::sockios::SIOCSIFNETMASK as c_ulong, &ifreq) }
     }
 
+    #[cfg(not(fuzzing))]
     pub fn mtu(&self) -> Result<i32> {
         let sock = create_unix_socket().map_err(Error::NetUtil)?;
 
         let ifreq = self.get_ifreq();
 
-        // ioctl is safe. Called with a valid sock fd, and we check the return.
-        let ret = unsafe { ioctl_with_ref(&sock, net_gen::sockios::SIOCGIFMTU as c_ulong, &ifreq) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
+        // SAFETY: ioctl is safe. Called with a valid sock fd, and we check the return.
+        unsafe { Self::ioctl_with_ref(&sock, net_gen::sockios::SIOCGIFMTU as c_ulong, &ifreq)? };
 
+        // SAFETY: access a union field
         let mtu = unsafe { ifreq.ifr_ifru.ifru_mtu };
 
         Ok(mtu)
     }
 
+    #[cfg(fuzzing)]
+    pub fn mtu(&self) -> Result<i32> {
+        // Consistent with the `virtio_devices::net::MIN_MTU`
+        Ok(1280)
+    }
+
     pub fn set_mtu(&self, mtu: i32) -> Result<()> {
         let sock = create_unix_socket().map_err(Error::NetUtil)?;
 
         let mut ifreq = self.get_ifreq();
         ifreq.ifr_ifru.ifru_mtu = mtu;
 
-        // ioctl is safe. Called with a valid sock fd, and we check the return.
-        let ret = unsafe { ioctl_with_ref(&sock, net_gen::sockios::SIOCSIFMTU as c_ulong, &ifreq) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
-
-        Ok(())
+        // SAFETY: ioctl is safe. Called with a valid sock fd, and we check the return.
+        unsafe { Self::ioctl_with_ref(&sock, net_gen::sockios::SIOCSIFMTU as c_ulong, &ifreq) }
     }
 
     /// Set the offload flags for the tap interface.
     pub fn set_offload(&self, flags: c_uint) -> Result<()> {
-        // ioctl is safe. Called with a valid tap fd, and we check the return.
-        let ret =
-            unsafe { ioctl_with_val(&self.tap_file, net_gen::TUNSETOFFLOAD(), flags as c_ulong) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
-
-        Ok(())
+        // SAFETY: ioctl is safe. Called with a valid tap fd, and we check the return.
+        unsafe { Self::ioctl_with_val(&self.tap_file, net_gen::TUNSETOFFLOAD(), flags as c_ulong) }
     }
 
     /// Enable the tap interface.
@@ -346,13 +348,11 @@ impl Tap {
 
         let mut ifreq = self.get_ifreq();
 
-        let ret =
-            unsafe { ioctl_with_ref(&sock, net_gen::sockios::SIOCGIFFLAGS as c_ulong, &ifreq) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
+        // SAFETY: IOCTL with correct arguments
+        unsafe { Self::ioctl_with_ref(&sock, net_gen::sockios::SIOCGIFFLAGS as c_ulong, &ifreq)? };
 
         // If TAP device is already up don't try and enable it
+        // SAFETY: access a union field
         let ifru_flags = unsafe { ifreq.ifr_ifru.ifru_flags };
         if ifru_flags & net_gen::net_device_flags_IFF_UP as i16
             == net_gen::net_device_flags_IFF_UP as i16
@@ -362,25 +362,14 @@ impl Tap {
 
         ifreq.ifr_ifru.ifru_flags = net_gen::net_device_flags_IFF_UP as i16;
 
-        // ioctl is safe. Called with a valid sock fd, and we check the return.
-        let ret =
-            unsafe { ioctl_with_ref(&sock, net_gen::sockios::SIOCSIFFLAGS as c_ulong, &ifreq) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
-
-        Ok(())
+        // SAFETY: ioctl is safe. Called with a valid sock fd, and we check the return.
+        unsafe { Self::ioctl_with_ref(&sock, net_gen::sockios::SIOCSIFFLAGS as c_ulong, &ifreq) }
     }
 
     /// Set the size of the vnet hdr.
     pub fn set_vnet_hdr_size(&self, size: c_int) -> Result<()> {
-        // ioctl is safe. Called with a valid tap fd, and we check the return.
-        let ret = unsafe { ioctl_with_ref(&self.tap_file, net_gen::TUNSETVNETHDRSZ(), &size) };
-        if ret < 0 {
-            return Err(Error::IoctlError(IoError::last_os_error()));
-        }
-
-        Ok(())
+        // SAFETY: ioctl is safe. Called with a valid tap fd, and we check the return.
+        unsafe { Self::ioctl_with_ref(&self.tap_file, net_gen::TUNSETVNETHDRSZ(), &size) }
     }
 
     fn get_ifreq(&self) -> net_gen::ifreq {
@@ -388,6 +377,7 @@ impl Tap {
 
         // This sets the name of the interface, which is the only entry
         // in a single-field union.
+        // SAFETY: access union fields and we're sure the copy is okay.
         unsafe {
             let ifrn_name = ifreq.ifr_ifrn.ifrn_name.as_mut();
             let name_slice = &mut ifrn_name[..self.if_name.len()];
@@ -400,6 +390,11 @@ impl Tap {
     pub fn get_if_name(&self) -> Vec<u8> {
         self.if_name.clone()
     }
+
+    #[cfg(fuzzing)]
+    pub fn new_for_fuzzing(tap_file: File, if_name: Vec<u8>) -> Self {
+        Tap { tap_file, if_name }
+    }
 }
 
 impl Read for Tap {

pci/Cargo.toml

@@ -10,20 +10,21 @@ kvm = ["vfio-ioctls/kvm"]
 mshv = ["vfio-ioctls/mshv"]
 
 [dependencies]
-anyhow = "1.0.66"
+anyhow = "1.0.70"
 byteorder = "1.4.3"
 hypervisor = { path = "../hypervisor" }
 vfio-bindings = { git = "https://github.com/rust-vmm/vfio", branch = "main", features = ["fam-wrappers"] }
 vfio-ioctls = { git = "https://github.com/rust-vmm/vfio", branch = "main", default-features = false }
-vfio_user = { path = "../vfio_user" }
+vfio_user = { git = "https://github.com/rust-vmm/vfio-user", branch = "main" }
+
 vmm-sys-util = "0.11.0"
-libc = "0.2.138"
+libc = "0.2.139"
 log = "0.4.17"
-serde = { version = "1.0.150", features = ["derive"] }
-thiserror = "1.0.37"
-versionize = "0.1.9"
+serde = { version = "1.0.156", features = ["derive"] }
+thiserror = "1.0.39"
+versionize = "0.1.10"
 versionize_derive = "0.1.4"
 vm-allocator = { path = "../vm-allocator" }
 vm-device = { path = "../vm-device" }
-vm-memory = "0.10.0"
-vm-migration = { path = "../vm-migration" }
+vm-memory = { version = "0.10.0", features = ["backend-mmap", "backend-atomic", "backend-bitmap"] }
+vm-migration = { path = "../vm-migration" }

pci/src/bus.rs

@@ -62,6 +62,7 @@ impl PciRoot {
                     0,
                     0,
                     None,
+                    None,
                 ),
             }
         }

pci/src/configuration.rs

@@ -32,6 +32,8 @@ const CAPABILITY_MAX_OFFSET: usize = 192;
 
 const INTERRUPT_LINE_PIN_REG: usize = 15;
 
+pub const PCI_CONFIGURATION_ID: &str = "pci_configuration";
+
 /// Represents the types of PCI headers allowed in the configuration registers.
 #[derive(Copy, Clone)]
 pub enum PciHeaderType {
@@ -394,7 +396,7 @@ fn decode_64_bits_bar_size(bar_size_hi: u32, bar_size_lo: u32) -> Option<u64> {
     None
 }
 
-#[derive(Default, Clone, Copy, Versionize)]
+#[derive(Debug, Default, Clone, Copy, Versionize)]
 struct PciBar {
     addr: u32,
     size: u32,
@@ -403,7 +405,7 @@ struct PciBar {
 }
 
