remove dead code

* Simple fast root snapshots

* clippy

* epd

* mips

.github/workflows/build_and_test.yml

@@ -20,12 +20,23 @@ jobs:
         with:
           profile: minimal
           toolchain: nightly
-      - uses: Swatinem/rust-cache@v1
+          override: true
       - name: install mdbook
-        run: cargo install mdbook
+        uses: baptiste0928/cargo-install@v1.3.0
+        with:
+          crate: mdbook
       - name: install linkcheck
-        run: cargo install mdbook-linkcheck
+        uses: baptiste0928/cargo-install@v1.3.0
-      - uses: actions/checkout@v2
+        with:
+          crate: mdbook-linkcheck
+      - uses: actions/checkout@v3
+      - uses: Swatinem/rust-cache@v2
+      - name: Install mimetype
+        if: runner.os == 'Linux'
+        run: sudo apt-get install libfile-mimeinfo-perl
+      - name: Check for binary blobs
+        if: runner.os == 'Linux'
+        run: ./scripts/check_for_blobs.sh
       - name: Build libafl debug
         run: cargo build -p libafl
       - name: Build the book
@@ -44,50 +55,59 @@ jobs:
   ubuntu:
     runs-on: ubuntu-22.04
     steps:
+    - name: Remove Dotnet & Haskell
+      run: rm -rf /usr/share/dotnet && rm -rf /opt/ghc
     - uses: actions-rs/toolchain@v1
       with:
         profile: minimal
         toolchain: stable
-    - uses: Swatinem/rust-cache@v1
     - name: set mold linker as default linker
       uses: rui314/setup-mold@v1
-    - name: Install deps
+    - name: Install and cache deps
-      run: sudo apt-get install -y llvm llvm-dev clang ninja-build clang-format-13 shellcheck
+      uses: awalsh128/cache-apt-pkgs-action@v1.1.0
+      with:
+        packages: llvm llvm-dev clang ninja-build clang-format-13 shellcheck libgtk-3-dev gcc-arm-linux-gnueabi g++-arm-linux-gnueabi libslirp-dev
     - name: get clang version
       run: command -v llvm-config && clang -v
     - name: Install cargo-hack
       run: curl -LsSf https://github.com/taiki-e/cargo-hack/releases/latest/download/cargo-hack-x86_64-unknown-linux-gnu.tar.gz | tar xzf - -C ~/.cargo/bin
     - name: Add nightly rustfmt and clippy
       run: rustup toolchain install nightly --component rustfmt --component clippy --allow-downgrade
-    - uses: actions/checkout@v2
+    - uses: actions/checkout@v3
-    - name: Run a normal build
+    - uses: Swatinem/rust-cache@v2
-      run: cargo build --verbose
+
-    # cargo-hack tests/checks each crate in the workspace
+    # ---- format check ----
-    #- name: Run tests
-    #  run: cargo hack test --all-features
-    # cargo-hack's --feature-powerset would be nice here but libafl has a too many knobs
-    - name: Check each feature
-      # Skipping python as it has to be built with the `maturin` tool
-      run: cargo hack check --feature-powerset --depth=2 --exclude-features=agpl,nautilus,python,sancov_pcguard_edges,arm,aarch64,i386 --no-dev-deps
     # pcguard edges and pcguard hitcounts are not compatible and we need to build them seperately
     - name: Check pcguard edges
       run: cargo check --features=sancov_pcguard_edges
-    - name: Build examples
-      run: cargo build --examples --verbose
-    - uses: actions/checkout@v2
     - name: Format
       run: cargo fmt -- --check
     - name: Run clang-format style check for C/C++ programs.
-      run: clang-format-13 -n -Werror --style=file $(find . -type f \( -name '*.cpp' -o -iname '*.hpp' -o -name '*.cc' -o -name '*.cxx' -o -name '*.cc' -o -name '*.h' \) | grep -v '/target/' | grep -v 'libpng-1\.6\.37' | grep -v 'stb_image\.h' | grep -v 'dlmalloc\.c')
+      run: clang-format-13 -n -Werror --style=file $(find . -type f \( -name '*.cpp' -o -iname '*.hpp' -o -name '*.cc' -o -name '*.cxx' -o -name '*.cc' -o -name '*.h' \) | grep -v '/target/' | grep -v 'libpng-1\.6\.37' | grep -v 'stb_image\.h' | grep -v 'dlmalloc\.c' | grep -v 'QEMU-Nyx')
     - name: run shellcheck
       run: shellcheck ./scripts/*.sh
-    - uses: actions/checkout@v2
+    - name: Run clippy
+      run: ./scripts/clippy.sh
+      
+    # ---- doc check ----
     - name: Build Docs
       run: cargo doc
     - name: Test Docs
       run: cargo +nightly test --doc --all-features
-    - name: Run clippy
+
-      run: ./scripts/clippy.sh
+    # ---- build and feature check ----
+    - name: Run a normal build
+      run: cargo build --verbose
+    # cargo-hack's --feature-powerset would be nice here but libafl has a too many knobs
+    - name: Check each feature
+      # Skipping `python` as it has to be built with the `maturin` tool
+      # `agpl`, `nautilus` require nightly
+      # `sancov_pcguard_edges` is tested seperately
+      run: cargo hack check --each-feature --clean-per-run --exclude-features=prelude,agpl,nautilus,python,sancov_pcguard_edges,arm,aarch64,i386,be,systemmode --no-dev-deps
+    - name: Check nightly features
+      run: cargo +nightly check --features=agpl && cargo +nightly check --features=nautilus
+    - name: Build examples
+      run: cargo build --examples --verbose
 
   ubuntu-concolic:
     runs-on: ubuntu-latest
@@ -96,8 +116,8 @@ jobs:
       with:
         profile: minimal
         toolchain: stable
-    - uses: Swatinem/rust-cache@v1
+    - uses: actions/checkout@v3
-    - uses: actions/checkout@v2
+    - uses: Swatinem/rust-cache@v2
     - name: Install smoke test deps
       run: sudo ./libafl_concolic/test/smoke_test_ubuntu_deps.sh 
     - name: Run smoke test
@@ -110,21 +130,18 @@ jobs:
       with:
         profile: minimal
         toolchain: stable
-    - uses: Swatinem/rust-cache@v1
     - name: set mold linker as default linker
       uses: rui314/setup-mold@v1
     - name: Install deps
       run: sudo apt-get install -y llvm llvm-dev clang ninja-build python3-dev python3-pip python3-venv
     - name: Install maturin
       run: python3 -m pip install maturin
-    - uses: actions/checkout@v2
+    - uses: actions/checkout@v3
+    - uses: Swatinem/rust-cache@v2
     - name: Run a maturin build
       run: cd ./bindings/pylibafl && maturin build
 
   fuzzers:
-    env:
-      CC: ccache clang # use ccache in default
-      CXX: ccache clang # use ccache in default
     strategy:
       matrix:
           os: [ubuntu-latest, macos-latest]
@@ -137,31 +154,38 @@ jobs:
     - name: set mold linker as default linker
       if: runner.os == 'Linux' # mold only support linux until now
       uses: rui314/setup-mold@v1
-    - name: add ccache 
-      uses: hendrikmuhs/ccache-action@v1.2
-      with:
-        key: ${{ github.job }}-${{ matrix.os }}
-        max-size: 2000M
-    - name: enable mult-thread for `make`
-      run: export MAKEFLAGS="-j$(expr $(nproc) \+ 1)"
-    - uses: Swatinem/rust-cache@v1
     - name: Add nightly rustfmt and clippy
       run: rustup toolchain install nightly --component rustfmt --component clippy --allow-downgrade
+    - name: Add no_std toolchain
+      run: rustup toolchain install nightly-x86_64-unknown-linux-gnu ; rustup component add rust-src --toolchain nightly-x86_64-unknown-linux-gnu
+    - name: Install python
+      if: runner.os == 'macOS'
+      run: brew install --force-bottle --overwrite python@3.11
     - uses: lyricwulf/abc@v1
       with: 
-        linux: llvm llvm-dev clang nasm ninja-build
+        # todo: remove afl++-clang when nyx support samcov_pcguard
+        linux: llvm llvm-dev clang nasm ninja-build gcc-arm-linux-gnueabi g++-arm-linux-gnueabi libgtk-3-dev afl++-clang pax-utils
         # update bash for macos to support `declare -A` command`
         macos: llvm libpng nasm coreutils z3 bash
+    - name: pip install
+      run: python3 -m pip install msgpack jinja2
+    # Note that nproc needs to have coreutils installed on macOS, so the order of CI commands matters.
+    - name: enable mult-thread for `make`
+      run: export MAKEFLAGS="-j$(expr $(nproc) \+ 1)"
     - name: install cargo-make
-      run: cargo install --force cargo-make
+      uses: baptiste0928/cargo-install@v1.3.0
-    - uses: actions/checkout@v2
+      with:
-    - name: Build and run example fuzzers
+        crate: cargo-make
+    - uses: actions/checkout@v3
+      with:
+        submodules: true # recursively checkout submodules
+    - uses: Swatinem/rust-cache@v2
+    - name: Build and run example fuzzers (Linux)
       if: runner.os == 'Linux'
       run: ./scripts/test_all_fuzzers.sh
-    - name: Build and run example fuzzers
+    - name: Build and run example fuzzers (macOS)
       if: runner.os == 'macOS' # use bash v4
       run: /usr/local/bin/bash ./scripts/test_all_fuzzers.sh
-    - run: ccache --show-stats
 
   nostd-build:
     runs-on: ubuntu-latest
@@ -170,10 +194,12 @@ jobs:
       with:
         profile: minimal
         toolchain: nightly
-    - uses: Swatinem/rust-cache@v1
+        override: true
-    - name: Add nightly rustfmt and clippy
+        components: rustfmt, clippy, rust-src
-      run: rustup toolchain install nightly && rustup target add --toolchain nightly aarch64-unknown-none && rustup component add --toolchain nightly rust-src && rustup target add thumbv6m-none-eabi
+    - uses: actions/checkout@v3
-    - uses: actions/checkout@v2
+    - uses: Swatinem/rust-cache@v2
+    - name: Add targets
+      run: rustup target add arm-linux-androideabi && rustup target add thumbv6m-none-eabi
     - name: Build aarch64-unknown-none
       run: cd ./fuzzers/baby_no_std && cargo +nightly build -Zbuild-std=core,alloc --target aarch64-unknown-none -v --release && cd ../..
     - name: run x86_64 until panic!
@@ -186,7 +212,7 @@ jobs:
   build-docker:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@v2
+    - uses: actions/checkout@v3
     - name: Build docker
       run: docker build -t libafl .
 
@@ -197,26 +223,25 @@ jobs:
       with:
         profile: minimal
         toolchain: stable
-    - uses: Swatinem/rust-cache@v1
+    - uses: actions/checkout@v3
-    - uses: actions/checkout@v2
+    - uses: Swatinem/rust-cache@v2
     - name: Windows Build
       run: cargo build --verbose
     - name: Run clippy
       uses: actions-rs/cargo@v1
       with:
         command: clippy
-    - name: Install LLVM and Clang # required for bindgen to work, see https://github.com/rust-lang/rust-bindgen/issues/1797
+    - name: Build docs
-      uses: KyleMayes/install-llvm-action@32c4866ebb71e0949e8833eb49beeebed48532bd
+      run: cargo doc
-      with:
-        version: "12.0"
-        directory: ${{ runner.temp }}/llvm
     - name: Set LIBCLANG_PATH
       run: echo "LIBCLANG_PATH=$((gcm clang).source -replace "clang.exe")" >> $env:GITHUB_ENV
     - name: install cargo-make
       run: cargo install --force cargo-make
     - uses: ilammy/msvc-dev-cmd@v1
-    - name: Build frida
+    - name: Build fuzzers/frida_libpng
       run: cd fuzzers/frida_libpng/ && cargo make test
+    - name: Build fuzzers/frida_gdiplus
+      run: cd fuzzers/frida_gdiplus/ && cargo make test
       
   macos:
     runs-on: macOS-latest
@@ -225,12 +250,12 @@ jobs:
       with:
         profile: minimal
         toolchain: stable
-    - uses: Swatinem/rust-cache@v1
     - name: Add nightly rustfmt and clippy
       run: rustup toolchain install nightly --component rustfmt --component clippy --allow-downgrade
     - name: Install deps
-      run: brew install z3
+      run: brew install z3 gtk+3
-    - uses: actions/checkout@v2
+    - uses: actions/checkout@v3
+    - uses: Swatinem/rust-cache@v2
     - name: MacOS Build
       run: cargo build --verbose
     - name: Run clippy
@@ -247,7 +272,6 @@ jobs:
       with:
         profile: minimal
         toolchain: stable
-    - uses: Swatinem/rust-cache@v1
     - uses: nttld/setup-ndk@v1
       with:
         ndk-version: r21e
@@ -257,7 +281,8 @@ jobs:
       run: rustup target add aarch64-linux-android
     - name: install cargo ndk
       run: cargo install cargo-ndk
-    - uses: actions/checkout@v2
+    - uses: actions/checkout@v3
+    - uses: Swatinem/rust-cache@v2
     - name: Build iOS
       run: cargo build --target aarch64-apple-ios
     - name: Build Android
@@ -271,3 +296,36 @@ jobs:
     #  run: clang -v
     #- name: Windows Test
     #  run: C:\Rust\.cargo\bin\cargo.exe test --verbose
+
+  freebsd:
+    runs-on: macos-12
+    name: Simple build in FreeBSD
+    steps:
+    - uses: actions/checkout@v3
+    - name: Test in FreeBSD
+      id: test
+      uses: vmactions/freebsd-vm@v0
+      with:
+        usesh: true
+        sync: rsync
+        copyback: false
+        mem: 2048
+        release: 13.1
+        prepare: |
+          pkg install -y curl bash sudo llvm14
+          curl https://sh.rustup.rs -sSf | sh -s -- -y
+
+        run: |
+          freebsd-version
+          . "$HOME/.cargo/env"
+          rustup toolchain install nightly
+          export LLVM_CONFIG=/usr/local/bin/llvm-config14
+          pwd
+          ls -lah
+          echo "local/bin"
+          ls -lah /usr/local/bin/
+          which llvm-config
+          chmod +x ./scripts/clippy.sh
+          bash ./scripts/shmem_limits_fbsd.sh
+          bash ./scripts/clippy.sh
+          cargo test

.github/workflows/codecov.yml

@@ -1,35 +0,0 @@
-on: [push]
-
-name: codecov
-
-jobs:
-  check:
-    name: code coverage test
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout repository
-        uses: actions/checkout@v2
-
-      - uses: Swatinem/rust-cache@v1
-      
-      - name: Install stable toolchain
-        uses: actions-rs/toolchain@v1
-        with:
-          toolchain: stable
-          override: true
-
-      - name: Run cargo-tarpaulin
-        uses: actions-rs/tarpaulin@v0.1
-        with:
-          version: '0.15.0'
-
-      - name: Upload to codecov.io
-        uses: codecov/codecov-action@v1.0.2
-        with:
-          token: ${{secrets.CODECOV_TOKEN}}
-
-      - name: Archive code coverage results
-        uses: actions/upload-artifact@v1
-        with:
-          name: code-coverage-report
-          path: cobertura.xml

.gitignore

@@ -1,4 +1,5 @@
 target
+target-bin
 out
 Cargo.lock
 vendor
@@ -7,6 +8,7 @@ vendor
 .env
 
 *.tmp
+*.swp
 *.o
 *.a
 *.so
@@ -16,6 +18,7 @@ vendor
 *.dll
 *.exe
 *.dSYM
+*.obj
 
 .cur_input
 .venv
@@ -45,3 +48,9 @@ __pycache__
 *.lafl_lock
 
 *atomic_file_testfile*
+**/libxml2
+**/corpus_discovered
+**/libxml2-*.tar.gz
+
+libafl_nyx/QEMU-Nyx
+libafl_nyx/packer

Cargo.toml

@@ -1,9 +1,3 @@
-[profile.release]
-lto = true
-codegen-units = 1
-opt-level = 3
-debug = true
-
 [workspace]
 members = [
     "libafl",
@@ -12,7 +6,9 @@ members = [
     "libafl_targets",
     "libafl_frida",
     "libafl_qemu",
+    "libafl_tinyinst",
     "libafl_sugar",
+    "libafl_nyx",
     "libafl_concolic/symcc_runtime",
     "libafl_concolic/symcc_libafl",
     "libafl_concolic/test/dump_constraints",
@@ -31,4 +27,16 @@ exclude = [
     "fuzzers",
     "bindings",
     "scripts",
+    "libafl_qemu/libafl_qemu_build",
+    "libafl_qemu/libafl_qemu_sys"
 ]
+
+[workspace.package]
+version = "0.8.2"
+
+[profile.release]
+lto = true
+codegen-units = 1
+opt-level = 3
+debug = true
+

Dockerfile

@@ -39,6 +39,12 @@ COPY libafl_frida/src/gettls.c libafl_frida/src/gettls.c
 COPY libafl_qemu/Cargo.toml libafl_qemu/build.rs libafl_qemu/
 COPY scripts/dummy.rs libafl_qemu/src/lib.rs
 
+COPY libafl_qemu/libafl_qemu_build/Cargo.toml libafl_qemu/libafl_qemu_build/
+COPY scripts/dummy.rs libafl_qemu/libafl_qemu_build/src/lib.rs
+
+COPY libafl_qemu/libafl_qemu_sys/Cargo.toml libafl_qemu/libafl_qemu_sys/build.rs libafl_qemu/libafl_qemu_sys/
+COPY scripts/dummy.rs libafl_qemu/libafl_qemu_sys/src/lib.rs
+
 COPY libafl_sugar/Cargo.toml libafl_sugar/
 COPY scripts/dummy.rs libafl_sugar/src/lib.rs
 
@@ -63,6 +69,12 @@ COPY scripts/dummy.rs libafl_concolic/symcc_runtime/src/lib.rs
 COPY libafl_concolic/symcc_libafl/Cargo.toml libafl_concolic/symcc_libafl/
 COPY scripts/dummy.rs libafl_concolic/symcc_libafl/src/lib.rs
 
+COPY libafl_nyx/Cargo.toml libafl_nyx/build.rs libafl_nyx/
+COPY scripts/dummy.rs libafl_nyx/src/lib.rs
+
+COPY libafl_tinyinst/Cargo.toml libafl_tinyinst/
+COPY scripts/dummy.rs libafl_tinyinst/src/lib.rs
+
 COPY utils utils
 
 RUN cargo build && cargo build --release
@@ -88,12 +100,18 @@ RUN touch libafl/src/lib.rs
 COPY libafl_targets/src libafl_targets/src
 RUN touch libafl_targets/src/lib.rs
 COPY libafl_frida/src libafl_frida/src
+RUN touch libafl_qemu/libafl_qemu_build/src/lib.rs
+COPY libafl_qemu/libafl_qemu_build/src libafl_qemu/libafl_qemu_build/src
+RUN touch libafl_qemu/libafl_qemu_sys/src/lib.rs
+COPY libafl_qemu/libafl_qemu_sys/src libafl_qemu/libafl_qemu_sys/src
 RUN touch libafl_qemu/src/lib.rs
 COPY libafl_qemu/src libafl_qemu/src
 RUN touch libafl_frida/src/lib.rs
 COPY libafl_concolic/symcc_libafl libafl_concolic/symcc_libafl
 COPY libafl_concolic/symcc_runtime libafl_concolic/symcc_runtime
 COPY libafl_concolic/test libafl_concolic/test
+COPY libafl_nyx/src libafl_nyx/src
+RUN touch libafl_nyx/src/lib.rs
 RUN cargo build && cargo build --release
 
 # Copy fuzzers over

README.md

@@ -34,6 +34,7 @@ LibAFL offers integrations with popular instrumentation frameworks. At the momen
 + SanitizerCoverage, in [libafl_targets](./libafl_targets)
 + Frida, in [libafl_frida](./libafl_frida)
 + QEMU user-mode, in [libafl_qemu](./libafl_qemu)
++ TinyInst, in [libafl_tinyinst](./libafl_tinyinst) by [elbiazo](https://github.com/elbiazo)
 
 ## Getting started
 
@@ -79,6 +80,12 @@ cd docs && mdbook serve
 We collect all example fuzzers in [`./fuzzers`](./fuzzers/).
 Be sure to read their documentation (and source), this is *the natural way to get started!*
 
+You can run each example fuzzer with
+```
+cargo make run
+```
+as long as the fuzzer directory has `Makefile.toml` file.
+
 The best-tested fuzzer is [`./fuzzers/libfuzzer_libpng`](./fuzzers/libfuzzer_libpng), a multicore libfuzzer-like fuzzer using LibAFL for a libpng harness.
 
 ## Resources
@@ -89,12 +96,16 @@ The best-tested fuzzer is [`./fuzzers/libfuzzer_libpng`](./fuzzers/libfuzzer_lib
 
 + The LibAFL book (WIP) [online](https://aflplus.plus/libafl-book) or in the [repo](./docs/src/)
 
++ Our research [paper](https://www.s3.eurecom.fr/docs/ccs22_fioraldi.pdf)
+
 + Our RC3 [talk](http://www.youtube.com/watch?v=3RWkT1Q5IV0 "Fuzzers Like LEGO") explaining the core concepts
 
 + Our Fuzzcon Europe [talk](https://www.youtube.com/watch?v=PWB8GIhFAaI "LibAFL: The Advanced Fuzzing Library") with a (a bit but not so much outdated) step-by-step discussion on how to build some example fuzzers
 
 + The Fuzzing101 [solutions](https://github.com/epi052/fuzzing-101-solutions) & series of [blog posts](https://epi052.gitlab.io/notes-to-self/blog/2021-11-01-fuzzing-101-with-libafl/) by [epi](https://github.com/epi052)
 
++ Blogpost on binary-only fuzzing lib libaf_qemu, [Hacking TMNF - Fuzzing the game server](https://blog.bricked.tech/posts/tmnf/part1/), by [RickdeJager](https://github.com/RickdeJager).
+
 ## Contributing
 
 For bugs, feel free to open issues or contact us directly. Thank you for your support. <3
@@ -107,6 +118,23 @@ Even though we will gladly assist you in finishing up your PR, try to
 
 Some of the parts in this list may be hard, don't be afraid to open a PR if you cannot fix them by yourself, so we can help.
 