 #[derive(Versionize)]
-struct PciConfigurationState {
+pub struct PciConfigurationState {
     registers: Vec<u32>,
     writable_bits: Vec<u32>,
     bars: Vec<PciBar>,
@@ -553,46 +555,78 @@ impl PciConfiguration {
         subsystem_vendor_id: u16,
         subsystem_id: u16,
         msix_config: Option<Arc<Mutex<MsixConfig>>>,
+        state: Option<PciConfigurationState>,
     ) -> Self {
-        let mut registers = [0u32; NUM_CONFIGURATION_REGISTERS];
-        let mut writable_bits = [0u32; NUM_CONFIGURATION_REGISTERS];
-        registers[0] = u32::from(device_id) << 16 | u32::from(vendor_id);
-        // TODO(dverkamp): Status should be write-1-to-clear
-        writable_bits[1] = 0x0000_ffff; // Status (r/o), command (r/w)
-        let pi = if let Some(pi) = programming_interface {
-            pi.get_register_value()
+        let (
+            registers,
+            writable_bits,
+            bars,
+            rom_bar_addr,
+            rom_bar_size,
+            rom_bar_used,
+            last_capability,
+            msix_cap_reg_idx,
+        ) = if let Some(state) = state {
+            (
+                state.registers.try_into().unwrap(),
+                state.writable_bits.try_into().unwrap(),
+                state.bars.try_into().unwrap(),
+                state.rom_bar_addr,
+                state.rom_bar_size,
+                state.rom_bar_used,
+                state.last_capability,
+                state.msix_cap_reg_idx,
+            )
         } else {
-            0
-        };
-        registers[2] = u32::from(class_code.get_register_value()) << 24
-            | u32::from(subclass.get_register_value()) << 16
-            | u32::from(pi) << 8
-            | u32::from(revision_id);
-        writable_bits[3] = 0x0000_00ff; // Cacheline size (r/w)
-        match header_type {
-            PciHeaderType::Device => {
-                registers[3] = 0x0000_0000; // Header type 0 (device)
-                writable_bits[15] = 0x0000_00ff; // Interrupt line (r/w)
-            }
-            PciHeaderType::Bridge => {
-                registers[3] = 0x0001_0000; // Header type 1 (bridge)
-                writable_bits[9] = 0xfff0_fff0; // Memory base and limit
-                writable_bits[15] = 0xffff_00ff; // Bridge control (r/w), interrupt line (r/w)
-            }
-        };
-        registers[11] = u32::from(subsystem_id) << 16 | u32::from(subsystem_vendor_id);
+            let mut registers = [0u32; NUM_CONFIGURATION_REGISTERS];
+            let mut writable_bits = [0u32; NUM_CONFIGURATION_REGISTERS];
+            registers[0] = u32::from(device_id) << 16 | u32::from(vendor_id);
+            // TODO(dverkamp): Status should be write-1-to-clear
+            writable_bits[1] = 0x0000_ffff; // Status (r/o), command (r/w)
+            let pi = if let Some(pi) = programming_interface {
+                pi.get_register_value()
+            } else {
+                0
+            };
+            registers[2] = u32::from(class_code.get_register_value()) << 24
+                | u32::from(subclass.get_register_value()) << 16
+                | u32::from(pi) << 8
+                | u32::from(revision_id);
+            writable_bits[3] = 0x0000_00ff; // Cacheline size (r/w)
+            match header_type {
+                PciHeaderType::Device => {
+                    registers[3] = 0x0000_0000; // Header type 0 (device)
+                    writable_bits[15] = 0x0000_00ff; // Interrupt line (r/w)
+                }
+                PciHeaderType::Bridge => {
+                    registers[3] = 0x0001_0000; // Header type 1 (bridge)
+                    writable_bits[9] = 0xfff0_fff0; // Memory base and limit
+                    writable_bits[15] = 0xffff_00ff; // Bridge control (r/w), interrupt line (r/w)
+                }
+            };
+            registers[11] = u32::from(subsystem_id) << 16 | u32::from(subsystem_vendor_id);
 
-        let bars = [PciBar::default(); NUM_BAR_REGS];
+            (
+                registers,
+                writable_bits,
+                [PciBar::default(); NUM_BAR_REGS],
+                0,
+                0,
+                false,
+                None,
+                None,
+            )
+        };
 
         PciConfiguration {
             registers,
             writable_bits,
             bars,
-            rom_bar_addr: 0,
-            rom_bar_size: 0,
-            rom_bar_used: false,
-            last_capability: None,
-            msix_cap_reg_idx: None,
+            rom_bar_addr,
+            rom_bar_size,
+            rom_bar_used,
+            last_capability,
+            msix_cap_reg_idx,
             msix_config,
         }
     }
@@ -610,18 +644,6 @@ impl PciConfiguration {
         }
     }
 
-    fn set_state(&mut self, state: &PciConfigurationState) {
-        self.registers.clone_from_slice(state.registers.as_slice());
-        self.writable_bits
-            .clone_from_slice(state.writable_bits.as_slice());
-        self.bars.clone_from_slice(state.bars.as_slice());
-        self.rom_bar_addr = state.rom_bar_addr;
-        self.rom_bar_size = state.rom_bar_size;
-        self.rom_bar_used = state.rom_bar_used;
-        self.last_capability = state.last_capability;
-        self.msix_cap_reg_idx = state.msix_cap_reg_idx;
-    }
-
     /// Reads a 32bit register from `reg_idx` in the register map.
     pub fn read_reg(&self, reg_idx: usize) -> u32 {
         *(self.registers.get(reg_idx).unwrap_or(&0xffff_ffff))
@@ -1046,16 +1068,11 @@ impl Pausable for PciConfiguration {}
 
 impl Snapshottable for PciConfiguration {
     fn id(&self) -> String {
-        String::from("pci_configuration")
+        String::from(PCI_CONFIGURATION_ID)
     }
 
     fn snapshot(&mut self) -> std::result::Result<Snapshot, MigratableError> {
-        Snapshot::new_from_versioned_state(&self.id(), &self.state())
-    }
-
-    fn restore(&mut self, snapshot: Snapshot) -> std::result::Result<(), MigratableError> {
-        self.set_state(&snapshot.to_versioned_state(&self.id())?);
-        Ok(())
+        Snapshot::new_from_versioned_state(&self.state())
     }
 }
 
@@ -1178,6 +1195,7 @@ mod tests {
             0xABCD,
             0x2468,
             None,
+            None,
         );
 
         // Add two capabilities with different contents.
@@ -1234,6 +1252,7 @@ mod tests {
             0xABCD,
             0x2468,
             None,
+            None,
         );
 
         let class_reg = cfg.read_reg(2);

pci/src/lib.rs

@@ -19,12 +19,13 @@ pub use self::configuration::{
     PciBarConfiguration, PciBarPrefetchable, PciBarRegionType, PciCapability, PciCapabilityId,
     PciClassCode, PciConfiguration, PciExpressCapabilityId, PciHeaderType, PciMassStorageSubclass,
     PciNetworkControllerSubclass, PciProgrammingInterface, PciSerialBusSubClass, PciSubclass,
+    PCI_CONFIGURATION_ID,
 };
 pub use self::device::{
     BarReprogrammingParams, DeviceRelocation, Error as PciDeviceError, PciDevice,
 };
 pub use self::msi::{msi_num_enabled_vectors, MsiCap, MsiConfig};
-pub use self::msix::{MsixCap, MsixConfig, MsixTableEntry, MSIX_TABLE_ENTRY_SIZE};
+pub use self::msix::{MsixCap, MsixConfig, MsixTableEntry, MSIX_CONFIG_ID, MSIX_TABLE_ENTRY_SIZE};
 pub use self::vfio::{VfioPciDevice, VfioPciError};
 pub use self::vfio_user::{VfioUserDmaMapping, VfioUserPciDevice, VfioUserPciDeviceError};
 use serde::de::Visitor;

pci/src/msi.rs

@@ -3,7 +3,6 @@
 // SPDX-License-Identifier: Apache-2.0 OR BSD-3-Clause
 //
 
-use anyhow::anyhow;
 use byteorder::{ByteOrder, LittleEndian};
 use std::io;
 use std::sync::Arc;
@@ -39,13 +38,15 @@ pub fn msi_num_enabled_vectors(msg_ctl: u16) -> usize {
 }
 
 #[derive(Error, Debug)]
-enum Error {
+pub enum Error {
     #[error("Failed enabling the interrupt route: {0}")]
     EnableInterruptRoute(io::Error),
     #[error("Failed updating the interrupt route: {0}")]
     UpdateInterruptRoute(io::Error),
 }
 
+pub const MSI_CONFIG_ID: &str = "msi_config";
+
 #[derive(Clone, Copy, Default, Versionize)]
 pub struct MsiCap {
     // Message Control Register
@@ -172,7 +173,7 @@ impl MsiCap {
 }
 
 #[derive(Versionize)]
-struct MsiConfigState {
+pub struct MsiConfigState {
     cap: MsiCap,
 }
 
@@ -184,51 +185,53 @@ pub struct MsiConfig {
 }
 
 impl MsiConfig {
-    pub fn new(msg_ctl: u16, interrupt_source_group: Arc<dyn InterruptSourceGroup>) -> Self {
-        let cap = MsiCap {
-            msg_ctl,
-            ..Default::default()
+    pub fn new(
+        msg_ctl: u16,
+        interrupt_source_group: Arc<dyn InterruptSourceGroup>,
+        state: Option<MsiConfigState>,
+    ) -> Result<Self, Error> {
+        let cap = if let Some(state) = state {
+            if state.cap.enabled() {
+                for idx in 0..state.cap.num_enabled_vectors() {
+                    let config = MsiIrqSourceConfig {
+                        high_addr: state.cap.msg_addr_hi,
+                        low_addr: state.cap.msg_addr_lo,
+                        data: state.cap.msg_data as u32,
+                        devid: 0,
+                    };
+
+                    interrupt_source_group
+                        .update(
+                            idx as InterruptIndex,
+                            InterruptSourceConfig::MsiIrq(config),
+                            state.cap.vector_masked(idx),
+                        )
+                        .map_err(Error::UpdateInterruptRoute)?;
+                }
+
+                interrupt_source_group
+                    .enable()
+                    .map_err(Error::EnableInterruptRoute)?;
+            }
+
+            state.cap
+        } else {
+            MsiCap {
+                msg_ctl,
+                ..Default::default()
+            }
         };
 
-        MsiConfig {
+        Ok(MsiConfig {
             cap,
             interrupt_source_group,
-        }
+        })
     }
 
     fn state(&self) -> MsiConfigState {
         MsiConfigState { cap: self.cap }
     }
 
-    fn set_state(&mut self, state: &MsiConfigState) -> Result<(), Error> {
-        self.cap = state.cap;
-
-        if self.enabled() {
-            for idx in 0..self.num_enabled_vectors() {
-                let config = MsiIrqSourceConfig {
-                    high_addr: self.cap.msg_addr_hi,
-                    low_addr: self.cap.msg_addr_lo,
-                    data: self.cap.msg_data as u32,
-                    devid: 0,
-                };
-
-                self.interrupt_source_group
-                    .update(
-                        idx as InterruptIndex,
-                        InterruptSourceConfig::MsiIrq(config),
-                        self.cap.vector_masked(idx),
-                    )
-                    .map_err(Error::UpdateInterruptRoute)?;
-            }
-
-            self.interrupt_source_group
-                .enable()
-                .map_err(Error::EnableInterruptRoute)?;
-        }
-
-        Ok(())
-    }
-
     pub fn enabled(&self) -> bool {
         self.cap.enabled()
     }
@@ -281,21 +284,10 @@ impl Pausable for MsiConfig {}
 
 impl Snapshottable for MsiConfig {
     fn id(&self) -> String {
-        String::from("msi_config")
+        String::from(MSI_CONFIG_ID)
     }
 
     fn snapshot(&mut self) -> std::result::Result<Snapshot, MigratableError> {
-        Snapshot::new_from_versioned_state(&self.id(), &self.state())
-    }
-
-    fn restore(&mut self, snapshot: Snapshot) -> std::result::Result<(), MigratableError> {
-        self.set_state(&snapshot.to_versioned_state(&self.id())?)
-            .map_err(|e| {
-                MigratableError::Restore(anyhow!(
-                    "Could not restore state for {}: {:?}",
-                    self.id(),
-                    e
-                ))
-            })
+        Snapshot::new_from_versioned_state(&self.state())
     }
 }

pci/src/msix.rs

@@ -4,7 +4,6 @@
 //
 
 use crate::{PciCapability, PciCapabilityId};
-use anyhow::anyhow;
 use byteorder::{ByteOrder, LittleEndian};
 use std::io;
 use std::result;
@@ -26,9 +25,10 @@ const MSIX_ENABLE_BIT: u8 = 15;
 const FUNCTION_MASK_MASK: u16 = (1 << FUNCTION_MASK_BIT) as u16;
 const MSIX_ENABLE_MASK: u16 = (1 << MSIX_ENABLE_BIT) as u16;
 pub const MSIX_TABLE_ENTRY_SIZE: usize = 16;
+pub const MSIX_CONFIG_ID: &str = "msix_config";
 
 #[derive(Debug)]
-enum Error {
+pub enum Error {
     /// Failed enabling the interrupt route.
     EnableInterruptRoute(io::Error),
     /// Failed updating the interrupt route.
@@ -61,7 +61,7 @@ impl Default for MsixTableEntry {
 }
 
 #[derive(Versionize)]
-struct MsixConfigState {
+pub struct MsixConfigState {
     table_entries: Vec<MsixTableEntry>,
     pba_entries: Vec<u64>,
     masked: bool,
@@ -84,23 +84,62 @@ impl MsixConfig {
         msix_vectors: u16,
         interrupt_source_group: Arc<dyn InterruptSourceGroup>,
         devid: u32,
-    ) -> Self {
+        state: Option<MsixConfigState>,
+    ) -> result::Result<Self, Error> {
         assert!(msix_vectors <= MAX_MSIX_VECTORS_PER_DEVICE);
 
-        let mut table_entries: Vec<MsixTableEntry> = Vec::new();
-        table_entries.resize_with(msix_vectors as usize, Default::default);
-        let mut pba_entries: Vec<u64> = Vec::new();
-        let num_pba_entries: usize = ((msix_vectors as usize) / BITS_PER_PBA_ENTRY) + 1;
-        pba_entries.resize_with(num_pba_entries, Default::default);
+        let (table_entries, pba_entries, masked, enabled) = if let Some(state) = state {
+            if state.enabled && !state.masked {
+                for (idx, table_entry) in state.table_entries.iter().enumerate() {
+                    if table_entry.masked() {
+                        continue;
+                    }
 
-        MsixConfig {
+                    let config = MsiIrqSourceConfig {
+                        high_addr: table_entry.msg_addr_hi,
+                        low_addr: table_entry.msg_addr_lo,
+                        data: table_entry.msg_data,
+                        devid,
+                    };
+
+                    interrupt_source_group
+                        .update(
+                            idx as InterruptIndex,
+                            InterruptSourceConfig::MsiIrq(config),
+                            state.masked,
+                        )
+                        .map_err(Error::UpdateInterruptRoute)?;
+
+                    interrupt_source_group
+                        .enable()
+                        .map_err(Error::EnableInterruptRoute)?;
+                }
+            }
+
+            (
+                state.table_entries,
+                state.pba_entries,
+                state.masked,
+                state.enabled,
+            )
+        } else {
+            let mut table_entries: Vec<MsixTableEntry> = Vec::new();
+            table_entries.resize_with(msix_vectors as usize, Default::default);
+            let mut pba_entries: Vec<u64> = Vec::new();
+            let num_pba_entries: usize = ((msix_vectors as usize) / BITS_PER_PBA_ENTRY) + 1;
+            pba_entries.resize_with(num_pba_entries, Default::default);
+
+            (table_entries, pba_entries, true, false)
+        };
+
+        Ok(MsixConfig {
             table_entries,
             pba_entries,
             devid,
             interrupt_source_group,
-            masked: true,
-            enabled: false,
-        }
+            masked,
+            enabled,
+        })
     }
 
     fn state(&self) -> MsixConfigState {
@@ -112,42 +151,6 @@ impl MsixConfig {
         }
     }
 
-    fn set_state(&mut self, state: &MsixConfigState) -> result::Result<(), Error> {
-        self.table_entries = state.table_entries.clone();
-        self.pba_entries = state.pba_entries.clone();
-        self.masked = state.masked;
-        self.enabled = state.enabled;
-
-        if self.enabled && !self.masked {
-            for (idx, table_entry) in self.table_entries.iter().enumerate() {
-                if table_entry.masked() {
-                    continue;
-                }
-
-                let config = MsiIrqSourceConfig {
-                    high_addr: table_entry.msg_addr_hi,
-                    low_addr: table_entry.msg_addr_lo,
-                    data: table_entry.msg_data,
-                    devid: self.devid,
-                };
-
-                self.interrupt_source_group
-                    .update(
-                        idx as InterruptIndex,
-                        InterruptSourceConfig::MsiIrq(config),
-                        self.masked,
-                    )
-                    .map_err(Error::UpdateInterruptRoute)?;
-
-                self.interrupt_source_group
-                    .enable()
-                    .map_err(Error::EnableInterruptRoute)?;
-            }
-        }
-
-        Ok(())
-    }
-
     pub fn masked(&self) -> bool {
         self.masked
     }
@@ -426,22 +429,11 @@ impl Pausable for MsixConfig {}
 
 impl Snapshottable for MsixConfig {
     fn id(&self) -> String {
-        String::from("msix_config")
+        String::from(MSIX_CONFIG_ID)
     }
 
     fn snapshot(&mut self) -> std::result::Result<Snapshot, MigratableError> {
-        Snapshot::new_from_versioned_state(&self.id(), &self.state())
-    }
-
-    fn restore(&mut self, snapshot: Snapshot) -> std::result::Result<(), MigratableError> {
-        self.set_state(&snapshot.to_versioned_state(&self.id())?)
-            .map_err(|e| {
-                MigratableError::Restore(anyhow!(
-                    "Could not restore state for {}: {:?}",
-                    self.id(),
-                    e
-                ))
-            })
+        Snapshot::new_from_versioned_state(&self.state())
     }
 }
 
@@ -519,6 +511,16 @@ impl MsixCap {
         self.pba & 0xffff_fff8
     }
 
+    pub fn table_set_offset(&mut self, addr: u32) {
+        self.table &= 0x7;
+        self.table += addr;
+    }
+
+    pub fn pba_set_offset(&mut self, addr: u32) {
+        self.pba &= 0x7;
+        self.pba += addr;
+    }
+
     pub fn table_bir(&self) -> u32 {
         self.table & 0x7
     }

pci/src/vfio.rs

@@ -3,11 +3,13 @@
 // SPDX-License-Identifier: Apache-2.0 OR BSD-3-Clause
 //
 
+use crate::msi::{MsiConfigState, MSI_CONFIG_ID};
+use crate::msix::MsixConfigState;
 use crate::{
     msi_num_enabled_vectors, BarReprogrammingParams, MsiCap, MsiConfig, MsixCap, MsixConfig,
     PciBarConfiguration, PciBarPrefetchable, PciBarRegionType, PciBdf, PciCapabilityId,
     PciClassCode, PciConfiguration, PciDevice, PciDeviceError, PciExpressCapabilityId,
-    PciHeaderType, PciSubclass, MSIX_TABLE_ENTRY_SIZE,
+    PciHeaderType, PciSubclass, MSIX_CONFIG_ID, MSIX_TABLE_ENTRY_SIZE, PCI_CONFIGURATION_ID,
 };
 use anyhow::anyhow;
 use byteorder::{ByteOrder, LittleEndian};
@@ -26,6 +28,9 @@ use vfio_bindings::bindings::vfio::*;
 use vfio_ioctls::{
     VfioContainer, VfioDevice, VfioIrq, VfioRegionInfoCap, VfioRegionSparseMmapArea,
 };
+use vm_allocator::page_size::{
+    align_page_size_down, align_page_size_up, is_4k_aligned, is_4k_multiple, is_page_size_aligned,
+};
 use vm_allocator::{AddressAllocator, SystemAllocator};
 use vm_device::interrupt::{
     InterruptIndex, InterruptManager, InterruptSourceGroup, MsiIrqGroupConfig,
@@ -37,6 +42,8 @@ use vm_migration::{
 };
 use vmm_sys_util::eventfd::EventFd;
 