+## Cite
+
+If you use LibAFL for your academic work, please cite the following paper:
+
+```bibtex
+@inproceedings{libafl,
+ author       = {Andrea Fioraldi and Dominik Maier and Dongjia Zhang and Davide Balzarotti},
+ title        = {{LibAFL: A Framework to Build Modular and Reusable Fuzzers}},
+ booktitle    = {Proceedings of the 29th ACM conference on Computer and communications security (CCS)},
+ series       = {CCS '22},
+ year         = {2022},
+ month        = {November},
+ location     = {Los Angeles, U.S.A.},
+ publisher    = {ACM},
+}
+```
+
 #### License
 
 <sup>

bindings/pylibafl/Cargo.toml

@@ -1,16 +1,16 @@
 [package]
 name = "pylibafl"
-version = "0.8.0"
+version = "0.8.2"
 edition = "2021"
 
 [dependencies]
-pyo3 = { version = "0.15", features = ["extension-module"] }
+pyo3 = { version = "0.17", features = ["extension-module"] }
-libafl_qemu = { path = "../../libafl_qemu", version = "0.8.0", features = ["python"] }
+libafl_qemu = { path = "../../libafl_qemu", version = "0.8.2", features = ["python"] }
-libafl_sugar = { path = "../../libafl_sugar", version = "0.8.0", features = ["python"] }
+libafl_sugar = { path = "../../libafl_sugar", version = "0.8.2", features = ["python"] }
-libafl = { path = "../../libafl", version = "0.8.0", features = ["python"] }
+libafl = { path = "../../libafl", version = "0.8.2", features = ["python"] }
 
 [build-dependencies]
-pyo3-build-config = { version = "0.15" }
+pyo3-build-config = { version = "0.17" }
 
 [lib]
 name = "pylibafl"

bindings/pylibafl/README.md

@@ -1,10 +1,10 @@
 # How to use python bindings
 ## First time setup
-```
+```bash
-# Create environment variable
-python -m venv .env 
 # Install maturin
 pip install maturin
+# Create virtual environment
+python3 -m venv .env
 ```
 ## Build bindings
 ```
@@ -13,7 +13,7 @@ source .env/bin/activate
 # Build python module
 maturin develop
 ```
-This is going to install `pylibafl` python module. 
+This is going to install `pylibafl` python module into this venv. 
 
 ## Use bindings
 ### Example: Running baby_fuzzer in fuzzers/baby_fuzzer/baby_fuzzer.py
@@ -28,4 +28,4 @@ Then simply run
 ```
 python PATH_TO_BABY_FUZZER/baby_fuzzer.py
 ```
-The crashes' directory will be created in the directory from which you ran the command.
+The crashes directory will be created in the directory from which you ran the command.

bindings/pylibafl/src/lib.rs

@@ -2,8 +2,7 @@ use libafl;
 #[cfg(target_os = "linux")]
 use libafl_qemu;
 use libafl_sugar;
-use pyo3::prelude::*;
+use pyo3::{prelude::*, types::PyDict};
-use pyo3::types::PyDict;
 
 const LIBAFL_CODE: &str = r#"
 class BaseObserver:

bindings/pylibafl/test.py

@@ -1,30 +1,38 @@
 from pylibafl.libafl import *
 import ctypes
 
+
 class FooObserver(BaseObserver):
     def __init__(self):
         self.n = 0
+
     def name(self):
         return "Foo"
+
     def pre_exec(self, state, input):
         if self.n % 10000 == 0:
             print("FOO!", self.n, input)
         self.n += 1
 
+
 class FooFeedback(BaseFeedback):
     def is_interesting(self, state, mgr, input, observers, exit_kind):
         ob = observers.match_name("Foo").unwrap_py()
         return ob.n % 10000 == 0
 
+
 class FooExecutor(BaseExecutor):
     def __init__(self, harness, observers: ObserversTuple):
         self.h = harness
         self.o = observers
+
     def observers(self):
         return self.o
+
     def run_target(self, fuzzer, state, mgr, input) -> ExitKind:
         return (self.h)(input)
 
+
 libc = ctypes.cdll.LoadLibrary("libc.so.6")
 
 area_ptr = libc.calloc(1, 4096)
@@ -33,34 +41,46 @@ observer = StdMapObserverI8("mymap", area_ptr, 4096)
 
 m = observer.as_map_observer()
 
-observers = ObserversTuple([observer.as_map_observer().as_observer(), FooObserver().as_observer()])
+observers = ObserversTuple(
+    [observer.as_map_observer().as_observer(), FooObserver().as_observer()]
+)
 
 feedback = feedback_or(MaxMapFeedbackI8(m).as_feedback(), FooFeedback().as_feedback())
 
-objective = feedback_and_fast(CrashFeedback().as_feedback(), MaxMapFeedbackI8(m).as_feedback())
+objective = feedback_and_fast(
+    CrashFeedback().as_feedback(), MaxMapFeedbackI8(m).as_feedback()
+)
 
 fuzzer = StdFuzzer(feedback, objective)
 
 rand = StdRand.with_current_nanos()
 
-state = StdState(rand.as_rand(), InMemoryCorpus().as_corpus(), InMemoryCorpus().as_corpus(), feedback, objective)
+state = StdState(
+    rand.as_rand(),
+    InMemoryCorpus().as_corpus(),
+    InMemoryCorpus().as_corpus(),
+    feedback,
+    objective,
+)
 
 monitor = SimpleMonitor(lambda s: print(s))
 
 mgr = SimpleEventManager(monitor.as_monitor())
 
+
 def harness(buf) -> ExitKind:
-    #print(buf)
+    # print(buf)
     m[0] = 1
-    if len(buf) > 0 and buf[0] == ord('a'):
+    if len(buf) > 0 and buf[0] == ord("a"):
         m[1] = 1
-        if len(buf) > 1 and buf[1] == ord('b'):
+        if len(buf) > 1 and buf[1] == ord("b"):
             m[2] = 1
-            if len(buf) > 2 and buf[2] == ord('c'):
+            if len(buf) > 2 and buf[2] == ord("c"):
                 m[3] = 1
                 return ExitKind.crash()
     return ExitKind.ok()
 
+
 # executor = InProcessExecutor(harness, observers, fuzzer, state, mgr.as_manager())
 
 executor = FooExecutor(harness, observers)
@@ -69,6 +89,6 @@ stage = StdMutationalStage(StdHavocMutator().as_mutator())
 
 stage_tuple_list = StagesTuple([stage.as_stage()])
 
-fuzzer.add_input(state, executor.as_executor(), mgr.as_manager(), b'\0\0')
+fuzzer.add_input(state, executor.as_executor(), mgr.as_manager(), b"\0\0")
 
 fuzzer.fuzz_loop(executor.as_executor(), state, mgr.as_manager(), stage_tuple_list)

docs/src/SUMMARY.md

@@ -24,6 +24,7 @@
 - [Design](./design/design.md)
   - [Architecture](./design/architecture.md)
   - [Metadata](./design/metadata.md)
+  - [Migrating from LibAFL <0.9 to 0.9](./design/migration-0.9.md)
 
 - [Message Passing](./message_passing/message_passing.md)
   - [Spawning Instances](./message_passing/spawn_instances.md)
@@ -33,5 +34,7 @@
   - [Introduction](./tutorial/intro.md)
 
 - [Advanced Features](./advanced_features/advanced_features.md)
-  - [Concolic Tracing & Hybrid Fuzzing](./advanced_features/concolic/concolic.md)
+  - [Binary-Only Fuzzing with `Frida`](./advanced_features/frida.md)
-  - [LibAFL in `no_std` environments (Kernels, Hypervisors, ...)](./advanced_features/no_std/no_std.md)
+  - [Concolic Tracing & Hybrid Fuzzing](./advanced_features/concolic.md)
+  - [LibAFL in `no_std` environments (Kernels, Hypervisors, ...)](./advanced_features/no_std.md)
+  - [Snapshot Fuzzing in Nyx](./advanced_features/nyx.md)

docs/src/advanced_features/advanced_features.md

@@ -1,3 +1,4 @@
 # Advanced Features
+
 In addition to core building blocks for fuzzers, LibAFL also has features for more advanced/niche fuzzing techniques.
-The following sections are dedicated to these features.
+The following sections are dedicated to some of these features.

docs/src/advanced_features/concolic.md

@@ -6,7 +6,9 @@ Then, we'll go through the relationship of SymCC and LibAFL concolic tracing.
 Finally, we'll walk through building a basic hybrid fuzzer using LibAFL.
 
 ## Concolic Tracing by Example
+
 Suppose you want to fuzz the following program:
+
 ```rust
 fn target(input: &[u8]) -> i32 {
     match &input {
@@ -19,6 +21,7 @@ fn target(input: &[u8]) -> i32 {
     }
 }
 ```
+
 A simple coverage-maximizing fuzzer that generates new inputs somewhat randomly will have a hard time finding an input that triggers the fictitious crashing input.
 Many techniques have been proposed to make fuzzing less random and more directly attempt to mutate the input to flip specific branches, such as the ones involved in crashing the above program.
 
@@ -27,6 +30,7 @@ In principle, concolic tracing works by observing all executed instructions in a
 To understand what this entails, we'll run an example with the above program.
 
 First, we'll simplify the program to simple if-then-else-statements:
+
 ```rust
 fn target(input: &[u8]) -> i32 {
     if input.len() == 4 {
@@ -56,8 +60,10 @@ fn target(input: &[u8]) -> i32 {
     }
 }
 ```
+
 Next, we'll trace the program on the input `[]`.
 The trace would look like this:
+
 ```rust,ignore
 Branch { // if input.len() == 4
     condition: Equals { 
@@ -74,6 +80,7 @@ Branch { // if input.len() == 0
     taken: true // This condition turned out to be true!
 }
 ```
+
 Using this trace, we can easily deduce that we can force the program to take a different path by having an input of length 4 or having an input with non-zero length.
 We do this by negating each branch condition and analytically solving the resulting 'expression'.
 In fact, we can create these expressions for any computation and give them to an [SMT](https://en.wikipedia.org/wiki/Satisfiability_modulo_theories)-Solver that will generate an input that satisfies the expression (as long as such an input exists).
@@ -81,13 +88,16 @@ In fact, we can create these expressions for any computation and give them to an
 In hybrid fuzzing, we combine this tracing + solving approach with more traditional fuzzing techniques.
 
 ## Concolic Tracing in LibAFL, SymCC and SymQEMU
+
 The concolic tracing support in LibAFL is implemented using SymCC.
 SymCC is a compiler plugin for clang that can be used as a drop-in replacement for a normal C or C++ compiler.
 SymCC will instrument the compiled code with callbacks into a runtime that can be supplied by the user.
 These callbacks allow the runtime to construct a trace that similar to the previous example.
 
 ### SymCC and its Runtimes
+
 SymCC ships with 2 runtimes: 
+
  * a 'simple' runtime that attempts to solve any branches it comes across using [Z3](https://github.com/Z3Prover/z3/wiki) and
  * a [QSym](https://github.com/sslab-gatech/qsym)-based runtime, which does a bit more filtering on the expressions and also solves using Z3.
 
@@ -96,15 +106,18 @@ This crate allows you to easily build a custom runtime out of the built-in build
 Checkout out the `symcc_runtime` docs for more information on how to build your own runtime.
 
 ### SymQEMU
+
 [SymQEMU](https://github.com/eurecom-s3/symqemu) is a sibling project to SymCC.
 Instead of instrumenting the target at compile-time, it inserts instrumentation via dynamic binary translation, building on top of the [`QEMU`](https://www.qemu.org) emulation stack.
 This means that using SymQEMU, any (x86) binary can be traced without the need to build in instrumentation ahead of time.
 The `symcc_runtime` crate supports this use case and runtimes built with `symcc_runtime` also work with SymQEMU.
 
 ## Hybrid Fuzzing in LibAFL
+
 The LibAFL repository contains an [example hybrid fuzzer](https://github.com/AFLplusplus/LibAFL/tree/main/fuzzers/libfuzzer_stb_image_concolic).
 
 There are three main steps involved with building a hybrid fuzzer using LibAFL:
+
 1. Building a runtime,
 2. choosing an instrumentation method and
 3. building the fuzzer.
@@ -113,11 +126,13 @@ Note that the order of these steps is important.
 For example, we need to have runtime ready before we can do instrumentation with SymCC.
 
 ### Building a Runtime
+
 Building a custom runtime can be done easily using the `symcc_runtime` crate.
 Note, that a custom runtime is a separate shared object file, which means that we need a separate crate for our runtime.
 Check out the [example hybrid fuzzer's runtime](https://github.com/AFLplusplus/LibAFL/tree/main/fuzzers/libfuzzer_stb_image_concolic/runtime) and the [`symcc_runtime` docs](https://docs.rs/symcc_runtime/0.1/symcc_runtime) for inspiration.
 
 ### Instrumentation
+
 There are two main instrumentation methods to make use of concolic tracing in LibAFL:
 * Using an **compile-time** instrumented target with **SymCC**. 
 This only works when the source is available for the target and the target is reasonably easy to build using the SymCC compiler wrapper.
@@ -127,6 +142,7 @@ It should be noted, however, that the 'quality' of the generated expressions can
 Therefore, it is recommended to use SymCC over SymQEMU when possible.
 
 #### Using SymCC
+
 The target needs to be instrumented ahead of fuzzing using SymCC.
 How exactly this is done does not matter.
 However, the SymCC compiler needs to be made aware of the location of the runtime that it should instrument against.
@@ -139,11 +155,13 @@ The [`symcc_libafl` crate](https://docs.rs/symcc_libafl) contains helper functio
 Make sure you satisfy the [build requirements](https://github.com/eurecom-s3/symcc#readme) of SymCC before attempting to build it.
 
 #### Using SymQEMU
+
 Build SymQEMU according to its [build instructions](https://github.com/eurecom-s3/symqemu#readme).
 By default, SymQEMU looks for the runtime in a sibling directory.
 Since we don't have a runtime there, we need to let it know the path to your runtime by setting `--symcc-build` argument of the `configure` script to the path of your runtime.
 
 ### Building the Fuzzer
+
 No matter the instrumentation method, the interface between the fuzzer and the instrumented target should now be consistent.
 The only difference between using SymCC and SymQEMU should be the binary that represents the target:
 In the case of SymCC it will be the binary that was build with instrumentation and with SymQEMU it will be the emulator binary (eg. `x86_64-linux-user/symqemu-x86_64`), followed by your uninstrumented target binary and arguments.
@@ -152,6 +170,7 @@ You can use the [`CommandExecutor`](https://docs.rs/libafl/0.6.0/libafl/executor
 When configuring the command, make sure you pass the `SYMCC_INPUT_FILE` environment variable the input file path, if your target reads input from a file (instead of standard input).
 
 #### Serialization and Solving
+
 While it is perfectly possible to build a custom runtime that also performs the solving step of hybrid fuzzing in the context of the target process, the intended use of the LibAFL concolic tracing support is to serialize the (filtered and pre-processed) branch conditions using the [`TracingRuntime`](https://docs.rs/symcc_runtime/0.1/symcc_runtime/tracing/struct.TracingRuntime.html).
 This serialized representation can be deserialized in the fuzzer process for solving using a [`ConcolicObserver`](https://docs.rs/libafl/0.6.0/libafl/observers/concolic/struct.ConcolicObserver.html) wrapped in a [`ConcolicTracingStage`](https://docs.rs/libafl/0.6.0/libafl/stages/concolic/struct.ConcolicTracingStage.html), which will attach a [`ConcolicMetadata`](https://docs.rs/libafl/0.6.0/libafl/observers/concolic/struct.ConcolicMetadata.html) to every [`TestCase`](https://docs.rs/libafl/0.6.0/libafl/corpus/testcase/struct.Testcase.html).
 
@@ -161,5 +180,5 @@ The [`SimpleConcolicMutationalStage`](https://docs.rs/libafl/0.6.0//libafl/stage
 It will attempt to solve all branches, like the original simple backend from SymCC, using Z3.
 
 ### Example
-The example fuzzer shows how to use the [`ConcolicTracingStage` together with the `SimpleConcolicMutationalStage`](https://github.com/AFLplusplus/LibAFL/blob/main/fuzzers/libfuzzer_stb_image_concolic/fuzzer/src/main.rs#L203) to build a basic hybrid fuzzer.
 
+The example fuzzer shows how to use the [`ConcolicTracingStage` together with the `SimpleConcolicMutationalStage`](https://github.com/AFLplusplus/LibAFL/blob/main/fuzzers/libfuzzer_stb_image_concolic/fuzzer/src/main.rs#L222) to build a basic hybrid fuzzer.

docs/src/advanced_features/frida.md

@@ -1,36 +1,43 @@
 # Binary-only Fuzzing with Frida
-LibAFL supports binary-only fuzzing with Frida; the dynamic instrumentation tool.
 
-In this section, we'll talk about some of the components in fuzzing with `libafl_frida`.
+LibAFL supports different instrumentation engines for binary-only fuzzing.
-You can take a look at a working example in our `fuzzers/frida_libpng` folder.
+A potent cross-platform (Windows, MacOS, Android, Linux, iOS) option for binary-only fuzzing is Frida; the dynamic instrumentation tool.
+
+In this section, we will talk about the components in fuzzing with `libafl_frida`.
+You can take a look at a working example in our [`fuzzers/frida_libpng`](https://github.com/AFLplusplus/LibAFL/tree/main/fuzzers/frida_libpng) folder for Linux, and [`fuzzers/frida_gdiplus`](https://github.com/AFLplusplus/LibAFL/tree/main/fuzzers/frida_gdiplus) for Windows.
+
+## Dependencies
 
-# Dependencies
 If you are on Linux or OSX, you'll need [libc++](https://libcxx.llvm.org/) for `libafl_frida` in addition to libafl's dependencies.
 If you are on Windows, you'll need to install llvm tools.
 
 
-# Harness & Instrumentation
+## Harness & Instrumentation
+
 LibAFL uses Frida's [__Stalker__](https://frida.re/docs/stalker/) to trace the execution of your program and instrument your harness.
-Thus you have to compile your harness to a dynamic library. Frida instruments your PUT after dynamically loading it.
+Thus, you have to compile your harness to a dynamic library. Frida instruments your PUT after dynamically loading it.
 
 For example in our `frida_libpng` example, we load the dynamic library and find the symbol to harness as follows:
-```rust
+
+```rust,ignore
         let lib = libloading::Library::new(module_name).unwrap();
         let target_func: libloading::Symbol<
             unsafe extern "C" fn(data: *const u8, size: usize) -> i32,
         > = lib.get(symbol_name.as_bytes()).unwrap();
 ```
 
+## `FridaInstrumentationHelper` and Runtimes
 
-# `FridaInstrumentationHelper` and Runtimes
 To use functionalities that Frida offers, we'll first need to obtain `Gum` object by `Gum::obtain()`.
 
-In LibAFL, We use struct `FridaInstrumentationHelper` to manage all the stuff related to Frida. `FridaInstrumentationHelper` is a key component that sets up the [__Transformer__](https://frida.re/docs/stalker/#transformer) that is used to to generate the instrumented code. It also initializes the `Runtimes` that offers various instrumentation.
+In LibAFL, we use the `FridaInstrumentationHelper` struct to manage frida-related state. `FridaInstrumentationHelper` is a key component that sets up the [__Transformer__](https://frida.re/docs/stalker/#transformer) that is used to generate the instrumented code. It also initializes the `Runtimes` that offer various instrumentation.
 
-We have `CoverageRuntime` that has tracks the edge coverage,  `AsanRuntime` for address sanitizer, `DrCovRuntime` that uses [__DrCov__](https://dynamorio.org/page_drcov.html) for coverage collection, and `CmpLogRuntime` for cmplog instrumentation. All these runtimes can be used by slotting them into `FridaInstrumentationHelper`
+We have `CoverageRuntime` that can track the edge coverage,  `AsanRuntime` for address sanitizer, `DrCovRuntime` that uses [__DrCov__](https://dynamorio.org/page_drcov.html) for coverage collection (to be imported in coverage tools like Lighthouse, bncov, dragondance,...), and `CmpLogRuntime` for cmplog instrumentation.
+All of these runtimes can be slotted into `FridaInstrumentationHelper` at build time.
 
-Combined with any `Runtime` you'd like to use, you can initialize the `FridaInstrumentationHelpe`r like this:
+Combined with any `Runtime` you'd like to use, you can initialize the `FridaInstrumentationHelper` like this:
-```rust
+
+```rust,ignore
 
         let gum = Gum::obtain();
         let frida_options = FridaOptions::parse_env_options();
@@ -44,17 +51,21 @@ Combined with any `Runtime` you'd like to use, you can initialize the `FridaInst
         );
 ```
 
-# Run the fuzzer
+## Running the Fuzzer
+
 After setting up the `FridaInstrumentationHelper`. You can obtain the pointer to the coverage map by calling `map_ptr_mut()`.
-```rust
+
+```rust,ignore
         let edges_observer = HitcountsMapObserver::new(StdMapObserver::new_from_ptr(
             "edges",
             frida_helper.map_ptr_mut().unwrap(),
             MAP_SIZE,
         ));
 ```
-You can link this observer to `FridaInProcessExecutor`,
+
-```rust
+You can then link this observer to `FridaInProcessExecutor` as follows:
+
+```rust,ignore
         let mut executor = FridaInProcessExecutor::new(
             &gum,
             InProcessExecutor::new(
@@ -71,4 +82,6 @@ You can link this observer to `FridaInProcessExecutor`,
             &mut frida_helper,
         );
 ```
-and finally you can run the fuzzer.
+
+And, finally you can run the fuzzer.
+See the `frida_` examples in [`./fuzzers`](https://github.com/AFLplusplus/LibAFL/tree/main/fuzzers/) for more information and, for linux or full-system, play around with `libafl_qemu`, another binary-only tracer.

docs/src/advanced_features/no_std.md

@@ -1,6 +1,6 @@
 # Using LibAFL in `no_std` environments
 
-It is possible to use LibAFL in `no_std` environments e.g. custom platforms like microcontrolles, kernels, hypervisors, and more.
+It is possible to use LibAFL in `no_std` environments e.g. custom platforms like microcontrollers, kernels, hypervisors, and more.
 
 You can simply add LibAFL to your `Cargo.toml` file:
 
@@ -8,7 +8,8 @@ You can simply add LibAFL to your `Cargo.toml` file:
 libafl = { path = "path/to/libafl/", default-features = false}
 ```
 
-Then build your project e.g. for `aarch64-unknown-none` using
+Then build your project e.g. for `aarch64-unknown-none` using:
+
 ```sh
 cargo build --no-default-features --target aarch64-unknown-none
 ```
@@ -18,18 +19,22 @@ cargo build --no-default-features --target aarch64-unknown-none
 The minimum amount of input LibAFL needs for `no_std` is a monotonically increasing timestamp.
 For this, anywhere in your project you need to implement the `external_current_millis` function, which returns the current time in milliseconds.
 