+pub(crate) const VFIO_COMMON_ID: &str = "vfio_common";
+
 #[derive(Debug, Error)]
 pub enum VfioPciError {
     #[error("Failed to create user memory region: {0}")]
@@ -59,6 +66,14 @@ pub enum VfioPciError {
     RegionAlignment,
     #[error("Invalid region size")]
     RegionSize,
+    #[error("Failed to retrieve MsiConfigState: {0}")]
+    RetrieveMsiConfigState(#[source] anyhow::Error),
+    #[error("Failed to retrieve MsixConfigState: {0}")]
+    RetrieveMsixConfigState(#[source] anyhow::Error),
+    #[error("Failed to retrieve PciConfigurationState: {0}")]
+    RetrievePciConfigurationState(#[source] anyhow::Error),
+    #[error("Failed to retrieve VfioCommonState: {0}")]
+    RetrieveVfioCommonState(#[source] anyhow::Error),
 }
 
 #[derive(Copy, Clone)]
@@ -407,6 +422,109 @@ pub(crate) struct VfioCommon {
 }
 
 impl VfioCommon {
+    pub(crate) fn new(
+        msi_interrupt_manager: Arc<dyn InterruptManager<GroupConfig = MsiIrqGroupConfig>>,
+        legacy_interrupt_group: Option<Arc<dyn InterruptSourceGroup>>,
+        vfio_wrapper: Arc<dyn Vfio>,
+        subclass: &dyn PciSubclass,
+        bdf: PciBdf,
+        snapshot: Option<Snapshot>,
+    ) -> Result<Self, VfioPciError> {
+        let pci_configuration_state =
+            vm_migration::versioned_state_from_id(snapshot.as_ref(), PCI_CONFIGURATION_ID)
+                .map_err(|e| {
+                    VfioPciError::RetrievePciConfigurationState(anyhow!(
+                        "Failed to get PciConfigurationState from Snapshot: {}",
+                        e
+                    ))
+                })?;
+
+        let configuration = PciConfiguration::new(
+            0,
+            0,
+            0,
+            PciClassCode::Other,
+            subclass,
+            None,
+            PciHeaderType::Device,
+            0,
+            0,
+            None,
+            pci_configuration_state,
+        );
+
+        let mut vfio_common = VfioCommon {
+            mmio_regions: Vec::new(),
+            configuration,
+            interrupt: Interrupt {
+                intx: None,
+                msi: None,
+                msix: None,
+            },
+            msi_interrupt_manager,
+            legacy_interrupt_group,
+            vfio_wrapper,
+            patches: HashMap::new(),
+        };
+
+        let state: Option<VfioCommonState> = snapshot
+            .as_ref()
+            .map(|s| s.to_versioned_state())
+            .transpose()
+            .map_err(|e| {
+                VfioPciError::RetrieveVfioCommonState(anyhow!(
+                    "Failed to get VfioCommonState from Snapshot: {}",
+                    e
+                ))
+            })?;
+        let msi_state = vm_migration::versioned_state_from_id(snapshot.as_ref(), MSI_CONFIG_ID)
+            .map_err(|e| {
+                VfioPciError::RetrieveMsiConfigState(anyhow!(
+                    "Failed to get MsiConfigState from Snapshot: {}",
+                    e
+                ))
+            })?;
+        let msix_state = vm_migration::versioned_state_from_id(snapshot.as_ref(), MSIX_CONFIG_ID)
+            .map_err(|e| {
+            VfioPciError::RetrieveMsixConfigState(anyhow!(
+                "Failed to get MsixConfigState from Snapshot: {}",
+                e
+            ))
+        })?;
+
+        if let Some(state) = state.as_ref() {
+            vfio_common.set_state(state, msi_state, msix_state)?;
+        } else {
+            vfio_common.parse_capabilities(bdf);
+            vfio_common.initialize_legacy_interrupt()?;
+        }
+
+        Ok(vfio_common)
+    }
+
+    /// In case msix table offset is not page size aligned, we need do some fixup to achive it.
+    /// Becuse we don't want the MMIO RW region and trap region overlap each other.
+    fn fixup_msix_region(&mut self, bar_id: u32, region_size: u64) -> u64 {
+        let msix = self.interrupt.msix.as_mut().unwrap();
+        let msix_cap = &mut msix.cap;
+
+        // Suppose table_bir equals to pba_bir here. Am I right?
+        let (table_offset, table_size) = msix_cap.table_range();
+        if is_page_size_aligned(table_offset) || msix_cap.table_bir() != bar_id {
+            return region_size;
+        }
+
+        let (pba_offset, pba_size) = msix_cap.pba_range();
+        let msix_sz = align_page_size_up(table_size + pba_size);
+        // Expand region to hold RW and trap region which both page size aligned
+        let size = std::cmp::max(region_size * 2, msix_sz * 2);
+        // let table starts from the middle of the region
+        msix_cap.table_set_offset((size / 2) as u32);
+        msix_cap.pba_set_offset((size / 2 + pba_offset - table_offset) as u32);
+
+        size
+    }
+
     pub(crate) fn allocate_bars(
         &mut self,
         allocator: &Arc<Mutex<SystemAllocator>>,
@@ -570,7 +688,9 @@ impl VfioCommon {
                         .ok_or(PciDeviceError::IoAllocationFailed(region_size))?
                 }
                 PciBarRegionType::Memory64BitRegion => {
-                    // BAR allocation must be naturally aligned
+                    // We need do some fixup to keep MMIO RW region and msix cap region page size
+                    // aligned.
+                    region_size = self.fixup_msix_region(bar_id, region_size);
                     mmio_allocator
                         .allocate(
                             restored_bar_addr,
@@ -656,7 +776,13 @@ impl VfioCommon {
         }
     }
 
-    pub(crate) fn initialize_msix(&mut self, msix_cap: MsixCap, cap_offset: u32, bdf: PciBdf) {
+    pub(crate) fn initialize_msix(
+        &mut self,
+        msix_cap: MsixCap,
+        cap_offset: u32,
+        bdf: PciBdf,
+        state: Option<MsixConfigState>,
+    ) {
         let interrupt_source_group = self
             .msi_interrupt_manager
             .create_group(MsiIrqGroupConfig {
@@ -669,7 +795,9 @@ impl VfioCommon {
             msix_cap.table_size(),
             interrupt_source_group.clone(),
             bdf.into(),
-        );
+            state,
+        )
+        .unwrap();
 
         self.interrupt.msix = Some(VfioMsix {
             bar: msix_config,
@@ -683,7 +811,12 @@ impl VfioCommon {
         self.vfio_wrapper.read_config_word((cap + 2).into())
     }
 
-    pub(crate) fn initialize_msi(&mut self, msg_ctl: u16, cap_offset: u32) {
+    pub(crate) fn initialize_msi(
+        &mut self,
+        msg_ctl: u16,
+        cap_offset: u32,
+        state: Option<MsiConfigState>,
+    ) {
         let interrupt_source_group = self
             .msi_interrupt_manager
             .create_group(MsiIrqGroupConfig {
@@ -692,7 +825,7 @@ impl VfioCommon {
             })
             .unwrap();
 
-        let msi_config = MsiConfig::new(msg_ctl, interrupt_source_group.clone());
+        let msi_config = MsiConfig::new(msg_ctl, interrupt_source_group.clone(), state).unwrap();
 
         self.interrupt.msi = Some(VfioMsi {
             cfg: msi_config,
@@ -701,6 +834,23 @@ impl VfioCommon {
         });
     }
 
+    pub(crate) fn get_msix_cap_idx(&self) -> Option<usize> {
+        let mut cap_next = self
+            .vfio_wrapper
+            .read_config_byte(PCI_CONFIG_CAPABILITY_OFFSET);
+
+        while cap_next != 0 {
+            let cap_id = self.vfio_wrapper.read_config_byte(cap_next.into());
+            if PciCapabilityId::from(cap_id) == PciCapabilityId::MsiX {
+                return Some(cap_next as usize);
+            } else {
+                cap_next = self.vfio_wrapper.read_config_byte((cap_next + 1).into());
+            }
+        }
+
+        None
+    }
+
     pub(crate) fn parse_capabilities(&mut self, bdf: PciBdf) {
         let mut cap_next = self
             .vfio_wrapper
@@ -719,7 +869,7 @@ impl VfioCommon {
                             // Parse capability only if the VFIO device
                             // supports MSI.
                             let msg_ctl = self.parse_msi_capabilities(cap_next);
-                            self.initialize_msi(msg_ctl, cap_next as u32);
+                            self.initialize_msi(msg_ctl, cap_next as u32, None);
                         }
                     }
                 }
@@ -730,7 +880,7 @@ impl VfioCommon {
                             // Parse capability only if the VFIO device
                             // supports MSI-X.
                             let msix_cap = self.parse_msix_capabilities(cap_next);
-                            self.initialize_msix(msix_cap, cap_next as u32, bdf);
+                            self.initialize_msix(msix_cap, cap_next as u32, bdf, None);
                         }
                     }
                 }
@@ -1055,6 +1205,15 @@ impl VfioCommon {
             return self.configuration.read_reg(reg_idx);
         }
 
+        if let Some(id) = self.get_msix_cap_idx() {
+            let msix = self.interrupt.msix.as_mut().unwrap();
+            if reg_idx * 4 == id + 4 {
+                return msix.cap.table;
+            } else if reg_idx * 4 == id + 8 {
+                return msix.cap.pba;
+            }
+        }
+
         // Since we don't support passing multi-functions devices, we should
         // mask the multi-function bit, bit 7 of the Header Type byte on the
         // register 3.
@@ -1097,7 +1256,12 @@ impl VfioCommon {
         }
     }
 
-    fn set_state(&mut self, state: &VfioCommonState) -> Result<(), VfioPciError> {
+    fn set_state(
+        &mut self,
+        state: &VfioCommonState,
+        msi_state: Option<MsiConfigState>,
+        msix_state: Option<MsixConfigState>,
+    ) -> Result<(), VfioPciError> {
         if let (Some(intx), Some(interrupt_source_group)) =
             (&state.intx_state, self.legacy_interrupt_group.clone())
         {
@@ -1112,11 +1276,11 @@ impl VfioCommon {
         }
 
         if let Some(msi) = &state.msi_state {
-            self.initialize_msi(msi.cap.msg_ctl, msi.cap_offset);
+            self.initialize_msi(msi.cap.msg_ctl, msi.cap_offset, msi_state);
         }
 
         if let Some(msix) = &state.msix_state {
-            self.initialize_msix(msix.cap, msix.cap_offset, msix.bdf.into());
+            self.initialize_msix(msix.cap, msix.cap_offset, msix.bdf.into(), msix_state);
         }
 
         Ok(())
@@ -1127,73 +1291,27 @@ impl Pausable for VfioCommon {}
 
 impl Snapshottable for VfioCommon {
     fn id(&self) -> String {
-        String::from("vfio_common")
+        String::from(VFIO_COMMON_ID)
     }
 
     fn snapshot(&mut self) -> std::result::Result<Snapshot, MigratableError> {
-        let mut vfio_common_snapshot =
-            Snapshot::new_from_versioned_state(&self.id(), &self.state())?;
+        let mut vfio_common_snapshot = Snapshot::new_from_versioned_state(&self.state())?;
 
         // Snapshot PciConfiguration
-        vfio_common_snapshot.add_snapshot(self.configuration.snapshot()?);
+        vfio_common_snapshot.add_snapshot(self.configuration.id(), self.configuration.snapshot()?);
 
         // Snapshot MSI
         if let Some(msi) = &mut self.interrupt.msi {
-            vfio_common_snapshot.add_snapshot(msi.cfg.snapshot()?);
+            vfio_common_snapshot.add_snapshot(msi.cfg.id(), msi.cfg.snapshot()?);
         }
 
         // Snapshot MSI-X
         if let Some(msix) = &mut self.interrupt.msix {
-            vfio_common_snapshot.add_snapshot(msix.bar.snapshot()?);
+            vfio_common_snapshot.add_snapshot(msix.bar.id(), msix.bar.snapshot()?);
         }
 
         Ok(vfio_common_snapshot)
     }
-
-    fn restore(&mut self, snapshot: Snapshot) -> std::result::Result<(), MigratableError> {
-        if let Some(vfio_common_section) = snapshot
-            .snapshot_data
-            .get(&format!("{}-section", self.id()))
-        {
-            // It has to be invoked first as we want Interrupt to be initialized
-            // correctly before we try to restore MSI and MSI-X configurations.
-            self.set_state(&vfio_common_section.to_versioned_state()?)
-                .map_err(|e| {
-                    MigratableError::Restore(anyhow!("Could not restore VFIO_COMMON state {:?}", e))
-                })?;
-
-            // Restore PciConfiguration
-            if let Some(pci_config_snapshot) = snapshot.snapshots.get(&self.configuration.id()) {
-                self.configuration.restore(*pci_config_snapshot.clone())?;
-            }
-
-            // Restore MSI
-            if let Some(msi) = &mut self.interrupt.msi {
-                if let Some(msi_snapshot) = snapshot.snapshots.get(&msi.cfg.id()) {
-                    msi.cfg.restore(*msi_snapshot.clone())?;
-                }
-                if msi.cfg.enabled() {
-                    self.enable_msi().unwrap();
-                }
-            }
-
-            // Restore MSI-X
-            if let Some(msix) = &mut self.interrupt.msix {
-                if let Some(msix_snapshot) = snapshot.snapshots.get(&msix.bar.id()) {
-                    msix.bar.restore(*msix_snapshot.clone())?;
-                }
-                if msix.bar.enabled() {
-                    self.enable_msix().unwrap();
-                }
-            }
-
-            return Ok(());
-        }
-
-        Err(MigratableError::Restore(anyhow!(
-            "Could not find VFIO_COMMON snapshot section"
-        )))
-    }
 }
 
 /// VfioPciDevice represents a VFIO PCI device.
@@ -1224,48 +1342,22 @@ impl VfioPciDevice {
         legacy_interrupt_group: Option<Arc<dyn InterruptSourceGroup>>,
         iommu_attached: bool,
         bdf: PciBdf,
-        restoring: bool,
         memory_slot: Arc<dyn Fn() -> u32 + Send + Sync>,
+        snapshot: Option<Snapshot>,
     ) -> Result<Self, VfioPciError> {
         let device = Arc::new(device);
         device.reset();
 
-        let configuration = PciConfiguration::new(
-            0,
-            0,
-            0,
-            PciClassCode::Other,
-            &PciVfioSubclass::VfioSubclass,
-            None,
-            PciHeaderType::Device,
-            0,
-            0,
-            None,
-        );
-
         let vfio_wrapper = VfioDeviceWrapper::new(Arc::clone(&device));
 
-        let mut common = VfioCommon {
-            mmio_regions: Vec::new(),
-            configuration,
-            interrupt: Interrupt {
-                intx: None,
-                msi: None,
-                msix: None,
-            },
+        let common = VfioCommon::new(
             msi_interrupt_manager,
             legacy_interrupt_group,
-            vfio_wrapper: Arc::new(vfio_wrapper) as Arc<dyn Vfio>,
-            patches: HashMap::new(),
-        };
-
-        // No need to parse capabilities from the device if on the restore path.
-        // The initialization will be performed later when restore() will be
-        // called.
-        if !restoring {
-            common.parse_capabilities(bdf);
-            common.initialize_legacy_interrupt()?;
-        }
+            Arc::new(vfio_wrapper) as Arc<dyn Vfio>,
+            &PciVfioSubclass::VfioSubclass,
+            bdf,
+            vm_migration::snapshot_from_id(snapshot.as_ref(), VFIO_COMMON_ID),
+        )?;
 
         let vfio_pci_device = VfioPciDevice {
             id,
@@ -1284,20 +1376,6 @@ impl VfioPciDevice {
         self.iommu_attached
     }
 
-    fn align_page_size(address: u64) -> u64 {
-        // SAFETY: FFI call. Trivially safe.
-        let page_size = unsafe { sysconf(_SC_PAGESIZE) as u64 };
-        (address + page_size - 1) & !(page_size - 1)
-    }
-
-    fn is_4k_aligned(address: u64) -> bool {
-        (address & 0xfff) == 0
-    }
-
-    fn is_4k_multiple(size: u64) -> bool {
-        (size & 0xfff) == 0
-    }
-
     fn generate_sparse_areas(
         caps: &[VfioRegionInfoCap],
         region_index: u32,
@@ -1309,14 +1387,14 @@ impl VfioPciDevice {
             match cap {
                 VfioRegionInfoCap::SparseMmap(sparse_mmap) => return Ok(sparse_mmap.areas.clone()),
                 VfioRegionInfoCap::MsixMappable => {
-                    if !Self::is_4k_aligned(region_start) {
+                    if !is_4k_aligned(region_start) {
                         error!(
                             "Region start address 0x{:x} must be at least aligned on 4KiB",
                             region_start
                         );
                         return Err(VfioPciError::RegionAlignment);
                     }
-                    if !Self::is_4k_multiple(region_size) {
+                    if !is_4k_multiple(region_size) {
                         error!(
                             "Region size 0x{:x} must be at least a multiple of 4KiB",
                             region_size
@@ -1328,7 +1406,8 @@ impl VfioPciDevice {
                     // the MSI-X PBA table, we must calculate the subregions
                     // around them, leading to a list of sparse areas.
                     // We want to make sure we will still trap MMIO accesses
-                    // to these MSI-X specific ranges.
+                    // to these MSI-X specific ranges. If these region don't align
+                    // with pagesize, we can achive it by enlarging its range.
                     //
                     // Using a BtreeMap as the list provided through the iterator is sorted
                     // by key. This ensures proper split of the whole region.
@@ -1336,10 +1415,14 @@ impl VfioPciDevice {
                     if let Some(msix) = vfio_msix {
                         if region_index == msix.cap.table_bir() {
                             let (offset, size) = msix.cap.table_range();
+                            let offset = align_page_size_down(offset);
+                            let size = align_page_size_up(size);
                             inter_ranges.insert(offset, size);
                         }
                         if region_index == msix.cap.pba_bir() {
                             let (offset, size) = msix.cap.pba_range();
+                            let offset = align_page_size_down(offset);
+                            let size = align_page_size_up(size);
                             inter_ranges.insert(offset, size);
                         }
                     }
@@ -1347,21 +1430,19 @@ impl VfioPciDevice {
                     let mut sparse_areas = Vec::new();
                     let mut current_offset = 0;
                     for (range_offset, range_size) in inter_ranges {
-                        let range_offset = Self::align_page_size(range_offset);
                         if range_offset > current_offset {
                             sparse_areas.push(VfioRegionSparseMmapArea {
                                 offset: current_offset,
                                 size: range_offset - current_offset,
                             });
                         }
-
-                        current_offset = Self::align_page_size(range_offset + range_size);
+                        current_offset = align_page_size_down(range_offset + range_size);
                     }
 
                     if region_size > current_offset {
                         sparse_areas.push(VfioRegionSparseMmapArea {
-                            offset: Self::align_page_size(current_offset),
-                            size: Self::align_page_size(region_size - current_offset),
+                            offset: current_offset,
+                            size: region_size - current_offset,
                         });
                     }
 