-// Assume this a clock source from a custom stdlib, which you want to use, which returns current time in seconds.
 ```c
+// Assume this a clock source from a custom stdlib, which you want to use, which returns current time in seconds.
 int my_real_seconds(void)
 {
     return *CLOCK;
 }
 ```
-and here we use it in Rust. `external_current_millis` is then called from LibAFL.
+
-Note that it needs to be `no_mangle` in order to get picked up by LibAFL at linktime.
+Here, we use it in Rust. `external_current_millis` is then called from LibAFL.
+Note that it needs to be `no_mangle` in order to get picked up by LibAFL at linktime:
+
 ```rust,ignore
 #[no_mangle]
 pub extern "C" fn external_current_millis() -> u64 {
     unsafe { my_real_seconds()*1000 }
 }
 ```
+
+See [./fuzzers/baby_no_std](https://github.com/AFLplusplus/LibAFL/tree/main/fuzzers/baby_no_std) for an example.

docs/src/advanced_features/nyx.md

@@ -0,0 +1,126 @@
+# Snapshot Fuzzing in Nyx
+
+NYX supports both source-based and binary-only fuzzing.
+
+Currently, `libafl_nyx` only supports [afl++](https://github.com/AFLplusplus/AFLplusplus)'s instruction. To install it, you can use `sudo apt install aflplusplus`. Or compile from the source:
+
+```bash
+git clone https://github.com/AFLplusplus/AFLplusplus
+cd AFLplusplus
+make all # this will not compile afl's additional extension
+```
+
+Then you should compile the target with the afl++ compiler wrapper:
+
+```bash
+export CC=afl-clang-fast
+export CXX=afl-clang-fast++
+# the following line depends on your target
+./configure --enable-shared=no
+make
+```
+
+For binary-only fuzzing, Nyx uses intel-PT(Intel® Processor Trace). You can find the supported CPU at <https://www.intel.com/content/www/us/en/support/articles/000056730/processors.html>.
+
+## Preparing Nyx working directory
+
+This step is used to pack the target into Nyx's kernel. Don't worry, we have a template shell script in our [example](https://github.com/AFLplusplus/LibAFL/blob/main/fuzzers/nyx_libxml2_parallel/setup_libxml2.sh):
+
+the parameter's meaning is listed below:
+
+```bash
+git clone https://github.com/nyx-fuzz/packer
+python3 "./packer/packer/nyx_packer.py" \
+    ./libxml2/xmllint \   # your target binary
+    /tmp/nyx_libxml2 \    # the nyx work directory
+    afl \                 # instruction type
+    instrumentation \
+    -args "/tmp/input" \  # the args of the program, means that we will run `xmllint /tmp/input` in each run.
+    -file "/tmp/input" \  # the input will be generated in `/tmp/input`. If no `--file`, then input will be passed through stdin
+    --fast_reload_mode \
+    --purge || exit
+```
+
+Then, you can generate the config file:
+
+```bash
+python3 ./packer/packer/nyx_config_gen.py /tmp/nyx_libxml2/ Kernel || exit
+```
+
+## Standalone fuzzing
+
+In the [example fuzzer](https://github.com/AFLplusplus/LibAFL/blob/main/fuzzers/nyx_libxml2_standalone/src/main.rs). First you need to run `./setup_libxml2.sh`, It will prepare your target and create your nyx work directory in `/tmp/libxml2`. After that, you can start write your code.
+
+First, to create `Nyxhelper`:
+
+```rust,ignore
+let share_dir = Path::new("/tmp/nyx_libxml2/");
+let cpu_id = 0; // use first cpu
+let parallel_mode = false; // close parallel_mode
+let mut helper = NyxHelper::new(share_dir, cpu_id, true, parallel_mode, None).unwrap(); // we don't the set the last parameter in standalone mode, we just use None, here
+```
+
+Then, fetch `trace_bits`, create an observer and the `NyxExecutor`:
+
+```rust,ignore
+let trace_bits = unsafe { std::slice::from_raw_parts_mut(helper.trace_bits, helper.map_size) };
+let observer = StdMapObserver::new("trace", trace_bits);
+let mut executor = NyxExecutor::new(&mut helper, tuple_list!(observer)).unwrap();
+```
+
+Finally, use them as normal and pass them into `fuzzer.fuzz_loop(&mut stages, &mut executor, &mut state, &mut mgr)` to start fuzzing.
+
+## Parallel fuzzing
+
+In the [example fuzzer](https://github.com/AFLplusplus/LibAFL/blob/main/fuzzers/nyx_libxml2_parallel/src/main.rs). First you need to run `./setup_libxml2.sh` as described before.
+
+Parallel fuzzing relies on [`Launcher`](../message_passing/spawn_instances.md), so spawn logic should be written in the scoop of anonymous function `run_client`:
+
+```rust,ignore
+let mut run_client = |state: Option<_>, mut restarting_mgr, _core_id: usize| {}
+```
+
+In `run_client`, you need to create `NyxHelper` first:
+
+```rust,ignore
+let share_dir = Path::new("/tmp/nyx_libxml2/");
+let cpu_id = _core_id as u32;
+let parallel_mode = true;
+let mut helper = NyxHelper::new(
+    share_dir, // nyx work directory
+    cpu_id,    // current cpu id
+    true,      // open snap_mode
+    parallel_mode, // open parallel mode
+    Some(parent_cpu_id.id as u32), // the cpu-id of master instance, there is only one master instance, other instances will be treated as slaved
+)
+.unwrap();
+```
+
+Then you can fetch the trace_bits and create an observer and `NyxExecutor`
+
+```rust,ignore
+let trace_bits =
+    unsafe { std::slice::from_raw_parts_mut(helper.trace_bits, helper.map_size) };
+let observer = StdMapObserver::new("trace", trace_bits);
+let mut executor = NyxExecutor::new(&mut helper, tuple_list!(observer)).unwrap();
+```
+
+Finally, open a `Launcher` as normal to start fuzzing:
+
+```rust,ignore
+match Launcher::builder()
+    .shmem_provider(shmem_provider)
+    .configuration(EventConfig::from_name("default"))
+    .monitor(monitor)
+    .run_client(&mut run_client)
+    .cores(&cores)
+    .broker_port(broker_port)
+    // .stdout_file(Some("/dev/null"))
+    .build()
+    .launch()
+{
+    Ok(()) => (),
+    Err(Error::ShuttingDown) => println!("Fuzzing stopped by user. Good bye."),
+    Err(err) => panic!("Failed to run launcher: {:?}", err),
+}
+```

docs/src/baby_fuzzer/baby_fuzzer.md

@@ -378,4 +378,4 @@ Bye!
 
 As you can see, after the panic message, the `objectives` count of the log increased by one and you will find the crashing input in `crashes/`.
 
-The complete code can be found in `./fuzzers/baby_fuzzer`.
+The complete code can be found in [`./fuzzers/baby_fuzzer`](https://github.com/AFLplusplus/LibAFL/tree/main/fuzzers/baby_fuzzer) alongside other `baby_` fuzzers.

docs/src/baby_fuzzer/more_examples.md

@@ -1,4 +1,5 @@
 # More Examples
+
 Examples can be found under `./fuzzer`.
 
 |fuzzer name|usage|

docs/src/core_concepts/corpus.md

@@ -2,7 +2,7 @@
 
 The Corpus is where testcases are stored. We define a Testcase as an Input and a set of related metadata like execution time for instance.
 
-A Corpus can store testcases in diferent ways, for example on disk, or in memory, or implement a cache to speedup on disk storage.
+A Corpus can store testcases in different ways, for example on disk, or in memory, or implement a cache to speedup on disk storage.
 
 Usually, a testcase is added to the Corpus when it is considered as interesting, but a Corpus is used also to store testcases that fulfill an objective (like crashing the tested program for instance).
 

docs/src/core_concepts/executor.md

@@ -11,13 +11,13 @@ In our model, it can also hold a set of Observers connected with each execution.
 
 In Rust, we bind this concept to the [`Executor`](https://docs.rs/libafl/0/libafl/executors/trait.Executor.html) trait. A structure implementing this trait must implement [`HasObservers`](https://docs.rs/libafl/0/libafl/executors/trait.HasObservers.html) too if wants to hold a set of Observers.
 
-By default, we implement some commonly used Executors such as [`InProcessExecutor`](https://docs.rs/libafl/0/libafl/executors/inprocess/struct.InProcessExecutor.html) is which the target is a harness function providing in-process crash detection. Another Executor is the [`ForkserverExecutor`](https://docs.rs/libafl/0/libafl/executors/forkserver/struct.ForkserverExecutor.html) that implements an AFL-like mechanism to spawn child processes to fuzz.
+By default, we implement some commonly used Executors such as [`InProcessExecutor`](https://docs.rs/libafl/0/libafl/executors/inprocess/struct.InProcessExecutor.html) in which the target is a harness function providing in-process crash detection. Another Executor is the [`ForkserverExecutor`](https://docs.rs/libafl/0/libafl/executors/forkserver/struct.ForkserverExecutor.html) that implements an AFL-like mechanism to spawn child processes to fuzz.
 
 A common pattern when creating an Executor is wrapping an existing one, for instance [`TimeoutExecutor`](https://docs.rs/libafl/0.6.1/libafl/executors/timeout/struct.TimeoutExecutor.html) wraps an executor and install a timeout callback before calling the original run function of the wrapped executor.
 
 ## InProcessExecutor
 Let's begin with the base case; `InProcessExecutor`.
-This executor uses [_SanitizerCoverage_](https://clang.llvm.org/docs/SanitizerCoverage.html) as its backend, as you can find the related code in `libafl_targets/src/sancov_pcguards`. Here we allocate a map called `EDGES_MAP` and then our compiler wrapper compiles the harness to write the coverage into this map.
+This executor executes the harness program (function) inside the fuzzer process.
 
 When you want to execute the harness as fast as possible, you will most probably want to use this `InprocessExecutor`.
 
@@ -27,6 +27,7 @@ When you want to execute the harness as fast as possible, you will most probably
 Next, we'll take a look at the `ForkserverExecutor`. In this case, it is `afl-cc` (from AFLplusplus/AFLplusplus) that compiles the harness code, and therefore, we can't use `EDGES_MAP` anymore. Hopefully, we have [_a way_](https://github.com/AFLplusplus/AFLplusplus/blob/2e15661f184c77ac1fbb6f868c894e946cbb7f17/instrumentation/afl-compiler-rt.o.c#L270) to tell the forkserver which map to record the coverage.
 
 As you can see from the forkserver example,
+
 ```rust,ignore
 //Coverage map shared between observer and executor
 let mut shmem = StdShMemProvider::new().unwrap().new_shmem(MAP_SIZE).unwrap();
@@ -34,15 +35,16 @@ let mut shmem = StdShMemProvider::new().unwrap().new_shmem(MAP_SIZE).unwrap();
 shmem.write_to_env("__AFL_SHM_ID").unwrap();
 let mut shmem_buf = shmem.as_mut_slice();
 ```
+
 Here we make a shared memory region; `shmem`, and write this to environmental variable `__AFL_SHM_ID`. Then the instrumented binary, or the forkserver, finds this shared memory region (from the aforementioned env var) to record its coverage. On your fuzzer side, you can pass this shmem map to your `Observer` to obtain coverage feedbacks combined with any `Feedback`.
 
 Another feature of the `ForkserverExecutor` to mention is the shared memory testcases. In normal cases, the mutated input is passed between the forkserver and the instrumented binary via `.cur_input` file. You can improve your forkserver fuzzer's performance by passing the input with shared memory.  
 
+If the target is configured to use shared memory testcases, the `ForkserverExecutor` will notice this during the handshake and will automatically set up things accordingly.
 See AFL++'s [_documentation_](https://github.com/AFLplusplus/AFLplusplus/blob/stable/instrumentation/README.persistent_mode.md#5-shared-memory-fuzzing) or the fuzzer example in `forkserver_simple/src/program.c` for reference.
 
-It is very simple, when you call `ForkserverExecutor::new()` with `use_shmem_testcase` true, the `ForkserverExecutor` sets things up and your harness can just fetch the input from `__AFL_FUZZ_TESTCASE_BUF`
-
 ## InprocessForkExecutor
+
 Finally, we'll talk about the `InProcessForkExecutor`.
 `InProcessForkExecutor` has only one difference from `InprocessExecutor`; It forks before running the harness and that's it.
 
@@ -50,7 +52,7 @@ But why do we want to do so? well, under some circumstances, you may find your h
 
 However, we have to take care of the shared memory, it's the child process that runs the harness code and writes the coverage to the map.
 
-We have to make the map shared between the parent process and the child process, so we'll use shared memory again. You should compile your harness with `pointer_maps` (for `libafl_targes`) features enabled, this way, we can have a pointer; `EDGES_MAP_PTR` that can point to any coverage map.
+We have to make the map shared between the parent process and the child process, so we'll use shared memory again. You should compile your harness with `pointer_maps` (for `libafl_targets`) features enabled, this way, we can have a pointer; `EDGES_MAP_PTR` that can point to any coverage map.
 
 On your fuzzer side, you can allocate a shared memory region and make the `EDGES_MAP_PTR` point to your shared memory.
 

docs/src/core_concepts/feedback.md

@@ -3,7 +3,7 @@
 The Feedback is an entity that classifies the outcome of an execution of the program under test as interesting or not.
 Typically, if an execution is interesting, the corresponding input used to feed the target program is added to a corpus.
 
-Most of the times, the notion of Feedback is deeply linked to the Observer, but they are different concepts.
+Most of the time, the notion of Feedback is deeply linked to the Observer, but they are different concepts.
 
 The Feedback, in most of the cases, processes the information reported by one or more observers to decide if the execution is interesting.
 The concept of "interestingness" is abstract, but typically it is related to a novelty search (i.e. interesting inputs are those that reach a previously unseen edge in the control flow graph).
@@ -11,8 +11,16 @@ The concept of "interestingness" is abstract, but typically it is related to a n
 As an example, given an Observer that reports all the sizes of memory allocations, a maximization Feedback can be used to maximize these sizes to sport pathological inputs in terms of memory consumption.
 
 In terms of code, the library offers the [`Feedback`](https://docs.rs/libafl/0/libafl/feedbacks/trait.Feedback.html) and the [`FeedbackState`](https://docs.rs/libafl/0/libafl/feedbacks/trait.FeedbackState.html) traits.
-The first is used to implement functors that, given the state of the obversers from the last execution, tells if the execution was interesting. The second is tied with `Feedback` and it is the state of the data that the feedback wants to persist in the fuzzers's state, for instance the cumulative map holding all the edges seen so far in the case of a feedback based on edge coverage.
+The first is used to implement functors that, given the state of the observers from the last execution, tells if the execution was interesting. The second is tied with `Feedback` and it is the state of the data that the feedback wants to persist in the fuzzers's state, for instance the cumulative map holding all the edges seen so far in the case of a feedback based on edge coverage.
 
-Multiple Feedbacks can be combined into boolean formula, considering for instance an execution as interesting if it triggers new code paths or execute in less time compared to the average execution time using [`feedback_or`](https://docs.rs/libafl/0/libafl/macro.feedback_or.html).
+Multiple Feedbacks can be combined into boolean formula, considering for instance an execution as interesting if it triggers new code paths or execute in less time compared to the average execution time using [`feedback_or`](https://docs.rs/libafl/*/libafl/macro.feedback_or.html).
 
-TODO objective feedbacks and fast feedback logic operators
+On top, logic operators like `feedback_or` and `feedback_and` have a `_fast` option (`feedback_or_fast` where the second feedback will not be evaluated, if the first part already answers the `interestingness` question, to save precious performance.
+
+Using `feedback_and_fast` in combination with [`ConstFeedback`](https://docs.rs/libafl/*/libafl/feedbacks/enum.ConstFeedback.html#method.new), certain feedbacks can be disabled dynamically.
+
+## Objectives
+
+While feedbacks are commonly used to decide if an [`Input`](https://docs.rs/libafl/*/libafl/inputs/trait.Input.html) should be kept for future mutations, they serve a double-purpose, as so-called `Objective Feedbacks`.
+In this case, the `interestingness` of a feedback indicates, if an `Objective` has been hit.
+Commonly, these would be a`crash or a timeout, but they can also be used to find specific parts of the program, for sanitization, or a differential fuzzing success.

docs/src/core_concepts/input.md

@@ -1,6 +1,6 @@
 # Input
 
-Formally, the input of a program is the data taken from external sources that affect the program behaviour.
+Formally, the input of a program is the data taken from external sources that affect the program behavior.
 
 In our model of an abstract fuzzer, we define the Input as the internal representation of the program input (or a part of it).
 
@@ -10,4 +10,6 @@ But it is not always the case. A program can expect inputs that are not byte arr
 
 In case of a grammar fuzzer for instance, the Input is generally an Abstract Syntax Tree because it is a data structure that can be easily manipulated while maintaining the validity, but the program expects a byte array as input, so just before the execution, the tree is serialized to a sequence of bytes.
 
-In the Rust code, an [`Input`](https://docs.rs/libafl/0/libafl/inputs/trait.Input.html) is a trait that can be implemented only by structures that are serializable and have only owned data as fields.
+In the Rust code, an [`Input`](https://docs.rs/libafl/*/libafl/inputs/trait.Input.html) is a trait that can be implemented only by structures that are serializable and have only owned data as fields.
+
+While most fuzzer use a normal `BytesInput`], more advanced inputs like inputs include special inputs for grammar fuzzing ([GramatronInput](https://docs.rs/libafl/*/libafl/inputs/gramatron/struct.GramatronInput.html) or `NautilusInput` on nightly), as well as the token-level [EncodedInput](https://docs.rs/libafl/*/libafl/inputs/encoded/struct.EncodedInput.html).

docs/src/core_concepts/mutator.md

@@ -2,8 +2,8 @@
 
 The Mutator is an entity that takes one or more Inputs and generates a new derived one.
 
-Mutators can be composed and they are generally linked to a specific Input type.
+Mutators can be composed, and they are generally linked to a specific Input type.
 
 There can be, for instance, a Mutator that applies more than a single type of mutation on the input. Consider a generic Mutator for a byte stream, bit flip is just one of the possible mutations but not the only one, there is also, for instance, the random replacement of a byte of the copy of a chunk.
 
-In LibAFL, [`Mutator`](https://docs.rs/libafl/0/libafl/mutators/trait.Mutator.html) is a trait.
+In LibAFL, [`Mutator`](https://docs.rs/libafl/*/libafl/mutators/trait.Mutator.html) is a trait.

docs/src/core_concepts/observer.md

@@ -1,12 +1,14 @@
 # Observer
 
-An Observer, or Observation Channel, is an entity that provides an information observed during the execution of the program under test to the fuzzer.
+An Observer is an entity that provides an information observed during the execution of the program under test to the fuzzer.
 
-The information contained in the Observer is not preserved across executions.
+The information contained in the Observer is not preserved across executions, but it may be serialized and passed on to other nodes if an `Input` is considered `interesting`, and added to the `Corpus`.
 
-As an example, the coverage shared map filled during the execution to report the executed edges used by fuzzers such as AFL and HonggFuzz can be considered an Observation Channel.
+As an example, the coverage map, filled during the execution to report the executed edges used by fuzzers such as AFL and `HonggFuzz` can be considered an observation. Another `Observer` can be the time spent executing a run, the program output, or more advanced observation, like maximum stack depth at runtime.
-This information is not preserved across runs and it is an observation of a dynamic property of the program.
+This information is not preserved across runs, and it is an observation of a dynamic property of the program.
 
 In terms of code, in the library this entity is described by the [`Observer`](https://docs.rs/libafl/0/libafl/observers/trait.Observer.html) trait.
 
-In addition to holding the volatile data connected with the last execution of the target, the structures implementing this trait can define some execution hooks that are executed before and after each fuzz case. In this hooks, the observer can modify the fuzzer's state.
+In addition to holding the volatile data connected with the last execution of the target, the structures implementing this trait can define some execution hooks that are executed before and after each fuzz case. In these hooks, the observer can modify the fuzzer's state.
+
+The fuzzer will act based on these observers through a [`Feedback`](./feedback.md), that reduces the observation to the choice if a testcase is `interesting` for the fuzzer, or not.

docs/src/core_concepts/stage.md

@@ -6,4 +6,4 @@ For instance, a Mutational Stage, given an input of the corpus, applies a Mutato
 
 A stage can also be an analysis stage, for instance, the Colorization stage of Redqueen that aims to introduce more entropy in a testcase or the Trimming stage of AFL that aims to reduce the size of a testcase.
 