@@ -1431,6 +1512,7 @@ impl VfioPciDevice {
                 )?;
 
                 for area in sparse_areas.iter() {
+                    // SAFETY: FFI call with correct arguments
                     let host_addr = unsafe {
                         libc::mmap(
                             null_mut(),
@@ -1452,6 +1534,15 @@ impl VfioPciDevice {
                         return Err(VfioPciError::MmapArea);
                     }
 
+                    if !is_page_size_aligned(area.size) || !is_page_size_aligned(area.offset) {
+                        warn!(
+                            "Could not mmap sparse area that is not page size aligned (offset = 0x{:x}, size = 0x{:x})",
+                            area.offset,
+                            area.size,
+                            );
+                        return Ok(());
+                    }
+
                     let user_memory_region = UserMemoryRegion {
                         slot: (self.memory_slot)(),
                         start: region.start.0 + area.offset,
@@ -1497,6 +1588,7 @@ impl VfioPciDevice {
                     error!("Could not remove the userspace memory region: {}", e);
                 }
 
+                // SAFETY: FFI call with correct arguments
                 let ret = unsafe {
                     libc::munmap(
                         user_memory_region.host_addr as *mut libc::c_void,
@@ -1708,28 +1800,13 @@ impl Snapshottable for VfioPciDevice {
     }
 
     fn snapshot(&mut self) -> std::result::Result<Snapshot, MigratableError> {
-        let mut vfio_pci_dev_snapshot = Snapshot::new(&self.id);
+        let mut vfio_pci_dev_snapshot = Snapshot::default();
 
         // Snapshot VfioCommon
-        vfio_pci_dev_snapshot.add_snapshot(self.common.snapshot()?);
+        vfio_pci_dev_snapshot.add_snapshot(self.common.id(), self.common.snapshot()?);
 
         Ok(vfio_pci_dev_snapshot)
     }
-
-    fn restore(&mut self, snapshot: Snapshot) -> std::result::Result<(), MigratableError> {
-        // Restore VfioCommon
-        if let Some(vfio_common_snapshot) = snapshot.snapshots.get(&self.common.id()) {
-            self.common.restore(*vfio_common_snapshot.clone())?;
-            self.map_mmio_regions().map_err(|e| {
-                MigratableError::Restore(anyhow!(
-                    "Could not map MMIO regions for VfioPciDevice on restore {:?}",
-                    e
-                ))
-            })?;
-        }
-
-        Ok(())
-    }
 }
 impl Transportable for VfioPciDevice {}
 impl Migratable for VfioPciDevice {}

pci/src/vfio_user.rs

@@ -3,15 +3,11 @@
 // SPDX-License-Identifier: Apache-2.0
 //
 
-use crate::vfio::{Interrupt, UserMemoryRegion, Vfio, VfioCommon, VfioError};
+use crate::vfio::{UserMemoryRegion, Vfio, VfioCommon, VfioError, VFIO_COMMON_ID};
 use crate::{BarReprogrammingParams, PciBarConfiguration, VfioPciError};
-use crate::{
-    PciBdf, PciClassCode, PciConfiguration, PciDevice, PciDeviceError, PciHeaderType, PciSubclass,
-};
-use anyhow::anyhow;
+use crate::{PciBdf, PciDevice, PciDeviceError, PciSubclass};
 use hypervisor::HypervisorVmError;
 use std::any::Any;
-use std::collections::HashMap;
 use std::os::unix::prelude::AsRawFd;
 use std::ptr::null_mut;
 use std::sync::{Arc, Barrier, Mutex};
@@ -51,6 +47,8 @@ pub enum VfioUserPciDeviceError {
     DmaUnmap(#[source] VfioUserError),
     #[error("Failed to initialize legacy interrupts: {0}")]
     InitializeLegacyInterrupts(#[source] VfioPciError),
+    #[error("Failed to create VfioCommon: {0}")]
+    CreateVfioCommon(#[source] VfioPciError),
 }
 
 #[derive(Copy, Clone)]
@@ -73,22 +71,9 @@ impl VfioUserPciDevice {
         msi_interrupt_manager: Arc<dyn InterruptManager<GroupConfig = MsiIrqGroupConfig>>,
         legacy_interrupt_group: Option<Arc<dyn InterruptSourceGroup>>,
         bdf: PciBdf,
-        restoring: bool,
         memory_slot: Arc<dyn Fn() -> u32 + Send + Sync>,
+        snapshot: Option<Snapshot>,
     ) -> Result<Self, VfioUserPciDeviceError> {
-        // This is used for the BAR and capabilities only
-        let configuration = PciConfiguration::new(
-            0,
-            0,
-            0,
-            PciClassCode::Other,
-            &PciVfioUserSubclass::VfioUserSubclass,
-            None,
-            PciHeaderType::Device,
-            0,
-            0,
-            None,
-        );
         let resettable = client.lock().unwrap().resettable();
         if resettable {
             client
@@ -102,29 +87,15 @@ impl VfioUserPciDevice {
             client: client.clone(),
         };
 
-        let mut common = VfioCommon {
-            mmio_regions: Vec::new(),
-            configuration,
-            interrupt: Interrupt {
-                intx: None,
-                msi: None,
-                msix: None,
-            },
+        let common = VfioCommon::new(
             msi_interrupt_manager,
             legacy_interrupt_group,
-            vfio_wrapper: Arc::new(vfio_wrapper) as Arc<dyn Vfio>,
-            patches: HashMap::new(),
-        };
-
-        // No need to parse capabilities from the device if on the restore path.
-        // The initialization will be performed later when restore() will be
-        // called.
-        if !restoring {
-            common.parse_capabilities(bdf);
-            common
-                .initialize_legacy_interrupt()
-                .map_err(VfioUserPciDeviceError::InitializeLegacyInterrupts)?;
-        }
+            Arc::new(vfio_wrapper) as Arc<dyn Vfio>,
+            &PciVfioUserSubclass::VfioUserSubclass,
+            bdf,
+            vm_migration::snapshot_from_id(snapshot.as_ref(), VFIO_COMMON_ID),
+        )
+        .map_err(VfioUserPciDeviceError::CreateVfioCommon)?;
 
         Ok(Self {
             id,
@@ -181,6 +152,7 @@ impl VfioUserPciDevice {
                 };
 
                 for s in mmaps.iter() {
+                    // SAFETY: FFI call with correct arguments
                     let host_addr = unsafe {
                         libc::mmap(
                             null_mut(),
@@ -247,6 +219,7 @@ impl VfioUserPciDevice {
                 }
 
                 // Remove mmaps
+                // SAFETY: FFI call with correct arguments
                 let ret = unsafe {
                     libc::munmap(
                         user_memory_region.host_addr as *mut libc::c_void,
@@ -562,28 +535,13 @@ impl Snapshottable for VfioUserPciDevice {
     }
 
     fn snapshot(&mut self) -> std::result::Result<Snapshot, MigratableError> {
-        let mut vfio_pci_dev_snapshot = Snapshot::new(&self.id);
+        let mut vfio_pci_dev_snapshot = Snapshot::default();
 
         // Snapshot VfioCommon
-        vfio_pci_dev_snapshot.add_snapshot(self.common.snapshot()?);
+        vfio_pci_dev_snapshot.add_snapshot(self.common.id(), self.common.snapshot()?);
 
         Ok(vfio_pci_dev_snapshot)
     }
-
-    fn restore(&mut self, snapshot: Snapshot) -> std::result::Result<(), MigratableError> {
-        // Restore VfioCommon
-        if let Some(vfio_common_snapshot) = snapshot.snapshots.get(&self.common.id()) {
-            self.common.restore(*vfio_common_snapshot.clone())?;
-            self.map_mmio_regions().map_err(|e| {
-                MigratableError::Restore(anyhow!(
-                    "Could not map MMIO regions for VfioUserPciDevice on restore {:?}",
-                    e
-                ))
-            })?;
-        }
-
-        Ok(())
-    }
 }
 impl Transportable for VfioUserPciDevice {}
 impl Migratable for VfioUserPciDevice {}

performance-metrics/Cargo.toml

@@ -3,16 +3,13 @@ name = "performance-metrics"
 version = "0.1.0"
 authors = ["The Cloud Hypervisor Authors"]
 edition = "2021"
-build = "build.rs"
+build = "../build.rs"
 
 [dependencies]
-clap = { version = "4.0.29", features = ["wrap_help","cargo"] }
-dirs = "4.0.0"
-serde = { version = "1.0.150", features = ["rc", "derive"] }
-serde_json = "1.0.89"
+argh = "0.1.9"
+dirs = "5.0.0"
+serde = { version = "1.0.156", features = ["rc", "derive"] }
+serde_json = "1.0.96"
 test_infra = { path = "../test_infra" }
-thiserror = "1.0.37"
+thiserror = "1.0.39"
 wait-timeout = "0.2.0"
-
-[build-dependencies]
-clap = { version = "4.0.29", features = ["cargo"] }

performance-metrics/build.rs

@@ -1,26 +0,0 @@
-// Copyright © 2020 Intel Corporation
-//
-// SPDX-License-Identifier: Apache-2.0
-//
-
-#[macro_use(crate_version)]
-extern crate clap;
-
-use std::process::Command;
-
-fn main() {
-    let mut git_human_readable = "v".to_owned() + crate_version!();
-    if let Ok(git_out) = Command::new("git").args(["describe", "--dirty"]).output() {
-        if git_out.status.success() {
-            if let Ok(git_out_str) = String::from_utf8(git_out.stdout) {
-                git_human_readable = git_out_str;
-            }
-        }
-    }
-
-    // This println!() has a special behavior, as it will set the environment
-    // variable GIT_HUMAN_READABLE, so that it can be reused from the binary.
-    // Particularly, this is used from the main.rs to display the exact
-    // version information.
-    println!("cargo:rustc-env=GIT_HUMAN_READABLE={git_human_readable}");
-}

performance-metrics/src/main.rs

@@ -5,16 +5,14 @@
 
 // Custom harness to run performance tests
 extern crate test_infra;
-#[macro_use(crate_authors)]
-extern crate clap;
 
 mod performance_tests;
 
-use clap::{Arg, ArgAction, Command as ClapCommand};
+use argh::FromArgs;
 use performance_tests::*;
 use serde::{Deserialize, Serialize};
 use std::{
-    env, fmt,
+    fmt,
     process::Command,
     sync::{mpsc::channel, Arc},
     thread,
@@ -112,25 +110,30 @@ impl Default for MetricsReport {
 #[derive(Default)]
 pub struct PerformanceTestOverrides {
     test_iterations: Option<u32>,
+    test_timeout: Option<u32>,
 }
 
 impl fmt::Display for PerformanceTestOverrides {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         if let Some(test_iterations) = self.test_iterations {
-            write!(f, "test_iterations = {test_iterations}")?;
+            write!(f, "test_iterations = {test_iterations}, ")?;
+        }
+        if let Some(test_timeout) = self.test_timeout {
+            write!(f, "test_timeout = {test_timeout}")?;
         }
 
         Ok(())
     }
 }
 
+#[derive(Clone)]
 pub struct PerformanceTestControl {
     test_timeout: u32,
     test_iterations: u32,
     num_queues: Option<u32>,
     queue_size: Option<u32>,
-    net_rx: Option<bool>,
-    fio_ops: Option<FioOps>,
+    net_control: Option<(bool, bool)>, // First bool is for RX(true)/TX(false), second bool is for bandwidth or PPS
+    fio_control: Option<(FioOps, bool)>, // Second parameter controls whether we want bandwidth or IOPS
     num_boot_vcpus: Option<u8>,
 }
 
@@ -146,11 +149,13 @@ impl fmt::Display for PerformanceTestControl {
         if let Some(o) = self.queue_size {
             output = format!("{output}, queue_size = {o}");
         }
-        if let Some(o) = self.net_rx {
-            output = format!("{output}, net_rx = {o}");
+        if let Some(o) = self.net_control {
+            let (rx, bw) = o;
+            output = format!("{output}, rx = {rx}, bandwidth = {bw}");
         }
-        if let Some(o) = &self.fio_ops {
-            output = format!("{output}, fio_ops = {o}");
+        if let Some(o) = &self.fio_control {
+            let (ops, bw) = o;
+            output = format!("{output}, fio_ops = {ops}, bandwidth = {bw}");
         }
 
         write!(f, "{output}")
@@ -164,8 +169,8 @@ impl PerformanceTestControl {
             test_iterations: 5,
             num_queues: None,
             queue_size: None,
-            net_rx: None,
-            fio_ops: None,
+            net_control: None,
+            fio_control: None,
             num_boot_vcpus: Some(1),
         }
     }
@@ -188,7 +193,14 @@ impl PerformanceTest {
             .test_iterations
             .unwrap_or(self.control.test_iterations)
         {
-            metrics.push((self.func_ptr)(&self.control));
+            // update the timeout in control if passed explicitly and run testcase with it
+            if let Some(test_timeout) = overrides.test_timeout {
+                let mut control: PerformanceTestControl = self.control.clone();
+                control.test_timeout = test_timeout;
+                metrics.push((self.func_ptr)(&control));
+            } else {
+                metrics.push((self.func_ptr)(&self.control));
+            }
         }
 
         let mean = (self.unit_adjuster)(mean(&metrics).unwrap());
@@ -207,9 +219,9 @@ impl PerformanceTest {
 
     // Calculate the timeout for each test
     // Note: To cover the setup/cleanup time, 20s is added for each iteration of the test
-    pub fn calc_timeout(&self, test_iterations: &Option<u32>) -> u64 {
-        ((self.control.test_timeout + 20) * test_iterations.unwrap_or(self.control.test_iterations))
-            as u64
+    pub fn calc_timeout(&self, test_iterations: &Option<u32>, test_timeout: &Option<u32>) -> u64 {
+        ((test_timeout.unwrap_or(self.control.test_timeout) + 20)
+            * test_iterations.unwrap_or(self.control.test_iterations)) as u64
     }
 }
 