-There are several stages in the LibAFL codebases implementing the [`Stage`](https://docs.rs/libafl/0/libafl/stages/trait.Stage.html) trait.
+There are several stages in the LibAFL codebase implementing the [`Stage`](https://docs.rs/libafl/*/libafl/stages/trait.Stage.html) trait.

docs/src/design/architecture.md

@@ -2,14 +2,14 @@
 
 The LibAFL architecture is built around some entities to allow code reuse and low-cost abstractions.
 
-Initially, we started thinking to implement LibAFL in an Object Oriented language, such C++. When we landed to Rust, we immediately changed our idea as we realized that, while Rust allows a sort of OOP pattern, we can build the library using a more sane approach like the one described in [this blogpost](https://kyren.github.io/2018/09/14/rustconf-talk.html) about game design in Rust.
+Initially, we started thinking about implementing LibAFL in a traditional Object-Oriented language, like C++. When we switched to Rust, we immediately changed our idea as we realized that, we can build the library using a more rust-y approach, namely the one described in [this blogpost](https://kyren.github.io/2018/09/14/rustconf-talk.html) about game design in Rust.
 
-The LibAFL code reuse meachanism is so based on components rather than sub-classes, but there are still some OOP patterns in the library.
+The LibAFL code reuse mechanism is based on components, rather than sub-classes, but there are still some OOP patterns in the library.
 
-Thinking about similar fuzzers, you can observe that most of the times the data structures that are modified are the ones related to testcases and the fuzzer global state.
+Thinking about similar fuzzers, you can observe that most of the time the data structures that are modified are the ones related to testcases and the fuzzer global state.
 
 Beside the entities previously described, we introduce the [`Testcase`](https://docs.rs/libafl/0.6/libafl/corpus/testcase/struct.Testcase.html) and [`State`](https://docs.rs/libafl/0.6/libafl/state/struct.StdState.html) entities. The Testcase is a container for an Input stored in the Corpus and its metadata (so, in the implementation, the Corpus stores Testcases) and the State contains all the metadata that are evolved while running the fuzzer, Corpus included.
 
-The State, in the implementation, contains only owned objects that are serializable and it is serializable itself. Some fuzzers may want to serialize its state when pausing or just, when doing in-process fuzzing, serialize on crash and deserialize in the new process to continue to fuzz with all the metadata preserved.
+The State, in the implementation, contains only owned objects that are serializable, and it is serializable itself. Some fuzzers may want to serialize its state when pausing or just, when doing in-process fuzzing, serialize on crash and deserialize in the new process to continue to fuzz with all the metadata preserved.
 
-Additionally, we group the entities that are "actions", like the CorpusScheduler and the Feedbacks, in a common place, the [`Fuzzer'](https://docs.rs/libafl/0.6.1/libafl/fuzzer/struct.StdFuzzer.html).
+Additionally, we group the entities that are "actions", like the `CorpusScheduler` and the `Feedbacks`, in a common place, the [`Fuzzer'](https://docs.rs/libafl/*/libafl/fuzzer/struct.StdFuzzer.html).

docs/src/design/metadata.md

@@ -1,6 +1,6 @@
 # Metadata
 
-A metadata in LibAFL is a self contained structure that holds associated data to the State or to a Testcase.
+A metadata in LibAFL is a self-contained structure that holds associated data to the State or to a Testcase.
 
 In terms of code, a metadata can be defined as a Rust struct registered in the SerdeAny register.
 

docs/src/design/migration-0.9.md

@@ -0,0 +1,162 @@
+# Migrating from LibAFL <0.9 to 0.9
+
+Internal APIs of LibAFL have changed in version 0.9 to prefer associated types in cases where components were "fixed" to
+particular versions of other components. As a result, many existing custom components will not be compatible between
+versions prior to 0.9 and version 0.9.
+
+## Reasons for this change
+
+When implementing a trait with a generic, it is possible to have more than one instantiation of that generic trait. As a
+result, everywhere where consistency across generic types was required to implement a trait, it needed to be properly
+and explicitly constrained at every point. This led to `impl`s which were at best difficult to debug and, at worst,
+incorrect and caused confusing bugs for users.
+
+For example, consider the  `MapCorpusMinimizer` implementation (from <0.9) below:
+
+```rust,ignore
+impl<E, I, O, S, TS> CorpusMinimizer<I, S> for MapCorpusMinimizer<E, I, O, S, TS>
+where
+    E: Copy + Hash + Eq,
+    I: Input,
+    for<'a> O: MapObserver<Entry = E> + AsIter<'a, Item = E>,
+    S: HasMetadata + HasCorpus<I>,
+    TS: TestcaseScore<I, S>,
+{
+    fn minimize<CS, EX, EM, OT, Z>(
+        &self,
+        fuzzer: &mut Z,
+        executor: &mut EX,
+        manager: &mut EM,
+        state: &mut S,
+    ) -> Result<(), Error>
+    where
+        CS: Scheduler<I, S>,
+        EX: Executor<EM, I, S, Z> + HasObservers<I, OT, S>,
+        EM: EventManager<EX, I, S, Z>,
+        OT: ObserversTuple<S>,
+        Z: Evaluator<EX, EM, I, S> + HasScheduler<CS, I, S>,
+    {
+        // --- SNIP ---
+    }
+}
+```
+
+It was previously necessary to constrain every generic using a slew of other generics; above, it is necessary to
+constrain the input type (`I`) for every generic, despite the fact that this was already made clear by the state (`S`)
+and that the input will necessarily be the same over every implementation for that type.
+
+Below is the same code, but with the associated types changes (note that some generic names have changed):
+
+```rust,ignore
+impl<E, O, T, TS> CorpusMinimizer<E> for MapCorpusMinimizer<E, O, T, TS>
+where
+    E: UsesState,
+    for<'a> O: MapObserver<Entry = T> + AsIter<'a, Item = T>,
+    E::State: HasMetadata + HasCorpus,
+    T: Copy + Hash + Eq,
+    TS: TestcaseScore<E::State>,
+{
+    fn minimize<CS, EM, Z>(
+        &self,
+        fuzzer: &mut Z,
+        executor: &mut E,
+        manager: &mut EM,
+        state: &mut E::State,
+    ) -> Result<(), Error>
+    where
+        E: Executor<EM, Z> + HasObservers,
+        CS: Scheduler<State=E::State>,
+        EM: UsesState<State=E::State>,
+        Z: HasScheduler<CS, State=E::State>,
+    {
+        // --- SNIP ---
+    }
+}
+```
+
+The executor is constrained to `EM` and `Z`, with each of their respective states being constrained to `E`'s state. It
+is no longer necessary to explicitly defined a generic for the input type, the state type, or the generic type, as these
+are all present as associated types for `E`. Additionally, we don't even need to specify any details about the observers
+(`OT` in the previous version) as the type does not need to be constrained and is not shared by other types.
+
+## Scope
+
+You are affected by this change if:
+
+- You specified explicit generics for a type (e.g., `MaxMapFeedback::<_, (), _>::new(...)`)
+- You implemented a custom component (e.g., `Mutator`, `Executor`, `State`, `Fuzzer`, `Feedback`, `Observer`, etc.)
+
+If you did neither of these, congrats! You are likely unaffected by these changes.
+
+### Migrating explicit generics
+
+Migrating specific generics should be a quite simple process; you should review the API documentation for details on the
+order of generics and replace them accordingly. Generally speaking, it should no longer be necessary to specify these
+generics.
+
+See `fuzzers/` for examples of these changes.
+
+### Migrating component types
+
+If you implemented a Mutator, Executor, State, or another kind of component, you must update your implementation. The
+main changes to the API are in the use of "Uses*" for associated types.
+
+In many scenarios, Input, Observers, and State generics have been moved into traits with associated types (namely,
+"UsesInput", "UsesObservers", and "UsesState". These traits are required for many existing traits now and are very
+straightforward to implement. In a majority of cases, you will have generics on your custom implementation or a fixed
+type to implement this with. Thankfully, Rust will let you know when you need to implement this type.
+
+As an example, `InMemoryCorpus` before 0.9 looked like this:
+
+```rust,ignore
+#[derive(Default, Serialize, Deserialize, Clone, Debug)]
+#[serde(bound = "I: serde::de::DeserializeOwned")]
+pub struct InMemoryCorpus<I>
+where
+    I: Input,
+{
+    entries: Vec<RefCell<Testcase<I>>>,
+    current: Option<usize>,
+}
+
+impl<I> Corpus<I> for InMemoryCorpus<I>
+where
+    I: Input,
+{
+    // --- SNIP ---
+}
+```
+
+After 0.9, all `Corpus` implementations are required to implement `UsesInput` and `Corpus` no longer has a generic for
+the input type (as it is now provided by the UsesInput impl). The migrated implementation is shown below:
+
+```rust,ignore
+#[derive(Default, Serialize, Deserialize, Clone, Debug)]
+#[serde(bound = "I: serde::de::DeserializeOwned")]
+pub struct InMemoryCorpus<I>
+where
+    I: Input,
+{
+    entries: Vec<RefCell<Testcase<I>>>,
+    current: Option<usize>,
+}
+
+impl<I> UsesInput for InMemoryCorpus<I>
+where
+    I: Input,
+{
+    type Input = I;
+}
+
+impl<I> Corpus for InMemoryCorpus<I>
+where
+    I: Input,
+{
+    // --- SNIP ---
+}
+```
+
+Now, `Corpus` cannot be accidentally implemented for another type other than that specified by `InMemoryCorpus`, as it
+is fixed to the associated type for `UsesInput`.
+
+A more complex example of migration can be found in the "Reasons for this change" section of this document.

docs/src/getting_started/crates.md

@@ -8,6 +8,8 @@ A crate is an individual library in Rust's Cargo build system, that you can use
 libafl = { version = "*" }
 ```
 
+## Crate List
+
 For LibAFL, each crate has its self-contained purpose, and the user may not need to use all of them in its project.
 Following the naming convention of the folders in the project's root, they are:
 
@@ -19,13 +21,13 @@ This crate has a number of feature flags that enable and disable certain aspects
 The features can be found in [LibAFL's `Cargo.toml`](https://github.com/AFLplusplus/LibAFL/blob/main/libafl/Cargo.toml) under "`[features]`", and are usually explained with comments there.
 Some features worthy of remark are:
 
-- `std` enables the parts of the code that use the Rust standard library. Without this flag, LibAFL is `no_std` compatible. This disables a range of features, but allows us to use LibAFL in embedded environments, read [the `no_std` section](../advanced_features/no_std/no_std.md) for further details.
+- `std` enables the parts of the code that use the Rust standard library. Without this flag, LibAFL is `no_std` compatible. This disables a range of features, but allows us to use LibAFL in embedded environments, read [the `no_std` section](../advanced_features/no_std.md) for further details.
 - `derive` enables the usage of the `derive(...)` macros defined in libafl_derive from libafl.
 - `rand_trait` allows you to use LibAFL's very fast (*but insecure!*) random number generator wherever compatibility with Rust's [`rand` crate](https://crates.io/crates/rand) is needed.
 - `llmp_bind_public` makes LibAFL's LLMP bind to a public TCP port, over which other fuzzers nodes can communicate with this instance.
 - `introspection` adds performance statistics to LibAFL.
 
-You can chose the features by using `features = ["feature1", "feature2", ...]` for LibAFL in your `Cargo.toml`.
+You can choose the features by using `features = ["feature1", "feature2", ...]` for LibAFL in your `Cargo.toml`.
 Out of this list, by default, `std`, `derive`, and `rand_trait` are already set.
 You can choose to disable them by setting `default-features = false` in your `Cargo.toml`.
 
@@ -64,10 +66,9 @@ To understand it deeper, look through the tutorials and examples.
 ### libafl_frida
 
 This library bridges LibAFL with Frida as instrumentation backend.
-
 With this crate, you can instrument targets on Linux/macOS/Windows/Android for coverage collection.
-
 Additionally, it supports CmpLog, and AddressSanitizer instrumentation and runtimes for aarch64.
+See further information, as well as usage instructions, [later in the book](../advanced_features/frida.md).
 
 ### libafl_qemu
 
@@ -75,3 +76,13 @@ This library bridges LibAFL with QEMU user-mode to fuzz ELF cross-platform binar
 
 It works on Linux and can collect edge coverage without collisions!
 It also supports a wide range of hooks and instrumentation options.
+
+### libafl_nyx
+
+[Nyx](https://nyx-fuzz.com/) is a KVM-based snapshot fuzzer. `libafl_nyx` adds these capabilities to LibAFL. There is a specific section explaining usage of libafl_nyx [later in the book](../advanced_features/nyx.md).
+
+### libafl_concolic
+
+Concolic fuzzing is the combination of fuzzing and a symbolic execution engine.
+This can reach greater depth than normal fuzzing, and is exposed in this crate.
+There is a specific section explaining usage of libafl_concolic [later in the book](../advanced_features/concolic.md).

docs/src/getting_started/getting_started.md

@@ -1,5 +1,5 @@
 # Getting Started
 
-To get startes with LibAFL, there are some initial steps to do.
+To get started with LibAFL, there are some initial steps to take.
 In this chapter, we discuss how to download and build LibAFL, using Rust's `cargo` command.
 We also describe the structure of LibAFL's components, so-called crates, and the purpose of each individual crate.

docs/src/getting_started/setup.md

@@ -22,7 +22,7 @@ $ git clone git@github.com:AFLplusplus/LibAFL.git
 You can alternatively, on a UNIX-like machine, download a compressed archive and extract it with:
 
 ```sh
-$ wget https://github.com/AFLplusplus/LibAFL/archive/main.tar.gz
+wget https://github.com/AFLplusplus/LibAFL/archive/main.tar.gz
 $ tar xvf LibAFL-main.tar.gz
 $ rm LibAFL-main.tar.gz
 $ ls LibAFL-main # this is the extracted folder

docs/src/introduction.md

@@ -18,6 +18,7 @@ Be it a specific target, a particular instrumentation backend, or a custom mutat
 
 LibAFL gives you many of the benefits of an off-the-shelf fuzzer, while being completely customizable.
 Some highlight features currently include:
+
 - `multi platform`: LibAFL works pretty much anywhere you can find a Rust compiler for. We already used it on *Windows*, *Android*, *MacOS*, and *Linux*, on *x86_64*, *aarch64*, ...
 - `portable`: `LibAFL` can be built in `no_std` mode.
 This means it does not require a specific OS-dependent runtime to function.

docs/src/message_passing/configurations.md

@@ -5,6 +5,5 @@ The chapter describes how to run nodes with different configurations
 in one fuzzing cluster.
 This allows, for example, a node compiled with ASAN, to know that it needs to rerun new testcases for a node without ASAN, while the same binary/configuration does not.
 
-> ## Under Construction!
+Fuzzers with the same configuration can exchange Observers for new testcases and reuse them without rerunning the input.
-> This section is under construction.
+A different configuration indicates, that only the raw input can be exchanged, it must be rerun on the other node to capture relevant observations.
-> Please check back later (or open a PR)

docs/src/message_passing/message_passing.md

@@ -36,6 +36,7 @@ It then sends the information needed to map the newly-allocated page in connecte
 Once the receiver maps the new page, flags it as safe for unmapping from the sending process (to avoid race conditions if we have more than a single EOP in a short time), and then continues to read from the new `ShMem`.
 
 The schema for client's maps to the broker is as follows:
+
 ```text
 [client0]        [client1]    ...    [clientN]
   |                  |                 /

docs/src/message_passing/spawn_instances.md

@@ -17,11 +17,13 @@ Launching nodes manually has the benefit that you can have multiple nodes with d
 
 While it's called "restarting" manager, it uses `fork` on Unix operating systems as optimization and only actually restarts from scratch on Windows.
 
-## Launcher
+
+## Automated, with Launcher
 
 The Launcher is the lazy way to do multiprocessing.
-You can use the Launcher builder to create a fuzzer that spawns multiple nodes, all using restarting event managers.
+You can use the Launcher builder to create a fuzzer that spawns multiple nodes with one click, all using restarting event managers and the same configuration.
-An example may look like this:
+
+To use launcher, first you need to write an anonymous function `let mut run_client = |state: Option<_>, mut mgr, _core_id|{}`, which uses three parameters to create individual fuzzer. Then you can specify the `shmem_provider`,`broker_port`,`monitor`,`cores` and other stuff through `Launcher::builder()`:
 
 ```rust,ignore
     Launcher::builder()
@@ -42,8 +44,16 @@ The value is a string indicating the cores to bind to, for example, `0,2,5` or `
 For each client, `run_client` will be called.
 On Windows, the Launcher will restart each client, while on Unix, it will use `fork`.
 
+Advanced use-cases:
+
+1. To connect multiple nodes together via TCP, you can use the `remote_broker_addr`. this requires the `llmp_bind_public` compile-time feature for `LibAFL`.
+2. To use multiple launchers for individual configurations, you can set `spawn_broker` to `false` on all but one.
+3. Launcher will not select the cores automatically, so you need to specify the `cores` that you want.
+
+For more examples, you can check out `qemu_launcher` and `libfuzzer_libpng_launcher` in [`./fuzzers/`](https://github.com/AFLplusplus/LibAFL/tree/main/fuzzers).
+
 ## Other ways
 
-The LlmpEvenManager family is the easiest way to do spawn instances, but for obscure targets, you may need to come up with other solutions.
+The `LlmpEventManager` family is the easiest way to spawn instances, but for obscure targets, you may need to come up with other solutions.
 LLMP is even, in theory, `no_std` compatible, and even completely different EventManagers can be used for message passing.
 If you are in this situation, please either read through the current implementations and/or reach out to us.

docs/src/tutorial/intro.md

@@ -1,5 +1,8 @@
 # Introduction
 
 > ## Under Construction!
+>
 > This section is under construction.
 > Please check back later (or open a PR)
+>
+> In the meantime, find the final Lain-based fuzzer in [the fuzzers folder](https://github.com/AFLplusplus/LibAFL/tree/main/fuzzers/tutorial)

fuzzers/FRET/.gitignore

@@ -0,0 +1,4 @@
+*.qcow2
+corpus
+*.axf
+demo

fuzzers/FRET/Cargo.toml

@@ -0,0 +1,41 @@
+[package]
+name = "fret"
+version = "0.8.2"
+authors = ["Andrea Fioraldi <andreafioraldi@gmail.com>", "Dominik Maier <domenukk@gmail.com>"]
+edition = "2021"
+
+[features]
+default = ["std", "snapshot_restore", "singlecore", "restarting", "feed_systemtrace", "fuzz_int" ]
+std = []
+snapshot_restore = []
+snapshot_fast = [ "snapshot_restore" ]
+singlecore = []
+restarting = ['singlecore']
+trace_abbs = []
+systemstate = []
+feed_systemgraph = [ "systemstate" ]
+feed_systemtrace = [ "systemstate" ]
+feed_longest = [ ]
+feed_afl = [ ]
+feed_genetic = [ ]
+fuzz_int = [ ]
+gensize_1 = [ ]
+gensize_10 = [ ]
+gensize_100 = [ ]
+observer_hitcounts = []
+no_hash_state = []
+run_until_saturation = []
+
+[profile.release]
+lto = true
+codegen-units = 1
+debug = true
+
+[dependencies]
+libafl = { path = "../../libafl/" }
+libafl_qemu = { path = "../../libafl_qemu/", features = ["arm", "systemmode"] }
+serde = { version = "1.0", default-features = false, features = ["alloc"] } # serialization lib
+hashbrown =  { version = "0.12", features = ["serde", "ahash-compile-time-rng"] } # A faster hashmap, nostd compatible
+petgraph = { version="0.6.0", features = ["serde-1"] }
+ron = "0.7" # write serialized data - including hashmaps
+rand = "0.5"

fuzzers/FRET/README.md

@@ -0,0 +1,26 @@
+# Qemu systemmode with launcher
+
+This folder contains an example fuzzer for the qemu systemmode, using LLMP for fast multi-process fuzzing and crash detection.
+
+## Build
+
+To build this example, run
+
+```bash
+cargo build --release
+cd example; sh build.sh; cd ..
+```
+
+This will build the the fuzzer (src/fuzzer.rs) and a small example binary based on FreeRTOS, which can run under a qemu emulation target.
+
+## Run
+
+Since the instrumentation is based on snapshtos QEMU needs a virtual drive (even if it is unused...).
+Create on and then run the fuzzer:
+```bash
+# create an image
+qemu-img create -f qcow2 dummy.qcow2 32M
+# run the fuzzer
+KERNEL=./example/example.elf target/release/qemu_systemmode -icount shift=auto,align=off,sleep=off -machine mps2-an385 -monitor null -kernel ./example/example.elf -serial null -nographic -snapshot -drive if=none,format=qcow2,file=dummy.qcow2 -S
+```
+Currently the ``KERNEL`` variable is needed because the fuzzer does not parse QEMUs arguments to find the binary.

fuzzers/FRET/benchmark/.gitignore

@@ -0,0 +1,12 @@
+*dump
+timedump*
+corpora
+build
+mnt
+.R*
+*.png
+*.pdf
+bins
+.snakemake
+*.zip
+*.tar.*

fuzzers/FRET/benchmark/Makefile

@@ -0,0 +1,57 @@
+TIME=7200
+
+corpora/%/seed:
+	mkdir -p $$(dirname $@)
+	LINE=$$(grep "^$$(basename $*)" target_symbols.csv); \
+	export \
+	KERNEL=benchmark/build/$*.elf \
+	FUZZ_MAIN=$$(echo $$LINE | cut -d, -f2) \
+	FUZZ_INPUT=$$(echo $$LINE | cut -d, -f3) \
+	FUZZ_INPUT_LEN=$$(echo $$LINE | cut -d, -f4) \
+	BREAKPOINT=$$(echo $$LINE | cut -d, -f5) \
+	SEED_DIR=benchmark/corpora/$* \
+	DUMP_SEED=seed; \
+	../fuzzer.sh
+
+timedump/%$(FUZZ_RANDOM)$(SUFFIX): corpora/%/seed
+	mkdir -p $$(dirname $@)
+	LINE=$$(grep "^$$(basename $*)" target_symbols.csv); \
+	export \
+	KERNEL=benchmark/build/$*.elf \
+	FUZZ_MAIN=$$(echo $$LINE | cut -d, -f2) \
+	FUZZ_INPUT=$$(echo $$LINE | cut -d, -f3) \
+	FUZZ_INPUT_LEN=$$(echo $$LINE | cut -d, -f4) \
+	BREAKPOINT=$$(echo $$LINE | cut -d, -f5) \
+	SEED_RANDOM=1 \
+	TIME_DUMP=benchmark/$@ \
+	CASE_DUMP=benchmark/$@; \
+	../fuzzer.sh + + + + + $(TIME) + + + > $@_log
+	#SEED_DIR=benchmark/corpora/$*
+
+all_sequential: timedump/sequential/mpeg2$(FUZZ_RANDOM) timedump/sequential/dijkstra$(FUZZ_RANDOM) timedump/sequential/epic$(FUZZ_RANDOM) \
+	timedump/sequential/g723_enc$(FUZZ_RANDOM) timedump/sequential/audiobeam$(FUZZ_RANDOM) \
+	timedump/sequential/gsm_enc$(FUZZ_RANDOM)
+
+all_kernel: timedump/kernel/bsort$(FUZZ_RANDOM) timedump/kernel/insertsort$(FUZZ_RANDOM) #timedump/kernel/fft$(FUZZ_RANDOM)
+
+all_app: timedump/app/lift$(FUZZ_RANDOM)
+
+all_system: timedump/lift$(FUZZ_RANDOM)$(SUFFIX)
+
+all_period: timedump/waters$(FUZZ_RANDOM)$(SUFFIX)
+
+tacle_rtos: timedump/tacle_rtos$(FUZZ_RANDOM)
+
+graphics:
+	Rscript --vanilla plot_comparison.r mnt/timedump/sequential audiobeam
+	Rscript --vanilla plot_comparison.r mnt/timedump/sequential dijkstra
+	Rscript --vanilla plot_comparison.r mnt/timedump/sequential epic
+	Rscript --vanilla plot_comparison.r mnt/timedump/sequential g723_enc
+	# Rscript --vanilla plot_comparison.r mnt/timedump/sequential gsm_enc
+	# Rscript --vanilla plot_comparison.r mnt/timedump/sequential huff_dec
+	Rscript --vanilla plot_comparison.r mnt/timedump/sequential mpeg2
+	# Rscript --vanilla plot_comparison.r mnt/timedump/sequential rijndael_dec
+	# Rscript --vanilla plot_comparison.r mnt/timedump/sequential rijndael_enc
+
+clean:
+	rm -rf corpora timedump

fuzzers/FRET/benchmark/Snakefile

@@ -0,0 +1,281 @@
+import csv
+import os
+def_flags="--no-default-features --features std,snapshot_restore,singlecore,restarting,run_until_saturation"
+remote="timedump_253048_1873f6_all/"
+RUNTIME=10
+TARGET_REPS_A=2
+TARGET_REPS_B=2
+NUM_NODES=2
+REP_PER_NODE_A=int(TARGET_REPS_A/NUM_NODES)
+REP_PER_NODE_B=int(TARGET_REPS_B/NUM_NODES)
+NODE_ID= 0 if os.getenv('NODE_ID') == None else int(os.environ['NODE_ID'])
+MY_RANGE_A=range(NODE_ID*REP_PER_NODE_A,(NODE_ID+1)*REP_PER_NODE_A)
+MY_RANGE_B=range(NODE_ID*REP_PER_NODE_B,(NODE_ID+1)*REP_PER_NODE_B)
+
+rule build_showmap:
+    output:
+        directory("bins/target_showmap")
+    shell:
+        "cargo build --target-dir {output} {def_flags},systemstate"
+
+rule build_random:
+    output:
+        directory("bins/target_random")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_longest"
+
+rule build_feedlongest:
+    output:
+        directory("bins/target_feedlongest")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_longest"
+
+rule build_frafl:
+    output:
+        directory("bins/target_frafl")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_afl,feed_longest"
+
+rule build_afl:
+    output:
+        directory("bins/target_afl")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_afl,observer_hitcounts"
+
+rule build_state:
+    output:
+        directory("bins/target_state")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_systemtrace"
+
+rule build_nohashstate:
+    output:
+        directory("bins/target_nohashstate")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_systemtrace,no_hash_state"
+
+rule build_graph:
+    output:
+        directory("bins/target_graph")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_systemgraph"
+
+rule build_showmap_int:
+    output:
+        directory("bins/target_showmap_int")
+    shell:
+        "cargo build --target-dir {output} {def_flags},systemstate,fuzz_int"
+
+rule build_random_int:
+    output:
+        directory("bins/target_random_int")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_longest,fuzz_int"
+
+rule build_state_int:
+    output:
+        directory("bins/target_state_int")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_systemtrace,fuzz_int"
+
+rule build_nohashstate_int:
+    output:
+        directory("bins/target_nohashstate_int")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_systemtrace,fuzz_int,no_hash_state"
+
+rule build_frafl_int:
+    output:
+        directory("bins/target_frafl_int")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_afl,feed_longest,fuzz_int"
+
+rule build_afl_int:
+    output:
+        directory("bins/target_afl_int")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_afl,fuzz_int,observer_hitcounts"
+
+rule build_feedlongest_int:
+    output:
+        directory("bins/target_feedlongest_int")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_longest,fuzz_int"
+
+rule build_feedgeneration1:
+    output:
+        directory("bins/target_feedgeneration1")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_genetic,gensize_1"
+
+rule build_feedgeneration1_int:
+    output:
+        directory("bins/target_feedgeneration1_int")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_genetic,fuzz_int,gensize_1"
+
+rule build_feedgeneration10:
+    output:
+        directory("bins/target_feedgeneration10")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_genetic,gensize_10"
+
+rule build_feedgeneration10_int:
+    output:
+        directory("bins/target_feedgeneration10_int")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_genetic,fuzz_int,gensize_10"
+
+rule build_feedgeneration100:
+    output:
+        directory("bins/target_feedgeneration100")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_genetic,gensize_100"
+
+rule build_feedgeneration100_int:
+    output:
+        directory("bins/target_feedgeneration100_int")
+    shell:
+        "cargo build --target-dir {output} {def_flags},feed_genetic,fuzz_int,gensize_100"
+
+rule run_bench:
+    input:
+        "build/{target}.elf",
+        "bins/target_{fuzzer}"
+    output:
+        multiext("timedump/{fuzzer}/{target}.{num}", "", ".log") # , ".case"
+    run:
+        with open('target_symbols.csv') as csvfile:
+            reader = csv.DictReader(csvfile)
+            line = next((x for x in reader if x['kernel']==wildcards.target), None)
+            if line == None:
+                return False
+            kernel=line['kernel']
+            fuzz_main=line['main_function']
+            fuzz_input=line['input_symbol']
+            fuzz_len=line['input_size']
+            bkp=line['return_function']
+            script="""
+                mkdir -p $(dirname {output[0]})
+                export KERNEL=$(pwd)/{input[0]}
+                export FUZZ_MAIN={fuzz_main}
+                export FUZZ_INPUT={fuzz_input}
+                export FUZZ_INPUT_LEN={fuzz_len}
+                export BREAKPOINT={bkp}
+                export SEED_RANDOM={wildcards.num}
+                export TIME_DUMP=$(pwd)/{output[0]}
+                export CASE_DUMP=$(pwd)/{output[0]}.case
+                export TRACE_DUMP=$(pwd)/{output[0]}.trace
+                export FUZZ_ITERS={RUNTIME}
+                export FUZZER=$(pwd)/{input[1]}/debug/fret
+                set +e
+                ../fuzzer.sh > {output[1]} 2>&1 
+                exit 0
+                """
+            if wildcards.fuzzer.find('random') >= 0:
+                script="export FUZZ_RANDOM={output[1]}\n"+script
+            shell(script)
+
+rule run_showmap:
+    input:
+        "{remote}build/{target}.elf",
+        "bins/target_showmap",
+        "bins/target_showmap_int",
+        "{remote}timedump/{fuzzer}/{target}.{num}.case"
+    output:
+        "{remote}timedump/{fuzzer}/{target}.{num}.trace.ron",
+        "{remote}timedump/{fuzzer}/{target}.{num}.case.time",
+    run:
+        with open('target_symbols.csv') as csvfile:
+            reader = csv.DictReader(csvfile)
+            line = next((x for x in reader if x['kernel']==wildcards.target), None)
+            if line == None:
+                return False
+            kernel=line['kernel']
+            fuzz_main=line['main_function']
+            fuzz_input=line['input_symbol']
+            fuzz_len=line['input_size']
+            bkp=line['return_function']
+            script=""
+            if wildcards.fuzzer.find('_int') > -1:
+                script="export FUZZER=$(pwd)/{input[2]}/debug/fret\n"
+            else:
+                script="export FUZZER=$(pwd)/{input[1]}/debug/fret\n"
+            script+="""
+                mkdir -p $(dirname {output})
+                export KERNEL=$(pwd)/{input[0]}
+                export FUZZ_MAIN={fuzz_main}
+                export FUZZ_INPUT={fuzz_input}
+                export FUZZ_INPUT_LEN={fuzz_len}
+                export BREAKPOINT={bkp}
+                export TRACE_DUMP=$(pwd)/{output[0]}
+                export DO_SHOWMAP=$(pwd)/{input[3]}
+                export TIME_DUMP=$(pwd)/{output[1]}
+                set +e
+                ../fuzzer.sh
+                exit 0
+                """
+            if wildcards.fuzzer.find('random') >= 0:
+                script="export FUZZ_RANDOM=1\n"+script
+            shell(script)
+
+rule tarnsform_trace:
+    input:
+        "{remote}timedump/{fuzzer}/{target}.{num}.trace.ron"
+    output:
+        "{remote}timedump/{fuzzer}/{target}.{num}.trace.csv"
+    shell:
+        "$(pwd)/../../../../state2gantt/target/debug/state2gantt {input} > {output[0]}"
+
+rule trace2gantt:
+    input:
+        "{remote}timedump/{fuzzer}/{target}.{num}.trace.csv"
+    output:
+        "{remote}timedump/{fuzzer}/{target}.{num}.trace.csv.png"
+    shell:
+        "Rscript --vanilla $(pwd)/../../../../state2gantt/gantt.R {input}"
+
+rule all_main:
+    input:
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['random','afl','feedgeneration10','state'], target=['waters','watersv2'],num=range(0,3))
+
+rule all_main_int:
+    input:
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['random_int','afl_int','feedgeneration10_int','state_int'], target=['waters_int','watersv2_int'],num=range(0,4))
+
+rule all_compare_feedgeneration:
+    input:
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['feedgeneration1','feedgeneration10','feedgeneration100'], target=['waters_int','watersv2'],num=range(0,10))
+
+rule all_compare_feedgeneration_int:
+    input:
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['feedgeneration1_int','feedgeneration10_int','feedgeneration100_int'], target=['waters_int','watersv2_int'],num=range(0,10))
+
+rule all_compare_afl:
+    input:
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['afl','frafl','feedlongest'], target=['waters','watersv2'],num=range(0,10))
+
+rule all_compare_afl_int:
+    input:
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['afl_int','frafl_int','feedlongest_int'], target=['waters_int','watersv2_int'],num=range(0,10))
+
+rule all_images:
+    input:
+        expand("{remote}timedump/{fuzzer}/{target}.{num}.trace.csv.png",remote=remote, fuzzer=['afl','feedgeneration10','state'], target=['waters','watersv2'],num=range(0,3))
+
+rule all_images_int:
+    input:
+        expand("{remote}timedump/{fuzzer}/{target}.{num}.trace.csv.png",remote=remote, fuzzer=['afl_int','feedgeneration10_int','state_int'], target=['waters_int','watersv2_int'],num=range(0,3))
+
+rule clusterfuzz:
+    input:
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['random','afl','feedgeneration10','state'], target=['waters','watersv2'],num=MY_RANGE_A),
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['random_int','afl_int','feedgeneration10_int','state_int'], target=['waters_int','watersv2_int'],num=MY_RANGE_A),
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['feedgeneration1','feedgeneration10','feedgeneration100'], target=['waters_int','watersv2'],num=MY_RANGE_B),
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['feedgeneration1_int','feedgeneration10_int','feedgeneration100_int'], target=['waters_int','watersv2_int'],num=MY_RANGE_B),
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['afl','frafl','feedlongest'], target=['waters','watersv2'],num=MY_RANGE_B),
+        expand("timedump/{fuzzer}/{target}.{num}", fuzzer=['afl_int','frafl_int','feedlongest_int'], target=['waters_int','watersv2_int'],num=MY_RANGE_B),
+
+rule all_bins:
+    input:
+        expand("bins/target_{target}{flag}",target=['random','afl','frafl','state','feedgeneration100'],flag=['','_int'])

fuzzers/FRET/benchmark/plot_comparison.r

@@ -0,0 +1,83 @@
+library("mosaic")
+args = commandArgs(trailingOnly=TRUE)
+
+#myolors=c("#339933","#0066ff","#993300") # grün, balu, rot
+myolors=c("dark green","dark blue","dark red", "yellow") # grün, balu, rot
+
+if (length(args)==0) {
+  runtype="timedump"
+  target="waters"
+  filename_1=sprintf("%s.png",target)
+  filename_2=sprintf("%s_maxline.png",target)
+  filename_3=sprintf("%s_hist.png",target)
+} else {
+  runtype=args[1]
+  target=args[2]
+  filename_1=sprintf("%s.png",args[2])
+  filename_2=sprintf("%s_maxline.png",args[2])
+  filename_3=sprintf("%s_hist.png",args[2])
+  # filename_1=args[3]
+}
+
+file_1=sprintf("~/code/FRET/LibAFL/fuzzers/FRET/benchmark/%s/%s_state",runtype,target)
+file_2=sprintf("~/code/FRET/LibAFL/fuzzers/FRET/benchmark/%s/%s_afl",runtype,target)
+file_3=sprintf("~/code/FRET/LibAFL/fuzzers/FRET/benchmark/%s/%s_random",runtype,target)
+file_4=sprintf("~/code/FRET/LibAFL/fuzzers/FRET/benchmark/%s/%s_graph",runtype,target)
+timetrace <- read.table(file_1, quote="\"", comment.char="")
+timetrace_afl <- read.table(file_2, quote="\"", comment.char="")
+timetrace_rand <- read.table(file_3, quote="\"", comment.char="")
+timetrace_graph <- read.table(file_4, quote="\"", comment.char="")
+timetrace[[2]]=seq_len(length(timetrace[[1]]))
+timetrace_afl[[2]]=seq_len(length(timetrace_afl[[1]]))
+timetrace_rand[[2]]=seq_len(length(timetrace_rand[[1]]))
+timetrace_graph[[2]]=seq_len(length(timetrace_graph[[1]]))
+names(timetrace)[1] <- "timetrace"
+names(timetrace)[2] <- "iter"
+names(timetrace_afl)[1] <- "timetrace"
+names(timetrace_afl)[2] <- "iter"
+names(timetrace_rand)[1] <- "timetrace"
+names(timetrace_rand)[2] <- "iter"
+names(timetrace_graph)[1] <- "timetrace"
+names(timetrace_graph)[2] <- "iter"
+
+png(file=filename_1)
+# pdf(file=filename_1,width=8, height=8)
+plot(timetrace[[2]],timetrace[[1]], col=myolors[1], xlab="iters", ylab="wcet", pch='.')
+points(timetrace_afl[[2]],timetrace_afl[[1]], col=myolors[2], pch='.')
+points(timetrace_rand[[2]],timetrace_rand[[1]], col=myolors[3], pch='.')
+points(timetrace_graph[[2]],timetrace_graph[[1]], col=myolors[4], pch='.')
+abline(lm(timetrace ~ iter, data=timetrace),col=myolors[1])
+abline(lm(timetrace ~ iter, data=timetrace_afl),col=myolors[2])
+abline(lm(timetrace ~ iter, data=timetrace_rand),col=myolors[3])
+dev.off()
+
+png(file=filename_3)
+gf_histogram(~ timetrace,data=timetrace, fill=myolors[1]) %>%
+gf_histogram(~ timetrace,data=timetrace_afl, fill=myolors[2]) %>%
+gf_histogram(~ timetrace,data=timetrace_rand, fill=myolors[3]) %>%
+gf_histogram(~ timetrace,data=timetrace_graph, fill=myolors[4])
+dev.off()
+
+# Takes a flat list
+trace2maxline <- function(tr) {
+  maxline = tr
+  for (var in seq_len(length(maxline))[2:length(maxline)]) {
+    maxline[var] = max(maxline[var],maxline[var-1])
+  }
+  #plot(seq_len(length(maxline)),maxline,"l",xlab="Index",ylab="WOET")
+  return(maxline)
+}
+timetrace[[1]] <- trace2maxline(timetrace[[1]])
+timetrace_afl[[1]] <- trace2maxline(timetrace_afl[[1]])
+timetrace_rand[[1]] <- trace2maxline(timetrace_rand[[1]])
+timetrace_graph[[1]] <- trace2maxline(timetrace_graph[[1]])
+
+png(file=filename_2)
+plot(timetrace[[2]],timetrace[[1]], col=myolors[1], xlab="iters", ylab="wcet", pch='.')
+points(timetrace_afl[[2]],timetrace_afl[[1]], col=myolors[2], pch='.')
+points(timetrace_rand[[2]],timetrace_rand[[1]], col=myolors[3], pch='.')
+points(timetrace_graph[[2]],timetrace_graph[[1]], col=myolors[4], pch='.')
+#abline(lm(timetrace ~ iter, data=timetrace),col=myolors[1])
+#abline(lm(timetrace ~ iter, data=timetrace_afl),col=myolors[2])
+#abline(lm(timetrace ~ iter, data=timetrace_rand),col=myolors[3])
+dev.off()