@@ -262,7 +274,7 @@ mod adjuster {
     }
 }
 
-const TEST_LIST: [PerformanceTest; 17] = [
+const TEST_LIST: [PerformanceTest; 29] = [
     PerformanceTest {
         name: "boot_time_ms",
         func_ptr: performance_boot_time,
@@ -321,7 +333,7 @@ const TEST_LIST: [PerformanceTest; 17] = [
         control: PerformanceTestControl {
             num_queues: Some(2),
             queue_size: Some(256),
-            net_rx: Some(true),
+            net_control: Some((true, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::bps_to_gbps,
@@ -332,7 +344,7 @@ const TEST_LIST: [PerformanceTest; 17] = [
         control: PerformanceTestControl {
             num_queues: Some(2),
             queue_size: Some(256),
-            net_rx: Some(false),
+            net_control: Some((false, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::bps_to_gbps,
@@ -343,7 +355,7 @@ const TEST_LIST: [PerformanceTest; 17] = [
         control: PerformanceTestControl {
             num_queues: Some(4),
             queue_size: Some(256),
-            net_rx: Some(true),
+            net_control: Some((true, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::bps_to_gbps,
@@ -354,18 +366,62 @@ const TEST_LIST: [PerformanceTest; 17] = [
         control: PerformanceTestControl {
             num_queues: Some(4),
             queue_size: Some(256),
-            net_rx: Some(false),
+            net_control: Some((false, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::bps_to_gbps,
     },
+    PerformanceTest {
+        name: "virtio_net_throughput_single_queue_rx_pps",
+        func_ptr: performance_net_throughput,
+        control: PerformanceTestControl {
+            num_queues: Some(2),
+            queue_size: Some(256),
+            net_control: Some((true, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
+    PerformanceTest {
+        name: "virtio_net_throughput_single_queue_tx_pps",
+        func_ptr: performance_net_throughput,
+        control: PerformanceTestControl {
+            num_queues: Some(2),
+            queue_size: Some(256),
+            net_control: Some((false, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
+    PerformanceTest {
+        name: "virtio_net_throughput_multi_queue_rx_pps",
+        func_ptr: performance_net_throughput,
+        control: PerformanceTestControl {
+            num_queues: Some(4),
+            queue_size: Some(256),
+            net_control: Some((true, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
+    PerformanceTest {
+        name: "virtio_net_throughput_multi_queue_tx_pps",
+        func_ptr: performance_net_throughput,
+        control: PerformanceTestControl {
+            num_queues: Some(4),
+            queue_size: Some(256),
+            net_control: Some((false, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
     PerformanceTest {
         name: "block_read_MiBps",
         func_ptr: performance_block_io,
         control: PerformanceTestControl {
             num_queues: Some(1),
             queue_size: Some(128),
-            fio_ops: Some(FioOps::Read),
+            fio_control: Some((FioOps::Read, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::Bps_to_MiBps,
@@ -376,7 +432,7 @@ const TEST_LIST: [PerformanceTest; 17] = [
         control: PerformanceTestControl {
             num_queues: Some(1),
             queue_size: Some(128),
-            fio_ops: Some(FioOps::Write),
+            fio_control: Some((FioOps::Write, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::Bps_to_MiBps,
@@ -387,7 +443,7 @@ const TEST_LIST: [PerformanceTest; 17] = [
         control: PerformanceTestControl {
             num_queues: Some(1),
             queue_size: Some(128),
-            fio_ops: Some(FioOps::RandomRead),
+            fio_control: Some((FioOps::RandomRead, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::Bps_to_MiBps,
@@ -398,7 +454,7 @@ const TEST_LIST: [PerformanceTest; 17] = [
         control: PerformanceTestControl {
             num_queues: Some(1),
             queue_size: Some(128),
-            fio_ops: Some(FioOps::RandomWrite),
+            fio_control: Some((FioOps::RandomWrite, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::Bps_to_MiBps,
@@ -409,7 +465,7 @@ const TEST_LIST: [PerformanceTest; 17] = [
         control: PerformanceTestControl {
             num_queues: Some(2),
             queue_size: Some(128),
-            fio_ops: Some(FioOps::Read),
+            fio_control: Some((FioOps::Read, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::Bps_to_MiBps,
@@ -420,7 +476,7 @@ const TEST_LIST: [PerformanceTest; 17] = [
         control: PerformanceTestControl {
             num_queues: Some(2),
             queue_size: Some(128),
-            fio_ops: Some(FioOps::Write),
+            fio_control: Some((FioOps::Write, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::Bps_to_MiBps,
@@ -431,7 +487,7 @@ const TEST_LIST: [PerformanceTest; 17] = [
         control: PerformanceTestControl {
             num_queues: Some(2),
             queue_size: Some(128),
-            fio_ops: Some(FioOps::RandomRead),
+            fio_control: Some((FioOps::RandomRead, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::Bps_to_MiBps,
@@ -442,11 +498,99 @@ const TEST_LIST: [PerformanceTest; 17] = [
         control: PerformanceTestControl {
             num_queues: Some(2),
             queue_size: Some(128),
-            fio_ops: Some(FioOps::RandomWrite),
+            fio_control: Some((FioOps::RandomWrite, true)),
             ..PerformanceTestControl::default()
         },
         unit_adjuster: adjuster::Bps_to_MiBps,
     },
+    PerformanceTest {
+        name: "block_read_IOPS",
+        func_ptr: performance_block_io,
+        control: PerformanceTestControl {
+            num_queues: Some(1),
+            queue_size: Some(128),
+            fio_control: Some((FioOps::Read, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
+    PerformanceTest {
+        name: "block_write_IOPS",
+        func_ptr: performance_block_io,
+        control: PerformanceTestControl {
+            num_queues: Some(1),
+            queue_size: Some(128),
+            fio_control: Some((FioOps::Write, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
+    PerformanceTest {
+        name: "block_random_read_IOPS",
+        func_ptr: performance_block_io,
+        control: PerformanceTestControl {
+            num_queues: Some(1),
+            queue_size: Some(128),
+            fio_control: Some((FioOps::RandomRead, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
+    PerformanceTest {
+        name: "block_random_write_IOPS",
+        func_ptr: performance_block_io,
+        control: PerformanceTestControl {
+            num_queues: Some(1),
+            queue_size: Some(128),
+            fio_control: Some((FioOps::RandomWrite, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
+    PerformanceTest {
+        name: "block_multi_queue_read_IOPS",
+        func_ptr: performance_block_io,
+        control: PerformanceTestControl {
+            num_queues: Some(2),
+            queue_size: Some(128),
+            fio_control: Some((FioOps::Read, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
+    PerformanceTest {
+        name: "block_multi_queue_write_IOPS",
+        func_ptr: performance_block_io,
+        control: PerformanceTestControl {
+            num_queues: Some(2),
+            queue_size: Some(128),
+            fio_control: Some((FioOps::Write, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
+    PerformanceTest {
+        name: "block_multi_queue_random_read_IOPS",
+        func_ptr: performance_block_io,
+        control: PerformanceTestControl {
+            num_queues: Some(2),
+            queue_size: Some(128),
+            fio_control: Some((FioOps::RandomRead, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
+    PerformanceTest {
+        name: "block_multi_queue_random_write_IOPS",
+        func_ptr: performance_block_io,
+        control: PerformanceTestControl {
+            num_queues: Some(2),
+            queue_size: Some(128),
+            fio_control: Some((FioOps::RandomWrite, false)),
+            ..PerformanceTestControl::default()
+        },
+        unit_adjuster: adjuster::identity,
+    },
 ];
 
 fn run_test_with_timeout(
@@ -455,6 +599,7 @@ fn run_test_with_timeout(
 ) -> Result<PerformanceTestResult, Error> {
     let (sender, receiver) = channel::<Result<PerformanceTestResult, Error>>();
     let test_iterations = overrides.test_iterations;
+    let test_timeout = overrides.test_timeout;
     let overrides = overrides.clone();
     thread::spawn(move || {
         println!(
@@ -477,7 +622,7 @@ fn run_test_with_timeout(
     });
 
     // Todo: Need to cleanup/kill all hanging child processes
-    let test_timeout = test.calc_timeout(&test_iterations);
+    let test_timeout = test.calc_timeout(&test_iterations, &test_timeout);
     receiver
         .recv_timeout(Duration::from_secs(test_timeout))
         .map_err(|_| {
@@ -494,39 +639,41 @@ fn date() -> String {
     String::from_utf8_lossy(&output.stdout).trim().to_string()
 }
 
+#[derive(FromArgs)]
+/// Generate the performance metrics data for Cloud Hypervisor
+struct Options {
+    #[argh(switch, long = "list-tests")]
+    /// print the list of available metrics tests
+    list_tests: bool,
+
+    #[argh(option, long = "test-filter")]
+    /// filter metrics tests to run based on provided keywords
+    keywords: Vec<String>,
+
+    #[argh(option, long = "report-file")]
+    /// report file. Stderr is used if not specified
+    report_file: Option<String>,
+
+    #[argh(option, long = "iterations")]
+    /// override number of test iterations
+    iterations: Option<u32>,
+
+    #[argh(option, long = "timeout")]
+    /// override test timeout, Ex. --timeout 5
+    timeout: Option<u32>,
+
+    #[argh(switch, short = 'V', long = "version")]
+    /// print version information
+    version: bool,
+}
+
 fn main() {
-    let cmd_arguments = ClapCommand::new("performance-metrics")
-        .version(env!("GIT_HUMAN_READABLE"))
-        .author(crate_authors!())
-        .about("Generate the performance metrics data for Cloud Hypervisor")
-        .arg(
-            Arg::new("test-filter")
-                .long("test-filter")
-                .help("Filter metrics tests to run based on provided keywords")
-                .num_args(1)
-                .required(false),
-        )
-        .arg(
-            Arg::new("list-tests")
-                .long("list-tests")
-                .help("Print the list of availale metrics tests")
-                .num_args(0)
-                .action(ArgAction::SetTrue)
-                .required(false),
-        )
-        .arg(
-            Arg::new("report-file")
-                .long("report-file")
-                .help("Report file. Standard error is used if not specified")
-                .num_args(1),
-        )
-        .arg(
-            Arg::new("iterations")
-                .long("iterations")
-                .help("Override number of test iterations")
-                .num_args(1),
-        )
-        .get_matches();
+    let opts: Options = argh::from_env();
+
+    if opts.version {
+        println!("{} {}", env!("CARGO_BIN_NAME"), env!("BUILD_VERSION"));
+        return;
+    }
 
     // It seems that the tool (ethr) used for testing the virtio-net latency
     // is not stable on AArch64, and therefore the latency test is currently
@@ -536,7 +683,7 @@ fn main() {
         .filter(|t| !(cfg!(target_arch = "aarch64") && t.name == "virtio_net_latency_us"))
         .collect();
 
-    if cmd_arguments.get_flag("list-tests") {
+    if opts.list_tests {
         for test in test_list.iter() {
             println!("\"{}\" ({})", test.name, test.control);
         }
@@ -544,10 +691,7 @@ fn main() {
         return;
     }
 
-    let test_filter = match cmd_arguments.get_many::<String>("test-filter") {
-        Some(s) => s.collect(),
-        None => Vec::new(),
-    };
+    let test_filter = opts.keywords.iter().collect::<Vec<&String>>();
 
     // Run performance tests sequentially and report results (in both readable/json format)
     let mut metrics_report: MetricsReport = Default::default();
@@ -555,11 +699,8 @@ fn main() {
     init_tests();
 
     let overrides = Arc::new(PerformanceTestOverrides {
-        test_iterations: cmd_arguments
-            .get_one::<String>("iterations")
-            .map(|s| s.parse())
-            .transpose()
-            .unwrap_or_default(),
+        test_iterations: opts.iterations,
+        test_timeout: opts.timeout,
     });
 
     for test in test_list.iter() {
@@ -578,19 +719,18 @@ fn main() {
 
     cleanup_tests();
 
-    let mut report_file: Box<dyn std::io::Write + Send> =
-        if let Some(file) = cmd_arguments.get_one::<String>("report-file") {
-            Box::new(
-                std::fs::File::create(std::path::Path::new(file))
-                    .map_err(|e| {
-                        eprintln!("Error opening report file: {file}: {e}");
-                        std::process::exit(1);
-                    })
-                    .unwrap(),
-            )
-        } else {
-            Box::new(std::io::stdout())
-        };
+    let mut report_file: Box<dyn std::io::Write + Send> = if let Some(ref file) = opts.report_file {
+        Box::new(
+            std::fs::File::create(std::path::Path::new(file))
+                .map_err(|e| {
+                    eprintln!("Error opening report file: {file}: {e}");
+                    std::process::exit(1);
+                })
+                .unwrap(),
+        )
+    } else {
+        Box::new(std::io::stdout())
+    };
 
     report_file
         .write_all(