fuzzers/FRET/benchmark/plot_multi.r

@@ -0,0 +1,327 @@
+library("mosaic")
+library("dplyr")
+library("foreach")
+library("doParallel")
+
+#setup parallel backend to use many processors
+cores=detectCores()
+cl <- makeCluster(cores[1]-1) #not to overload your computer
+registerDoParallel(cl)
+
+args = commandArgs(trailingOnly=TRUE)
+
+if (length(args)==0) {
+  runtype="timedump_253048_1873f6_all/timedump"
+  target="waters_int"
+  outputpath="~/code/FRET/LibAFL/fuzzers/FRET/benchmark/"
+  #MY_SELECTION <- c('state', 'afl', 'graph', 'random')
+  SAVE_FILE=TRUE
+} else {
+  runtype=args[1]
+  target=args[2]
+  outputpath=args[3]
+  MY_SELECTION <- args[4:length(args)]
+  SAVE_FILE=TRUE
+}
+worst_cases <- list(waters=0, waters_int=0, tmr=405669, micro_longint=0)
+worst_case <- worst_cases[[target]]
+if (is.null(worst_case)) {
+  worst_case = 0
+}
+
+#MY_COLORS=c("green","blue","red", "orange", "pink", "black")
+MY_COLORS <- c("green", "blue", "red", "magenta", "orange", "cyan", "pink", "gray", "orange", "black", "yellow","brown")
+BENCHDIR=sprintf("~/code/FRET/LibAFL/fuzzers/FRET/benchmark/%s",runtype)
+BASENAMES=Filter(function(x) x!="" && substr(x,1,1)!='.',list.dirs(BENCHDIR,full.names=FALSE))
+PATTERNS="%s.[0-9]*$"
+#RIBBON='sd'
+#RIBBON='span'
+RIBBON='both'
+DRAW_WC = worst_case > 0
+LEGEND_POS="topright"
+#LEGEND_POS="bottomright"
+CONTINUE_LINE_TO_END=FALSE
+
+# https://www.r-bloggers.com/2013/04/how-to-change-the-alpha-value-of-colours-in-r/
+alpha <- function(col, alpha=1){
+  if(missing(col))
+    stop("Please provide a vector of colours.")
+  apply(sapply(col, col2rgb)/255, 2, 
+        function(x) 
+          rgb(x[1], x[2], x[3], alpha=alpha))  
+}
+
+# Trimm a list of data frames to common length
+trim_data <- function(input,len=NULL) {
+  if (is.null(len)) {
+    len <- min(sapply(input, function(v) dim(v)[1]))
+  }
+  return(lapply(input, function(d) slice_head(d,n=len)))
+}
+
+length_of_data <- function(input) {
+  min(sapply(input, function(v) dim(v)[1]))
+}
+
+# Takes a flat list
+trace2maxline <- function(tr) {
+  maxline = tr
+  for (var in seq_len(length(maxline))[2:length(maxline)]) {
+    #if (maxline[var]>1000000000) {
+    #  maxline[var]=maxline[var-1]
+    #} else {
+      maxline[var] = max(maxline[var],maxline[var-1]) 
+    #}
+  }
+  #plot(seq_len(length(maxline)),maxline,"l",xlab="Index",ylab="WOET")
+  return(maxline)
+}
+
+# Take a list of data frames, output same form but maxlines
+data2maxlines <- function(tr) {
+  min_length <- min(sapply(tr, function(v) dim(v)[1]))
+  maxline <- tr
+  for (var in seq_len(length(tr))) {
+    maxline[[var]][[1]]=trace2maxline(tr[[var]][[1]])
+  }
+  return(maxline)
+}
+# Take a multi-column data frame, output same form but maxlines
+frame2maxlines <- function(tr) {
+  for (var in seq_len(length(tr))) {
+    tr[[var]]=trace2maxline(tr[[var]])
+  }
+  return(tr)
+}
+
+trace2maxpoints <- function(tr) {
+  minval = tr[1,1]
+  collect = tr[1,]
+  for (i in seq_len(dim(tr)[1])) {
+    if (minval < tr[i,1]) {
+      collect = rbind(collect,tr[i,])
+      minval = tr[i,1]
+    }
+  }
+  tmp = tr[dim(tr)[1],]
+  tmp[1] = minval[1]
+  collect = rbind(collect,tmp)
+  return(collect)
+}
+
+sample_maxpoints <- function(tr,po) {
+  index = 1
+  collect=NULL
+  endpoint = dim(tr)[1]
+  for (p in po) {
+    if (p<=tr[1,2]) {
+      tmp = tr[index,]
+      tmp[2] = p
+      collect = rbind(collect, tmp)
+    } else if (p>=tr[endpoint,2]) {
+      tmp = tr[endpoint,]
+      tmp[2] = p
+      collect = rbind(collect, tmp)
+    } else {
+      for (i in seq(index,endpoint)-1) {
+        if (p >= tr[i,2] && p<tr[i+1,2]) {
+          tmp = tr[i,]
+          tmp[2] = p
+          collect = rbind(collect, tmp)
+          index = i
+          break
+        }
+      } 
+    }
+  }
+  return(collect)
+}
+
+#https://www.r-bloggers.com/2012/01/parallel-r-loops-for-windows-and-linux/
+all_runtypetables <- foreach (bn=BASENAMES) %do% {
+  runtypefiles <- list.files(file.path(BENCHDIR,bn),pattern=sprintf(PATTERNS,target),full.names = TRUE)
+  if (length(runtypefiles) > 0) {
+    runtypetables_reduced <- foreach(i=seq_len(length(runtypefiles))) %dopar% {
+      rtable = read.csv(runtypefiles[[i]], col.names=c(sprintf("%s%d",bn,i),sprintf("times%d",i)))
+      trace2maxpoints(rtable)
+    }
+    #runtypetables <- lapply(seq_len(length(runtypefiles)),
+    #    function(i)read.csv(runtypefiles[[i]], col.names=c(sprintf("%s%d",bn,i),sprintf("times%d",i))))
+    #runtypetables_reduced <- lapply(runtypetables, trace2maxpoints)
+    runtypetables_reduced
+    #all_runtypetables = c(all_runtypetables, list(runtypetables_reduced))
+  }
+}
+all_runtypetables = all_runtypetables[lapply(all_runtypetables, length) > 0]
+all_min_points = foreach(rtt=all_runtypetables,.combine = cbind) %do% {
+  bn = substr(names(rtt[[1]])[1],1,nchar(names(rtt[[1]])[1])-1)
+  ret = data.frame(min(unlist(lapply(rtt, function(v) v[dim(v)[1],2]))))
+  names(ret)[1] = bn
+  ret/(3600 * 1000)
+}
+all_max_points = foreach(rtt=all_runtypetables,.combine = cbind) %do% {
+  bn = substr(names(rtt[[1]])[1],1,nchar(names(rtt[[1]])[1])-1)
+  ret = data.frame(max(unlist(lapply(rtt, function(v) v[dim(v)[1],2]))))
+  names(ret)[1] = bn
+  ret/(3600 * 1000)
+}
+all_points = sort(unique(Reduce(c, lapply(all_runtypetables, function(v) Reduce(c, lapply(v, function(w) w[[2]]))))))
+all_maxlines <- foreach (rtt=all_runtypetables) %do% {
+  bn = substr(names(rtt[[1]])[1],1,nchar(names(rtt[[1]])[1])-1)
+  runtypetables_sampled = foreach(v=rtt) %dopar% {
+    sample_maxpoints(v, all_points)[1]
+  }
+  #runtypetables_sampled = lapply(rtt, function(v) sample_maxpoints(v, all_points)[1])
+  tmp_frame <- Reduce(cbind, runtypetables_sampled)
+  statframe <- data.frame(rowMeans(tmp_frame),apply(tmp_frame, 1, sd),apply(tmp_frame, 1, min),apply(tmp_frame, 1, max), apply(tmp_frame, 1, median))
+  names(statframe) <- c(bn, sprintf("%s_sd",bn), sprintf("%s_min",bn), sprintf("%s_max",bn), sprintf("%s_med",bn))
+  #statframe[sprintf("%s_times",bn)] = all_points
+  round(statframe)
+  #all_maxlines = c(all_maxlines, list(round(statframe))) 
+}
+one_frame<-data.frame(all_maxlines)
+one_frame[length(one_frame)+1] <- all_points/(3600 * 1000)
+names(one_frame)[length(one_frame)] <- 'time'
+
+typenames = names(one_frame)[which(names(one_frame) != 'time')]
+typenames = typenames[which(!endsWith(typenames, "_sd"))]
+typenames = typenames[which(!endsWith(typenames, "_med"))]
+ylow=min(one_frame[typenames])
+yhigh=max(one_frame[typenames],worst_case)
+typenames = typenames[which(!endsWith(typenames, "_min"))]
+typenames = typenames[which(!endsWith(typenames, "_max"))]
+
+ml2lines <- function(ml,lim) {
+  lines = NULL
+  last = 0
+  for (i in seq_len(dim(ml)[1])) {
+    if (!CONTINUE_LINE_TO_END && lim<ml[i,2]) {
+      break
+    }
+    lines = rbind(lines, cbind(X=last, Y=ml[i,1]))
+    lines = rbind(lines, cbind(X=ml[i,2], Y=ml[i,1]))
+    last = ml[i,2]
+  }
+  return(lines)
+}
+
+plotting <- function(selection, filename, MY_COLORS_) {
+# filter out names of iters and sd cols
+typenames = names(one_frame)[which(names(one_frame) != 'times')]
+typenames = typenames[which(!endsWith(typenames, "_sd"))]
+typenames = typenames[which(!endsWith(typenames, "_med"))]
+typenames = typenames[which(!endsWith(typenames, "_min"))]
+typenames = typenames[which(!endsWith(typenames, "_max"))]
+typenames = selection[which(selection %in% typenames)]
+if (length(typenames) == 0) {return()}
+
+h_ = 500
+w_ = h_*4/3
+
+if (SAVE_FILE) {png(file=sprintf("%s%s_%s.png",outputpath,target,filename), width=w_, height=h_)}
+par(mar=c(4,4,1,1))
+par(oma=c(0,0,0,0))
+
+plot(c(1,max(one_frame['time'])),c(ylow,yhigh), col='white', xlab="Time [h]", ylab="WORT [insn]", pch='.')
+
+for (t in seq_len(length(typenames))) {
+  #proj = one_frame[seq(1, dim(one_frame)[1], by=max(1, length(one_frame[[1]])/(10*w_))),]
+  #points(proj[c('iters',typenames[t])], col=MY_COLORS_[t], pch='.')
+  avglines = ml2lines(one_frame[c(typenames[t],'time')],all_max_points[typenames[t]])
+  #lines(avglines, col=MY_COLORS_[t])
+  medlines = ml2lines(one_frame[c(sprintf("%s_med",typenames[t]),'time')],all_max_points[typenames[t]])
+  lines(medlines, col=MY_COLORS_[t], lty='solid')
+  milines = NULL
+  malines = NULL
+  milines = ml2lines(one_frame[c(sprintf("%s_min",typenames[t]),'time')],all_max_points[typenames[t]])
+  malines = ml2lines(one_frame[c(sprintf("%s_max",typenames[t]),'time')],all_max_points[typenames[t]])
+  if (exists("RIBBON") && ( RIBBON=='max' )) {
+    #lines(milines, col=MY_COLORS_[t], lty='dashed')
+    lines(malines, col=MY_COLORS_[t], lty='dashed')
+    #points(proj[c('iters',sprintf("%s_min",typenames[t]))], col=MY_COLORS_[t], pch='.')
+    #points(proj[c('iters',sprintf("%s_max",typenames[t]))], col=MY_COLORS_[t], pch='.')
+  }
+  if (exists("RIBBON") && RIBBON != '') {
+  for (i in seq_len(dim(avglines)[1]-1)) {
+    if (RIBBON=='both') {
+      # draw boxes
+      x_l <- milines[i,][['X']]
+      x_r <- milines[i+1,][['X']]
+      y_l <- milines[i,][['Y']]
+      y_h <- malines[i,][['Y']]
+      rect(x_l, y_l, x_r, y_h, col=alpha(MY_COLORS_[t], alpha=0.1), lwd=0)
+    }
+    if (FALSE && RIBBON=='span') {
+      # draw boxes
+      x_l <- milines[i,][['X']]
+      x_r <- milines[i+1,][['X']]
+      y_l <- milines[i,][['Y']]
+      y_h <- malines[i,][['Y']]
+      rect(x_l, y_l, x_r, y_h, col=alpha(MY_COLORS_[t], alpha=0.1), lwd=0)
+    }
+    #if (FALSE && RIBBON=='both' || RIBBON=='sd') {
+    #  # draw sd
+    #  x_l <- avglines[i,][['X']]
+    #  x_r <- avglines[i+1,][['X']]
+    #  y_l <- avglines[i,][['Y']]-one_frame[ceiling(i/2),][[sprintf("%s_sd",typenames[t])]]
+    #  y_h <- avglines[i,][['Y']]+one_frame[ceiling(i/2),][[sprintf("%s_sd",typenames[t])]]
+    #  if (x_r != x_l) {
+    #    rect(x_l, y_l, x_r, y_h, col=alpha(MY_COLORS_[t], alpha=0.1), lwd=0) 
+    #  }
+    #}
+    #sd_ <- row[sprintf("%s_sd",typenames[t])][[1]]
+    #min_ <- row[sprintf("%s_min",typenames[t])][[1]]
+    #max_ <- row[sprintf("%s_max",typenames[t])][[1]]
+    #if (exists("RIBBON")) {
+    #  switch (RIBBON,
+    #    'sd' = arrows(x_, y_-sd_, x_, y_+sd_, length=0, angle=90, code=3, col=alpha(MY_COLORS_[t], alpha=0.03)),
+    #    'both' = arrows(x_, y_-sd_, x_, y_+sd_, length=0, angle=90, code=3, col=alpha(MY_COLORS_[t], alpha=0.05)),
+    #    'span' = #arrows(x_, min_, x_, max_, length=0, angle=90, code=3, col=alpha(MY_COLORS_[t], alpha=0.03))
+    #  )
+    #}
+    ##arrows(x_, y_-sd_, x_, y_+sd_, length=0.05, angle=90, code=3, col=alpha(MY_COLORS[t], alpha=0.1))
+  }
+  }
+}
+leglines=typenames
+if (DRAW_WC) {
+  lines(c(0,length(one_frame[[1]])),y=c(worst_case,worst_case), lty='dotted')
+  leglines=c(typenames, 'worst observed') 
+}
+legend(LEGEND_POS, legend=leglines,#"topleft"
+       col=c(MY_COLORS_[1:length(typenames)],"black"),
+       lty=c(rep("solid",length(typenames)),"dotted"))
+
+if (SAVE_FILE) {dev.off()}
+}
+stopCluster(cl)
+
+par(mar=c(3.8,3.8,0,0))
+par(oma=c(0,0,0,0))
+
+#RIBBON='both'
+#MY_SELECTION = c('state_int','generation100_int')
+#MY_SELECTION = c('state_int','frafl_int')
+
+if (exists("MY_SELECTION")) {
+  plotting(MY_SELECTION, 'custom', MY_COLORS[c(1,2)])
+} else {
+  # MY_SELECTION=c('state', 'afl', 'random', 'feedlongest', 'feedgeneration', 'feedgeneration10')
+  #MY_SELECTION=c('state_int', 'afl_int', 'random_int', 'feedlongest_int', 'feedgeneration_int', 'feedgeneration10_int')
+  #MY_SELECTION=c('state', 'frAFL', 'statenohash', 'feedgeneration10')
+  #MY_SELECTION=c('state_int', 'frAFL_int', 'statenohash_int', 'feedgeneration10_int')
+  MY_SELECTION=typenames
+  RIBBON='both'
+  for (i in seq_len(length(MY_SELECTION))) {
+    n <- MY_SELECTION[i]
+    plotting(c(n), n, c(MY_COLORS[i]))
+  }
+  RIBBON='max'
+  plotting(MY_SELECTION,'all', MY_COLORS)
+}
+
+for (t in seq_len(length(typenames))) {
+  li = one_frame[dim(one_frame)[1],]
+  pear = (li[[typenames[[t]]]]-li[[sprintf("%s_med",typenames[[t]])]])/li[[sprintf("%s_sd",typenames[[t]])]]
+  print(sprintf("%s pearson: %g",typenames[[t]],pear))
+}

fuzzers/FRET/benchmark/target_symbols.csv

@@ -0,0 +1,24 @@
+kernel,main_function,input_symbol,input_size,return_function
+mpeg2,mpeg2_main,mpeg2_oldorgframe,90112,mpeg2_return
+audiobeam,audiobeam_main,audiobeam_input,11520,audiobeam_return
+epic,epic_main,epic_image,4096,epic_return
+dijkstra,dijkstra_main,dijkstra_AdjMatrix,10000,dijkstra_return
+fft,fft_main,fft_twidtable,2046,fft_return
+bsort,bsort_main,bsort_Array,400,bsort_return
+insertsort,insertsort_main,insertsort_a,400,insertsort_return
+g723_enc,g723_enc_main,g723_enc_INPUT,1024,g723_enc_return
+rijndael_dec,rijndael_dec_main,rijndael_dec_data,32768,rijndael_dec_return
+rijndael_enc,rijndael_enc_main,rijndael_enc_data,31369,rijndael_enc_return
+huff_dec,huff_dec_main,huff_dec_encoded,419,huff_dec_return
+huff_enc,huff_enc_main,huff_enc_plaintext,600,huff_enc_return
+gsm_enc,gsm_enc_main,gsm_enc_pcmdata,6400,gsm_enc_return
+tmr,main,FUZZ_INPUT,32,trigger_Qemu_break
+tacle_rtos,prvStage0,FUZZ_INPUT,604,trigger_Qemu_break
+lift,main_lift,FUZZ_INPUT,100,trigger_Qemu_break
+waters,main_waters,FUZZ_INPUT,4096,trigger_Qemu_break
+watersv2,main_waters,FUZZ_INPUT,4096,trigger_Qemu_break
+waters_int,main_waters,FUZZ_INPUT,4096,trigger_Qemu_break
+watersv2_int,main_waters,FUZZ_INPUT,4096,trigger_Qemu_break
+micro_branchless,main_branchless,FUZZ_INPUT,4,trigger_Qemu_break
+micro_int,main_int,FUZZ_INPUT,16,trigger_Qemu_break
+micro_longint,main_micro_longint,FUZZ_INPUT,16,trigger_Qemu_break

fuzzers/FRET/example/build.sh

@@ -0,0 +1,2 @@
+#!/bin/sh
+arm-none-eabi-gcc -ggdb -ffreestanding -nostartfiles -lgcc -T mps2_m3.ld -mcpu=cortex-m3 main.c startup.c -o example.elf

fuzzers/FRET/example/main.c

@@ -0,0 +1,38 @@
+int BREAKPOINT() {
+  for (;;)
+  {
+  }
+}
+
+int LLVMFuzzerTestOneInput(unsigned int* Data, unsigned int Size) {
+  //if (Data[3] == 0) {while(1){}}  // cause a timeout
+  for (int i=0; i<Size; i++) {
+    // if (Data[i] > 0xFFd0 && Data[i] < 0xFFFF) {return 1;}    // cause qemu to crash
+    for (int j=i+1; j<Size; j++) {
+      if (Data[j] == 0) {continue;}
+      if (Data[j]>Data[i]) {
+        int tmp = Data[i];
+        Data[i]=Data[j];
+        Data[j]=tmp;
+        if (Data[i] <= 100) {j--;}
+      }
+    }
+  }
+  return BREAKPOINT();
+}
+    unsigned int FUZZ_INPUT[] = {
+    101,201,700,230,860,
+    234,980,200,340,678,
+    230,134,900,236,900,
+    123,800,123,658,607,
+    246,804,567,568,207,
+    407,246,678,457,892,
+    834,456,878,246,699,
+    854,234,844,290,125,
+    324,560,852,928,910,
+    790,853,345,234,586,
+    };
+
+int main() {
+  LLVMFuzzerTestOneInput(FUZZ_INPUT, 50);
+}

fuzzers/FRET/example/mps2_m3.ld

@@ -0,0 +1,143 @@
+/*
+ * FreeRTOS V202112.00
+ * Copyright (C) 2020 Amazon.com, Inc. or its affiliates.  All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy of
+ * this software and associated documentation files (the "Software"), to deal in
+ * the Software without restriction, including without limitation the rights to
+ * use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of
+ * the Software, and to permit persons to whom the Software is furnished to do so,
+ * subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
+ * FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
+ * COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+ * IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * https://www.FreeRTOS.org
+ * https://github.com/FreeRTOS
+ *
+ */
+
+MEMORY
+{
+    RAM (xrw)  : ORIGIN = 0x00000000, LENGTH = 4M
+    /* Originally                                    */
+    /* FLASH (xr) : ORIGIN = 0x00000000, LENGTH = 4M */
+    /* RAM (xrw)  : ORIGIN = 0x20000000, LENGTH = 4M */
+}
+ENTRY(Reset_Handler)
+
+_Min_Heap_Size = 0x300000 ;        /* Required amount of heap. */
+_Min_Stack_Size = 0x4000 ;       /* Required amount of stack. */
+M_VECTOR_RAM_SIZE = (16 + 48) * 4;
+_estack = ORIGIN(RAM) + LENGTH(RAM);
+
+SECTIONS
+{
+
+    .isr_vector :
+    {
+        __vector_table = .;
+        KEEP(*(.isr_vector))
+        . = ALIGN(4);
+    } > RAM /* FLASH */
+
+    .text :
+    {
+        . = ALIGN(4);
+        *(.text*)
+        KEEP (*(.init))
+        KEEP (*(.fini))
+        KEEP(*(.eh_frame))
+        *(.rodata*)
+        . = ALIGN(4);
+        _etext = .;
+    } > RAM /* FLASH */
+
+    .ARM.extab :
+    {
+        . = ALIGN(4);
+        *(.ARM.extab* .gnu.linkonce.armextab.*)
+        . = ALIGN(4);
+    } >RAM /* FLASH */
+
+    .ARM :
+    {
+        . = ALIGN(4);
+        __exidx_start = .;
+        *(.ARM.exidx* .gnu.linkonce.armexidx.*)
+        __exidx_end = .;
+        . = ALIGN(4);
+    } >RAM /* FLASH */
+
+    .interrupts_ram :
+    {
+        . = ALIGN(4);
+        __VECTOR_RAM__ = .;
+        __interrupts_ram_start__ = .;
+        . += M_VECTOR_RAM_SIZE;
+        . = ALIGN(4);
+        __interrupts_ram_end = .;
+    } > RAM
+
+    _sidata = LOADADDR(.data);
+
+    .data : /* AT ( _sidata ) */
+    {
+        . = ALIGN(4);
+        _sdata = .;
+        *(.data*)
+        . = ALIGN(4);
+        _edata = .;
+    } > RAM /* RAM AT > FLASH */
+
+    .uninitialized (NOLOAD):
+    {
+        . = ALIGN(32);
+        __uninitialized_start = .;
+        *(.uninitialized)
+        KEEP(*(.keep.uninitialized))
+        . = ALIGN(32);
+        __uninitialized_end = .;
+    } > RAM
+
+    .bss :
+    {
+        . = ALIGN(4);
+        _sbss = .;
+        __bss_start__ = _sbss;
+        *(.bss*)
+        *(COMMON)
+        . = ALIGN(4);
+        _ebss = .;
+        __bss_end__ = _ebss;
+    } >RAM
+
+    .heap :
+    {
+        . = ALIGN(8);
+        PROVIDE ( end = . );
+        PROVIDE ( _end = . );
+        _heap_bottom = .;
+        . = . + _Min_Heap_Size;
+        _heap_top = .;
+        . = . + _Min_Stack_Size;
+        . = ALIGN(8);
+       } >RAM
+
+   /* Set stack top to end of RAM, and stack limit move down by
+    * size of stack_dummy section */
+   __StackTop = ORIGIN(RAM) + LENGTH(RAM);
+   __StackLimit = __StackTop - _Min_Stack_Size;
+   PROVIDE(__stack = __StackTop);
+     
+  /* Check if data + heap + stack exceeds RAM limit */
+  ASSERT(__StackLimit >= _heap_top, "region RAM overflowed with stack")
+}
+

fuzzers/FRET/example/startup.c

@@ -0,0 +1,114 @@
+/*
+ * FreeRTOS V202112.00
+ * Copyright (C) 2020 Amazon.com, Inc. or its affiliates.  All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy of
+ * this software and associated documentation files (the "Software"), to deal in
+ * the Software without restriction, including without limitation the rights to
+ * use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of
+ * the Software, and to permit persons to whom the Software is furnished to do so,
+ * subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
+ * FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
+ * COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+ * IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * https://www.FreeRTOS.org
+ * https://github.com/FreeRTOS
+ *
+ */
+
+typedef unsigned int uint32_t;
+
+extern int main();
+
+extern uint32_t _estack, _sidata, _sdata, _edata, _sbss, _ebss;
+
+/* Prevent optimization so gcc does not replace code with memcpy */
+__attribute__( ( optimize( "O0" ) ) )
+__attribute__( ( naked ) )
+void Reset_Handler( void )
+{
+    /* set stack pointer */
+    __asm volatile ( "ldr r0, =_estack" );
+    __asm volatile ( "mov sp, r0" );
+
+    /* copy .data section from flash to RAM */
+    // Not needed for this example, see linker script
+    // for( uint32_t * src = &_sidata, * dest = &_sdata; dest < &_edata; )
+    // {
+    //     *dest++ = *src++;
+    // }
+
+    /* zero out .bss section */
+    for( uint32_t * dest = &_sbss; dest < &_ebss; )
+    {
+        *dest++ = 0;
+    }
+
+    /* jump to board initialisation */
+    void _start( void );
+    _start();
+}
+
+const uint32_t * isr_vector[] __attribute__( ( section( ".isr_vector" ) ) ) =
+{
+    ( uint32_t * ) &_estack,
+    ( uint32_t * ) &Reset_Handler,       /* Reset                -15 */
+    0,     /* NMI_Handler          -14 */
+    0,    /* HardFault_Handler    -13 */
+    0,    /* MemManage_Handler    -12 */
+    0,    /* BusFault_Handler     -11 */
+    0,    /* UsageFault_Handler   -10 */
+    0,                                   /* reserved */
+    0,                                   /* reserved */
+    0,                                   /* reserved */
+    0,                                   /* reserved   -6 */
+    0,     /* SVC_Handler              -5 */
+    0,    /* DebugMon_Handler         -4 */
+    0,                                   /* reserved */
+    0,  /* PendSV handler    -2 */
+    0, /* SysTick_Handler   -1 */
+    0,                                   /* uart0 receive 0 */
+    0,                                   /* uart0 transmit */
+    0,                                   /* uart1 receive */
+    0,                                   /* uart1 transmit */
+    0,                                   /* uart 2 receive */
+    0,                                   /* uart 2 transmit */
+    0,                                   /* GPIO 0 combined interrupt */
+    0,                                   /* GPIO 2 combined interrupt */
+    0,                                   /* Timer 0 */
+    0,                                   /* Timer 1 */
+    0,                                   /* Dial Timer */
+    0,                                   /* SPI0 SPI1 */
+    0,                                   /* uart overflow 1, 2,3 */
+    0,                                   /* Ethernet   13 */
+};
+
+__attribute__( ( naked ) ) void exit(__attribute__((unused)) int status )
+{
+    /* Force qemu to exit using ARM Semihosting */
+    __asm volatile (
+        "mov r1, r0\n"
+        "cmp r1, #0\n"
+        "bne .notclean\n"
+        "ldr r1, =0x20026\n" /* ADP_Stopped_ApplicationExit, a clean exit */
+        ".notclean:\n"
+        "movs r0, #0x18\n"   /* SYS_EXIT */
+        "bkpt 0xab\n"
+        "end: b end\n"
+        );
+}
+
+void _start( void )
+{
+    main( );
+    exit( 0 );
+}
+

fuzzers/FRET/fuzzer.sh

@@ -0,0 +1,25 @@
+#!/usr/bin/env bash
+parent_path=$( cd "$(dirname "${BASH_SOURCE[0]}")" ; pwd -P )
+cd "$parent_path"
+
+[ -n "$1" -a "$1" != "+" -a -z "$KERNEL" ] && export KERNEL="$1"
+[ -n "$2" -a "$2" != "+" -a -z "$FUZZ_MAIN" ]   && export FUZZ_MAIN="$2"
+[ -n "$3" -a "$3" != "+" -a -z "$FUZZ_INPUT" ]  && export FUZZ_INPUT="$3"
+[ -n "$4" -a "$4" != "+" -a -z "$FUZZ_INPUT_LEN" ]  && export FUZZ_INPUT_LEN="$4"
+[ -n "$5" -a "$5" != "+" -a -z "$BREAKPOINT" ]  && export BREAKPOINT="$5"
+[ -n "$6" -a "$6" != "+" -a -z "$FUZZ_ITERS" ]  && export FUZZ_ITERS="$6"
+[ -n "$7" -a "$7" != "+" -a -z "$TIME_DUMP" ]  && export TIME_DUMP="$7"
+[ -n "$8" -a "$8" != "+" -a -z "$CASE_DUMP" ]  && export CASE_DUMP="$8"
+[ -n "$9" -a "$9" != "+" -a -z "$DO_SHOWMAP" ]  && export DO_SHOWMAP="$9"
+[ -n "${10}" -a "${10}" != "+" -a -z "$SHOWMAP_TEXTINPUT" ]  && export SHOWMAP_TEXTINPUT="${10}"
+[ -n "${11}" -a "${11}" != "+" -a -z "$TRACE_DUMP" ]  && export TRACE_DUMP="${11}"
+
+[ -z "$FUZZER" ] && export FUZZER=target/debug/fret
+set +e
+$FUZZER -icount shift=4,align=off,sleep=off -machine mps2-an385 -monitor null -kernel $KERNEL -serial null -nographic -S -semihosting --semihosting-config enable=on,target=native -snapshot -drive if=none,format=qcow2,file=dummy.qcow2
+if [ "$exitcode" = "101" ]
+then
+    exit 101
+else
+    exit 0
+fi

fuzzers/FRET/src/clock.rs

@@ -0,0 +1,344 @@
+use hashbrown::{hash_map::Entry, HashMap};
+use libafl::{
+    bolts::{
+        current_nanos,
+        rands::StdRand,
+        tuples::{tuple_list},
+    },
+    executors::{ExitKind},
+    fuzzer::{StdFuzzer},
+    inputs::{BytesInput, HasTargetBytes},
+    observers::{Observer,VariableMapObserver},
+    state::{StdState, HasNamedMetadata},
+    Error,
+    observers::ObserversTuple, prelude::UsesInput, impl_serdeany,
+};
+use serde::{Deserialize, Serialize};
+use std::{cell::UnsafeCell, cmp::max, env, fs::OpenOptions, io::Write, time::Instant};
+use libafl::bolts::tuples::Named;
+
+use libafl_qemu::{
+    emu,
+    emu::Emulator,
+    executor::QemuExecutor,
+    helper::{QemuHelper, QemuHelperTuple, QemuInstrumentationFilter},
+};
+use libafl::events::EventFirer;
+use libafl::state::HasClientPerfMonitor;
+use libafl::inputs::Input;
+use libafl::feedbacks::Feedback;
+use libafl::SerdeAny;
+use libafl::state::HasMetadata;
+use libafl::corpus::testcase::Testcase;
+use core::{fmt::Debug, time::Duration};
+// use libafl::feedbacks::FeedbackState;
+// use libafl::state::HasFeedbackStates;
+use libafl::bolts::tuples::MatchName;
+use std::time::{SystemTime, UNIX_EPOCH};
+
+pub static mut FUZZ_START_TIMESTAMP : SystemTime = UNIX_EPOCH;
+
+//========== Metadata
+#[derive(Debug, SerdeAny, Serialize, Deserialize)]
+pub struct QemuIcountMetadata {
+    runtime: u64,
+}
+
+/// Metadata for [`QemuClockIncreaseFeedback`]
+#[derive(Debug, Serialize, Deserialize, SerdeAny)]
+pub struct MaxIcountMetadata {
+    pub max_icount_seen: u64,
+    pub name: String,
+}
+
+// impl FeedbackState for MaxIcountMetadata
+// {
+//     fn reset(&mut self) -> Result<(), Error> {
+//         self.max_icount_seen = 0;
+//         Ok(())
+//     }
+// }
+
+impl Named for MaxIcountMetadata
+{
+    #[inline]
+    fn name(&self) -> &str {
+        self.name.as_str()
+    }
+}
+
+impl MaxIcountMetadata
+{
+    /// Create new `MaxIcountMetadata`
+    #[must_use]
+    pub fn new(name: &'static str) -> Self {
+        Self {
+            max_icount_seen: 0,
+            name: name.to_string(),
+        }
+    }
+}
+
+impl Default for MaxIcountMetadata {
+    fn default() -> Self {
+        Self::new("MaxClock")
+    }
+}
+
+/// A piece of metadata tracking all icounts
+#[derive(Debug, SerdeAny, Serialize, Deserialize)]
+pub struct IcHist (pub Vec<(u64, u128)>, pub (u64,u128));
+
+//========== Observer
+
+/// A simple observer, just overlooking the runtime of the target.
+#[derive(Serialize, Deserialize, Debug, Clone)]
+pub struct QemuClockObserver {
+    name: String,
+    start_tick: u64,
+    end_tick: u64,
+}
+
+impl QemuClockObserver {
+    /// Creates a new [`QemuClockObserver`] with the given name.
+    #[must_use]
+    pub fn new(name: &'static str) -> Self {
+        Self {
+            name: name.to_string(),
+            start_tick: 0,
+            end_tick: 0,
+        }
+    }
+
+    /// Gets the runtime for the last execution of this target.
+    #[must_use]
+    pub fn last_runtime(&self) -> u64 {
+        self.end_tick - self.start_tick
+    }
+}
+
+impl<S> Observer<S> for QemuClockObserver
+where
+    S: UsesInput + HasMetadata,
+{
+    fn pre_exec(&mut self, _state: &mut S, _input: &S::Input) -> Result<(), Error> {
+        // Only remember the pre-run ticks if presistent mode ist used
+        #[cfg(not(feature = "snapshot_restore"))]
+        unsafe {
+            self.start_tick=emu::icount_get_raw();
+            self.end_tick=self.start_tick;
+        }
+        // unsafe {
+        //     println!("clock pre {}",emu::icount_get_raw());
+        // }
+        Ok(())
+    }
+
+    fn post_exec(&mut self, _state: &mut S, _input: &S::Input, _exit_kind: &ExitKind) -> Result<(), Error> {
+        unsafe { self.end_tick = emu::icount_get_raw() };
+        // println!("clock post {}", self.end_tick);
+        // println!("Number of Ticks: {} <- {} {}",self.end_tick - self.start_tick, self.end_tick, self.start_tick);
+        let metadata =_state.metadata_mut();
+        let hist = metadata.get_mut::<IcHist>();
+        let timestamp = SystemTime::now().duration_since(unsafe {FUZZ_START_TIMESTAMP}).unwrap().as_millis();
+        match hist {
+            None => {
+                metadata.insert(IcHist(vec![(self.end_tick - self.start_tick, timestamp)],
+                (self.end_tick - self.start_tick, timestamp)));
+            }
+            Some(v) => {
+                v.0.push((self.end_tick - self.start_tick, timestamp));
+                if (v.1.0 < self.end_tick-self.start_tick) {
+                    v.1 = (self.end_tick - self.start_tick, timestamp);
+                }
+                if v.0.len() >= 100 {
+                    if let Ok(td) = env::var("TIME_DUMP") {
+                        let mut file = OpenOptions::new()
+                            .read(true)
+                            .write(true)
+                            .create(true)
+                            .append(true)
+                            .open(td).expect("Could not open timedump");
+                        let newv : Vec<(u64, u128)> = Vec::with_capacity(100);
+                        for i in std::mem::replace(&mut v.0, newv).into_iter() {
+                            writeln!(file, "{},{}", i.0, i.1).expect("Write to dump failed");
+                        }
+                    } else {
+                        // If we don't write out values we don't need to remember them at all
+                        v.0.clear();
+                    }
+                }
+            }
+        }
+        Ok(())
+    }
+}
+
+impl Named for QemuClockObserver {
+    #[inline]
+    fn name(&self) -> &str {
+        &self.name
+    }
+}
+
+impl Default for QemuClockObserver {
+    fn default() -> Self {
+        Self {
+            name: String::from("clock"),
+            start_tick: 0,
+            end_tick: 0,
+        }
+    }
+}
+
+//========== Feedback
+/// Nop feedback that annotates execution time in the new testcase, if any
+/// for this Feedback, the testcase is never interesting (use with an OR).
+/// It decides, if the given [`QemuClockObserver`] value of a run is interesting.
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct ClockTimeFeedback {
+    exec_time: Option<Duration>,
+    name: String,
+}
+
+impl<S> Feedback<S> for ClockTimeFeedback
+where
+    S: UsesInput + HasClientPerfMonitor + HasMetadata,
+{
+    #[allow(clippy::wrong_self_convention)]
+    fn is_interesting<EM, OT>(
+        &mut self,
+        _state: &mut S,
+        _manager: &mut EM,
+        _input: &S::Input,
+        observers: &OT,
+        _exit_kind: &ExitKind,
+    ) -> Result<bool, Error>
+    where
+        EM: EventFirer<State = S>,
+        OT: ObserversTuple<S>,
+    {
+        // TODO Replace with match_name_type when stable
+        let observer = observers.match_name::<QemuClockObserver>(self.name()).unwrap();
+        self.exec_time = Some(Duration::from_nanos(observer.last_runtime() << 4)); // Assume a somewhat realistic multiplier of clock, it does not matter
+        Ok(false)
+    }
+
+    /// Append to the testcase the generated metadata in case of a new corpus item
+    #[inline]
+    fn append_metadata(
+        &mut self,
+        _state: &mut S,
+        testcase: &mut Testcase<S::Input>,
+    ) -> Result<(), Error> {
+        *testcase.exec_time_mut() = self.exec_time;
+        self.exec_time = None;
+        Ok(())
+    }
+
+    /// Discard the stored metadata in case that the testcase is not added to the corpus
+    #[inline]
+    fn discard_metadata(&mut self, _state: &mut S, _input: &S::Input) -> Result<(), Error> {
+        self.exec_time = None;
+        Ok(())
+    }
+}
+
+impl Named for ClockTimeFeedback {
+    #[inline]
+    fn name(&self) -> &str {
+        self.name.as_str()
+    }
+}
+
+impl ClockTimeFeedback {
+    /// Creates a new [`ClockFeedback`], deciding if the value of a [`QemuClockObserver`] with the given `name` of a run is interesting.
+    #[must_use]
+    pub fn new(name: &'static str) -> Self {
+        Self {
+            exec_time: None,
+            name: name.to_string(),
+        }
+    }
+
+    /// Creates a new [`ClockFeedback`], deciding if the given [`QemuClockObserver`] value of a run is interesting.
+    #[must_use]
+    pub fn new_with_observer(observer: &QemuClockObserver) -> Self {
+        Self {
+            exec_time: None,
+            name: observer.name().to_string(),
+        }
+    }
+}
+
+/// A [`Feedback`] rewarding increasing the execution cycles on Qemu.
+#[derive(Debug)]
+pub struct QemuClockIncreaseFeedback {
+    name: String,
+}
+
+impl<S> Feedback<S> for QemuClockIncreaseFeedback
+where
+    S: UsesInput + HasNamedMetadata + HasClientPerfMonitor + Debug,
+{
+    fn is_interesting<EM, OT>(
+        &mut self,
+        state: &mut S,
+        _manager: &mut EM,
+        _input: &S::Input,
+        _observers: &OT,
+        _exit_kind: &ExitKind,
+    ) -> Result<bool, Error>
+    where
+        EM: EventFirer<State = S>,
+        OT: ObserversTuple<S>,
+    {
+        let observer = _observers.match_name::<QemuClockObserver>("clock")
+            .expect("QemuClockObserver not found");
+        let clock_state = state
+            .named_metadata_mut()
+            .get_mut::<MaxIcountMetadata>(&self.name)
+            .unwrap();
+        if observer.last_runtime() > clock_state.max_icount_seen {
+            // println!("Clock improving {}",observer.last_runtime());
+            clock_state.max_icount_seen = observer.last_runtime();
+            return Ok(true);
+        }
+        Ok(false)
+    }
+
+    /// Append to the testcase the generated metadata in case of a new corpus item
+    #[inline]
+    fn append_metadata(&mut self, _state: &mut S, testcase: &mut Testcase<S::Input>) -> Result<(), Error> {
+        // testcase.metadata_mut().insert(QemuIcountMetadata{runtime: self.last_runtime});
+        Ok(())
+    }
+
+    /// Discard the stored metadata in case that the testcase is not added to the corpus
+    #[inline]
+    fn discard_metadata(&mut self, _state: &mut S, _input: &S::Input) -> Result<(), Error> {
+        Ok(())
+    }
+
+}
+
+impl Named for QemuClockIncreaseFeedback {
+    #[inline]
+    fn name(&self) -> &str {
+        &self.name
+    }
+}
+
+impl QemuClockIncreaseFeedback {
+    /// Creates a new [`HitFeedback`]
+    #[must_use]
+    pub fn new(name: &'static str) -> Self {
+        Self {name: String::from(name)}
+    }
+}
+
+impl Default for QemuClockIncreaseFeedback {
+    fn default() -> Self {
+        Self::new("MaxClock")
+    }
+}

fuzzers/FRET/src/fuzzer.rs

@@ -0,0 +1,715 @@
+//! A fuzzer using qemu in systemmode for binary-only coverage of kernels
+//!
+use core::time::Duration;
+use std::{env, path::PathBuf, process::{self, abort}, io::{Read, Write}, fs::{self, OpenOptions}, cmp::{min, max}, mem::transmute_copy, collections::btree_map::Range};
+
+use libafl::{
+    bolts::{
+        core_affinity::Cores,
+        current_nanos,
+        launcher::Launcher,
+        rands::StdRand,
+        shmem::{ShMemProvider, StdShMemProvider},
+        tuples::tuple_list,
+        AsSlice,
+    },
+    corpus::{Corpus, InMemoryCorpus, OnDiskCorpus},
+    events::EventConfig,
+    executors::{ExitKind, TimeoutExecutor},
+    feedback_or,
+    feedback_or_fast,
+    feedbacks::{CrashFeedback, MaxMapFeedback, TimeoutFeedback},
+    fuzzer::{Fuzzer, StdFuzzer},
+    inputs::{BytesInput, HasTargetBytes},
+    monitors::MultiMonitor,
+    observers::{VariableMapObserver},
+    schedulers::{IndexesLenTimeMinimizerScheduler, QueueScheduler},
+    state::{HasCorpus, StdState, HasMetadata, HasNamedMetadata},
+    Error,
+    prelude::{SimpleMonitor, SimpleEventManager, AsMutSlice, RandBytesGenerator, Generator, SimpleRestartingEventManager, HasBytesVec, minimizer::TopRatedsMetadata, havoc_mutations, StdScheduledMutator, HitcountsMapObserver}, Evaluator, stages::StdMutationalStage,
+};
+use libafl_qemu::{
+    edges, edges::QemuEdgeCoverageHelper, elf::EasyElf, emu::Emulator, GuestPhysAddr, QemuExecutor,
+    QemuHooks, Regs, QemuInstrumentationFilter, GuestAddr,
+    emu::libafl_qemu_set_native_breakpoint, emu::libafl_qemu_remove_native_breakpoint,
+};
+use rand::{SeedableRng, StdRng, Rng};
+use crate::{
+    clock::{QemuClockObserver, ClockTimeFeedback, QemuClockIncreaseFeedback, IcHist, FUZZ_START_TIMESTAMP},
+    qemustate::QemuStateRestoreHelper,
+    systemstate::{helpers::QemuSystemStateHelper, observers::QemuSystemStateObserver, feedbacks::{DumpSystraceFeedback, NovelSystemStateFeedback}, graph::{SysMapFeedback, SysGraphFeedbackState, GraphMaximizerCorpusScheduler}, schedulers::{LongestTraceScheduler, GenerationScheduler}}, worst::{TimeMaximizerCorpusScheduler, ExecTimeIncFeedback, TimeStateMaximizerCorpusScheduler, AlwaysTrueFeedback},
+    mutational::MyStateStage,
+    mutational::{MINIMUM_INTER_ARRIVAL_TIME}, 
+};
+use std::time::{SystemTime, UNIX_EPOCH};
+
+pub static mut RNG_SEED: u64 = 1;
+
+pub static mut LIMIT : u32 = u32::MAX;
+
+pub const MAX_NUM_INTERRUPT: usize = 32;
+pub const DO_NUM_INTERRUPT: usize = 32;
+pub static mut MAX_INPUT_SIZE: usize = 32;
+/// Read ELF program headers to resolve physical load addresses.
+fn virt2phys(vaddr: GuestPhysAddr, tab: &EasyElf) -> GuestPhysAddr {
+    let ret;
+    for i in &tab.goblin().program_headers {
+        if i.vm_range().contains(&vaddr.try_into().unwrap()) {
+            ret = vaddr - TryInto::<GuestPhysAddr>::try_into(i.p_vaddr).unwrap()
+                + TryInto::<GuestPhysAddr>::try_into(i.p_paddr).unwrap();
+            return ret - (ret % 2);
+        }
+    }
+    return vaddr;
+}
+
+extern "C" {
+    static mut libafl_interrupt_offsets : [u32; 32];
+    static mut libafl_num_interrupts : usize;
+}
+
+pub fn fuzz() {
+    unsafe {FUZZ_START_TIMESTAMP = SystemTime::now();}
+    let mut starttime = std::time::Instant::now();
+    if let Ok(s) = env::var("FUZZ_SIZE") {
+        str::parse::<usize>(&s).expect("FUZZ_SIZE was not a number");
+    };
+    // Hardcoded parameters
+    let timeout = Duration::from_secs(10);
+    let broker_port = 1337;
+    let cores = Cores::from_cmdline("1").unwrap();
+    let corpus_dirs = [PathBuf::from("./corpus")];
+    let objective_dir = PathBuf::from("./crashes");
+
+    let mut elf_buffer = Vec::new();
+    let elf = EasyElf::from_file(
+        env::var("KERNEL").expect("KERNEL env not set"),
+        &mut elf_buffer,
+    )
+    .unwrap();
+
+    // the main address where the fuzzer starts
+    // if this is set for freeRTOS it has an influence on where the data will have to be written,
+    // since the startup routine copies the data segemnt to it's virtual address
+    let main_addr = elf
+        .resolve_symbol(&env::var("FUZZ_MAIN").unwrap_or_else(|_| "FUZZ_MAIN".to_owned()), 0);
+    if let Some(main_addr) = main_addr {
+        println!("main address = {:#x}", main_addr);
+    }
+
+    let input_addr = elf
+        .resolve_symbol(
+            &env::var("FUZZ_INPUT").unwrap_or_else(|_| "FUZZ_INPUT".to_owned()),
+            0,
+        )
+        .expect("Symbol or env FUZZ_INPUT not found") as GuestPhysAddr;
+    let input_addr = virt2phys(input_addr,&elf) as GuestPhysAddr;
+    println!("FUZZ_INPUT @ {:#x}", input_addr);
+
+    let test_length_ptr = elf
+        .resolve_symbol("FUZZ_LENGTH", 0).map(|x| x as GuestPhysAddr);
+    let test_length_ptr = Option::map_or(test_length_ptr, None, |x| Some(virt2phys(x,&elf)));
+
+    let input_counter_ptr = elf
+        .resolve_symbol(&env::var("FUZZ_POINTER").unwrap_or_else(|_| "FUZZ_POINTER".to_owned()), 0)
+        .map(|x| x as GuestPhysAddr);
+    let input_counter_ptr = Option::map_or(input_counter_ptr, None, |x| Some(virt2phys(x,&elf)));
+
+    #[cfg(feature = "systemstate")]
+    let curr_tcb_pointer = elf              // loads to the address specified in elf, without respecting program headers
+        .resolve_symbol("pxCurrentTCB", 0)
+        .expect("Symbol pxCurrentTCBC not found");
+    // let curr_tcb_pointer = virt2phys(curr_tcb_pointer,&elf);
+    #[cfg(feature = "systemstate")]
+    println!("TCB pointer at {:#x}", curr_tcb_pointer);
+    #[cfg(feature = "systemstate")]
+    let task_queue_addr = elf
+        .resolve_symbol("pxReadyTasksLists", 0)
+        .expect("Symbol pxReadyTasksLists not found");
+    // let task_queue_addr = virt2phys(task_queue_addr,&elf.goblin());
+    #[cfg(feature = "systemstate")]
+    println!("Task Queue at {:#x}", task_queue_addr);
+    #[cfg(feature = "systemstate")]
+    let svh = elf
+        .resolve_symbol("xPortPendSVHandler", 0)
+        .expect("Symbol xPortPendSVHandler not found");
+    // let svh=virt2phys(svh, &elf);
+    // let svh = elf
+    //     .resolve_symbol("vPortEnterCritical", 0)
+    //     .expect("Symbol vPortEnterCritical not found");
+    #[cfg(feature = "systemstate")]
+    let app_start = elf
+        .resolve_symbol("__APP_CODE_START__", 0)
+        .expect("Symbol __APP_CODE_START__ not found");
+    #[cfg(feature = "systemstate")]
+    let app_end = elf
+        .resolve_symbol("__APP_CODE_END__", 0)
+        .expect("Symbol __APP_CODE_END__ not found");
+    #[cfg(feature = "systemstate")]
+    let app_range = app_start..app_end;
+    #[cfg(feature = "systemstate")]
+    dbg!(app_range.clone());
+
+    let breakpoint = elf
+        .resolve_symbol(
+            &env::var("BREAKPOINT").unwrap_or_else(|_| "BREAKPOINT".to_owned()),
+            0,
+        )
+        .expect("Symbol or env BREAKPOINT not found");
+    println!("Breakpoint address = {:#x}", breakpoint);
+    unsafe {
+        libafl_num_interrupts = 0;
+    }
+
+    if let Ok(input_len) = env::var("FUZZ_INPUT_LEN") {
+        unsafe {MAX_INPUT_SIZE = str::parse::<usize>(&input_len).expect("FUZZ_INPUT_LEN was not a number");}
+    }
+    unsafe {dbg!(MAX_INPUT_SIZE);}
+
+    if let Ok(seed) = env::var("SEED_RANDOM") {
+        unsafe {RNG_SEED = str::parse::<u64>(&seed).expect("SEED_RANDOM must be an integer.");}
+    }
+
+    let mut run_client = |state: Option<_>, mut mgr, _core_id| {
+        // Initialize QEMU
+        let args: Vec<String> = env::args().collect();
+        let env: Vec<(String, String)> = env::vars().collect();
+        let emu = Emulator::new(&args, &env);
+
+        if let Some(main_addr) = main_addr {
+            unsafe {
+                libafl_qemu_set_native_breakpoint(main_addr);
+                emu.run();
+                libafl_qemu_remove_native_breakpoint(main_addr);
+            }
+        }
+
+        unsafe { libafl_qemu_set_native_breakpoint(breakpoint); }// BREAKPOINT 
+
+        // The wrapped harness function, calling out to the LLVM-style harness
+        let mut harness = |input: &BytesInput| {
+            let target = input.target_bytes();
+            let mut buf = target.as_slice();
+            let mut len = buf.len();
+            unsafe {
+                #[cfg(feature = "fuzz_int")]
+                {
+                    let mut start_tick : u32 = 0;
+                    for i in 0..DO_NUM_INTERRUPT {
+                        let mut t : [u8; 4] = [0,0,0,0];
+                        if len > (i+1)*4 {
+                            for j in 0 as usize..4 as usize {
+                                t[j]=buf[i*4+j];
+                            }
+                            if i == 0 || true {
+                                unsafe {start_tick = u32::from_le_bytes(t) % LIMIT;}
+                            } else {
+                                start_tick = u32::saturating_add(start_tick,max(MINIMUM_INTER_ARRIVAL_TIME,u32::from_le_bytes(t)));
+                            }
+                            libafl_interrupt_offsets[i] = start_tick;
+                            libafl_num_interrupts = i+1;
+                        }
+                    }
+                    if buf.len() > libafl_num_interrupts*4 {
+                        buf = &buf[libafl_num_interrupts*4..];
+                        len = buf.len();
+                    }
+                    // println!("Load: {:?}", libafl_interrupt_offsets[0..libafl_num_interrupts].to_vec());
+                }
+                if len > MAX_INPUT_SIZE {
+                    buf = &buf[0..MAX_INPUT_SIZE];
+                    len = MAX_INPUT_SIZE;
+                }
+
+                emu.write_phys_mem(input_addr, buf);
+                if let Some(s) = test_length_ptr {
+                    emu.write_phys_mem(s as u64, &len.to_le_bytes())
+                }
+
+                emu.run();
+
+                // If the execution stops at any point other then the designated breakpoint (e.g. a breakpoint on a panic method) we consider it a crash
+                let mut pcs = (0..emu.num_cpus())
+                    .map(|i| emu.cpu_from_index(i))
+                    .map(|cpu| -> Result<u32, String> { cpu.read_reg(Regs::Pc) });
+                match pcs
+                    .find(|pc| (breakpoint..breakpoint + 5).contains(pc.as_ref().unwrap_or(&0)))
+                {
+                    Some(_) => ExitKind::Ok,
+                    None => ExitKind::Crash,
+                }
+            }
+        };
+
+        // Create an observation channel using the coverage map
+        let edges = unsafe { &mut edges::EDGES_MAP };
+        let edges_counter = unsafe { &mut edges::MAX_EDGES_NUM };
+        let edges_observer = VariableMapObserver::new("edges", edges, edges_counter);
+        #[cfg(feature = "observer_hitcounts")]
+        let edges_observer = HitcountsMapObserver::new(edges_observer);
+
+        // Create an observation channel to keep track of the execution time
+        let clock_time_observer = QemuClockObserver::new("clocktime");
+
+        let systemstate_observer = QemuSystemStateObserver::new();
+
+        // Feedback to rate the interestingness of an input
+        // This one is composed by two Feedbacks in OR
+        let mut feedback = feedback_or!(
+            // Time feedback, this one does not need a feedback state
+            ClockTimeFeedback::new_with_observer(&clock_time_observer)
+        );
+        #[cfg(feature = "feed_genetic")]
+        let mut feedback = feedback_or!(
+            feedback,
+            AlwaysTrueFeedback::new()
+        );
+        #[cfg(feature = "feed_afl")]
+        let mut feedback = feedback_or!(
+            feedback,
+            // New maximization map feedback linked to the edges observer and the feedback state
+            MaxMapFeedback::new_tracking(&edges_observer, true, true)
+        );
+        #[cfg(feature = "feed_longest")]
+        let mut feedback = feedback_or!(
+            // afl feedback needs to be activated first for MapIndexesMetadata
+            feedback,
+            // Feedback to reward any input which increses the execution time
+            ExecTimeIncFeedback::new()
+        );
+        #[cfg(all(feature = "systemstate",not(any(feature = "feed_systemgraph",feature = "feed_systemtrace"))))]
+        let mut feedback = feedback_or!(
+            feedback,
+            DumpSystraceFeedback::with_dump(env::var("TRACE_DUMP").ok().map(PathBuf::from))
+        );
+        #[cfg(feature = "feed_systemtrace")]
+        let mut feedback = feedback_or!(
+            feedback,
+            // AlwaysTrueFeedback::new(),
+            NovelSystemStateFeedback::default()
+        );
+        #[cfg(feature = "feed_systemgraph")]
+        let mut feedback = feedback_or!(
+            feedback,
+            SysMapFeedback::default()
+        );
+
+        // A feedback to choose if an input is a solution or not
+        let mut objective = feedback_or_fast!(CrashFeedback::new(), TimeoutFeedback::new());
+
+        // If not restarting, create a State from scratch
+        let mut state = state.unwrap_or_else(|| {
+            StdState::new(
+                // RNG
+                unsafe {StdRand::with_seed(RNG_SEED) },
+                // Corpus that will be evolved, we keep it in memory for performance
+                InMemoryCorpus::new(),
+                // Corpus in which we store solutions (crashes in this example),
+                // on disk so the user can get them after stopping the fuzzer
+                OnDiskCorpus::new(objective_dir.clone()).unwrap(),
+                // States of the feedbacks.
+                // The feedbacks can report the data that should persist in the State.
+                &mut feedback,
+                // Same for objective feedbacks
+                &mut objective,
+            )
+            .unwrap()
+        });
+
+        // A minimization+queue policy to get testcasess from the corpus
+        #[cfg(not(any(feature = "feed_afl",feature = "feed_systemgraph",feature = "feed_systemtrace", feature = "feed_genetic")))]
+        let scheduler = QueueScheduler::new();
+        #[cfg(all(feature = "feed_afl",not(any(feature = "feed_systemgraph",feature = "feed_systemtrace"))))]
+        let scheduler = TimeMaximizerCorpusScheduler::new(QueueScheduler::new());
+        #[cfg(feature = "feed_systemtrace")]
+        let scheduler = LongestTraceScheduler::new(TimeStateMaximizerCorpusScheduler::new(QueueScheduler::new()));
+        #[cfg(feature = "feed_systemgraph")]
+        let scheduler = GraphMaximizerCorpusScheduler::new(QueueScheduler::new());
+        #[cfg(feature = "feed_genetic")]
+        let scheduler = GenerationScheduler::new();
+
+        // A fuzzer with feedbacks and a corpus scheduler
+        let mut fuzzer = StdFuzzer::new(scheduler, feedback, objective);
+        #[cfg(not(feature = "systemstate"))]
+        let qhelpers = tuple_list!(
+                QemuEdgeCoverageHelper::default(),
+                QemuStateRestoreHelper::new()
+            );
+        #[cfg(feature = "systemstate")]
+        let qhelpers = tuple_list!(
+                QemuEdgeCoverageHelper::default(),
+                QemuStateRestoreHelper::new(),
+                QemuSystemStateHelper::new(svh,curr_tcb_pointer,task_queue_addr,input_counter_ptr,app_range.clone())
+            );
+        let mut hooks = QemuHooks::new(&emu,qhelpers);
+
+        #[cfg(not(feature = "systemstate"))]
+        let observer_list = tuple_list!(edges_observer, clock_time_observer);
+        #[cfg(feature = "systemstate")]
+        let observer_list = tuple_list!(edges_observer, clock_time_observer, systemstate_observer);
+
+        // Create a QEMU in-process executor
+        let executor = QemuExecutor::new(
+            &mut hooks,
+            &mut harness,
+            observer_list,
+            &mut fuzzer,
+            &mut state,
+            &mut mgr,
+        )
+        .expect("Failed to create QemuExecutor");
+
+        // Wrap the executor to keep track of the timeout
+        let mut executor = TimeoutExecutor::new(executor, timeout);
+
+        let mutations = havoc_mutations();
+        // Setup an havoc mutator with a mutational stage
+        let mutator = StdScheduledMutator::new(mutations);
+        // #[cfg(not(all(feature = "feed_systemtrace", feature = "fuzz_int")))]
+        // let mut stages = tuple_list!(StdMutationalStage::new(mutator));
+        // #[cfg(all(feature = "feed_systemtrace", feature = "fuzz_int"))]
+        #[cfg(feature = "fuzz_int")]
+        let mut stages = tuple_list!(StdMutationalStage::new(mutator),MyStateStage::new());
+        #[cfg(not(feature = "fuzz_int"))]
+        let mut stages = tuple_list!(StdMutationalStage::new(mutator));
+
+        if env::var("DO_SHOWMAP").is_ok() {
+            let s = &env::var("DO_SHOWMAP").unwrap();
+            let show_input = if s=="-" {
+                    let mut buf = Vec::<u8>::new();
+                    std::io::stdin().read_to_end(&mut buf).expect("Could not read Stdin");
+                    buf
+                } else if s=="$" {
+                    env::var("SHOWMAP_TEXTINPUT").expect("SHOWMAP_TEXTINPUT not set").as_bytes().to_owned()
+                } else {
+                    fs::read(s).expect("Input file for DO_SHOWMAP can not be read")
+                };
+            fuzzer.evaluate_input(&mut state, &mut executor, &mut mgr, BytesInput::new(show_input))
+                .unwrap();
+            if let Ok(td) = env::var("TIME_DUMP") {
+                let mut file = OpenOptions::new()
+                    .read(true)
+                    .write(true)
+                    .create(true)
+                    .append(true)
+                    .open(td).expect("Could not open timedump");
+                if let Some(ichist) = state.metadata_mut().get_mut::<IcHist>() {
+                    for i in ichist.0.drain(..) {
+                        writeln!(file, "{},{}", i.0, i.1).expect("Write to dump failed");
+                    }
+                }
+            }
+        } else {
+            if let Ok(_) = env::var("SEED_RANDOM") {
+                unsafe {
+                    let mut rng = StdRng::seed_from_u64(RNG_SEED);
+                    for i in 0..100 {
+                        let inp = BytesInput::new(vec![rng.gen::<u8>(); MAX_INPUT_SIZE]);
+                        fuzzer.evaluate_input(&mut state, &mut executor, &mut mgr, inp).unwrap();
+                    }
+                }
+            }
+            else if let Ok(sf) = env::var("SEED_DIR") {
+                state
+                    .load_initial_inputs(&mut fuzzer, &mut executor, &mut mgr, &[PathBuf::from(&sf)])
+                    .unwrap_or_else(|_| {
+                        println!("Failed to load initial corpus at {:?}", &corpus_dirs);
+                        process::exit(0);
+                    });
+                println!("We imported {} inputs from seedfile.", state.corpus().count());
+            } else if state.corpus().count() < 1 {
+                state
+                    .load_initial_inputs(&mut fuzzer, &mut executor, &mut mgr, &corpus_dirs)
+                    .unwrap_or_else(|_| {
+                        println!("Failed to load initial corpus at {:?}", &corpus_dirs);
+                        process::exit(0);
+                    });
+                println!("We imported {} inputs from disk.", state.corpus().count());
+            }
+
+            match env::var("FUZZ_ITERS") {
+                Err(_) => {
+                    fuzzer
+                        .fuzz_loop(&mut stages, &mut executor, &mut state, &mut mgr)
+                        .unwrap();
+                },
+                Ok(t) => {
+                    println!("Iterations {}",t);
+                    let num = str::parse::<u64>(&t).expect("FUZZ_ITERS was not a number");
+                    if let Ok(s) = env::var("FUZZ_RANDOM") { unsafe {
+                        if s.contains("watersv2_int") {
+                            println!("V2");
+                            LIMIT=7000000;
+                        } else {
+                            println!("V1");
+                            LIMIT=5000000;
+                        }
+                        println!("Random Fuzzing, ignore corpus");
+                        // let mut generator = RandBytesGenerator::new(MAX_INPUT_SIZE);
+                        let target_duration = Duration::from_secs(num);
+                        let start_time = std::time::Instant::now();
+                        let mut rng = StdRng::seed_from_u64(RNG_SEED);
+                        while start_time.elapsed() < target_duration {
+                            // let inp = generator.generate(&mut state).unwrap();
+                            // libafl's generator is too slow
+                            let inp = BytesInput::new(vec![rng.gen::<u8>(); MAX_INPUT_SIZE]);
+                            fuzzer.evaluate_input(&mut state, &mut executor, &mut mgr, inp).unwrap();
+                        }
+                    }} else {
+                        // fuzzer
+                        //     .fuzz_loop_for_duration(&mut stages, &mut executor, &mut state, &mut mgr, Duration::from_secs(num))
+                        //     .unwrap();
+                        fuzzer
+                            .fuzz_loop_until(&mut stages, &mut executor, &mut state, &mut mgr, starttime.checked_add(Duration::from_secs(num)).unwrap())
+                            .unwrap();
+                        #[cfg(feature = "run_until_saturation")]
+                        {
+                            {
+                            let mut dumper = |marker : String| {
+                            if let Ok(td) = env::var("TIME_DUMP") {
+                                let mut file = OpenOptions::new()
+                                    .read(true)
+                                    .write(true)
+                                    .create(true)
+                                    .append(true)
+                                    .open(td).expect("Could not open timedump");
+                                if let Some(ichist) = state.metadata_mut().get_mut::<IcHist>() {
+                                    for i in ichist.0.drain(..) {
+                                        writeln!(file, "{},{}", i.0, i.1).expect("Write to dump failed");
+                                    }
+                                }
+                            }
+                            if let Ok(td) = env::var("CASE_DUMP") {
+                                println!("Dumping worst case to {:?}", td);
+                                let corpus = state.corpus();
+                                let mut worst = Duration::new(0,0);
+                                let mut worst_input = None;
+                                for i in 0..corpus.count() {
+                                    let tc = corpus.get(i).expect("Could not get element from corpus").borrow();
+                                    if worst < tc.exec_time().expect("Testcase missing duration") {
+                                        worst_input = Some(tc.input().as_ref().unwrap().bytes().to_owned());
+                                        worst = tc.exec_time().expect("Testcase missing duration");
+                                    }
+                                }
+                                match worst_input {
+                                    Some(wi) => {
+                                        // let cd = format!("{}.case",&td);
+                                        let mut cd = td.clone();
+                                        cd.push_str(&marker);
+                                        fs::write(&cd,wi).expect("Failed to write worst corpus element");
+                                    },
+                                    None => (),
+                                }
+                                #[cfg(feature = "feed_systemgraph")]
+                                {
+                                    let mut gd = String::from(&td);
+                                    gd.push_str(&format!(".graph{}", marker));
+                                    if let Some(md) = state.named_metadata_mut().get_mut::<SysGraphFeedbackState>("SysMap") {
+                                        fs::write(&gd,ron::to_string(&md).expect("Failed to serialize graph")).expect("Failed to write graph");
+                                    }
+                                }
+                                {
+                                    let mut gd = String::from(&td);
+                                    if let Some(md) = state.metadata_mut().get_mut::<TopRatedsMetadata>() {
+                                        let mut uniq: Vec<usize> = md.map.values().map(|x| x.clone()).collect();
+                                        uniq.sort();
+                                        uniq.dedup();
+                                        gd.push_str(&format!(".{}.toprated{}", uniq.len(), marker));
+                                        fs::write(&gd,ron::to_string(&md.map).expect("Failed to serialize metadata")).expect("Failed to write graph");
+                                    }
+                                }
+                            }
+                            };
+                            dumper(format!(".iter_{}",t));
+                            }
+                            println!("Start running until saturation");
+                            let mut last = state.metadata().get::<IcHist>().unwrap().1;
+                            while SystemTime::now().duration_since(unsafe {FUZZ_START_TIMESTAMP}).unwrap().as_millis() < last.1 + Duration::from_secs(10800).as_millis() {
+                                starttime=starttime.checked_add(Duration::from_secs(30)).unwrap();
+                                fuzzer
+                                    .fuzz_loop_until(&mut stages, &mut executor, &mut state, &mut mgr, starttime)
+                                    .unwrap();
+                                let after = state.metadata().get::<IcHist>().unwrap().1;
+                                if after.0 > last.0 {
+                                    last=after;
+                                }
+                                if let Ok(td) = env::var("CASE_DUMP") {
+                                    println!("Dumping worst case to {:?}", td);
+                                    let corpus = state.corpus();
+                                    let mut worst = Duration::new(0,0);
+                                    let mut worst_input = None;
+                                    for i in 0..corpus.count() {
+                                        let tc = corpus.get(i).expect("Could not get element from corpus").borrow();
+                                        if worst < tc.exec_time().expect("Testcase missing duration") {
+                                            worst_input = Some(tc.input().as_ref().unwrap().bytes().to_owned());
+                                            worst = tc.exec_time().expect("Testcase missing duration");
+                                        }
+                                    }
+                                    match worst_input {
+                                        Some(wi) => {
+                                            // let cd = format!("{}.case",&td);
+                                            let cd = td.clone();
+                                            fs::write(&cd,wi).expect("Failed to write worst corpus element");
+                                        },
+                                        None => (),
+                                    }
+                                    #[cfg(feature = "feed_systemgraph")]
+                                    {
+                                        let mut gd = String::from(&td);
+                                        gd.push_str(".graph" );
+                                        if let Some(md) = state.named_metadata_mut().get_mut::<SysGraphFeedbackState>("SysMap") {
+                                            fs::write(&gd,ron::to_string(&md).expect("Failed to serialize graph")).expect("Failed to write graph");
+                                        }
+                                    }
+                                    {
+                                        let mut gd = String::from(&td);
+                                        if let Some(md) = state.metadata_mut().get_mut::<TopRatedsMetadata>() {
+                                            let mut uniq: Vec<usize> = md.map.values().map(|x| x.clone()).collect();
+                                            uniq.sort();
+                                            uniq.dedup();
+                                            gd.push_str(&format!(".{}.toprated", uniq.len()));
+                                            fs::write(&gd,ron::to_string(&md.map).expect("Failed to serialize metadata")).expect("Failed to write graph");
+                                        }
+                                    }
+                                }
+                            }
+                        }
+                    }
+                    if let Ok(td) = env::var("TIME_DUMP") {
+                        let mut file = OpenOptions::new()
+                            .read(true)
+                            .write(true)
+                            .create(true)
+                            .append(true)
+                            .open(td).expect("Could not open timedump");
+                        if let Some(ichist) = state.metadata_mut().get_mut::<IcHist>() {
+                            for i in ichist.0.drain(..) {
+                                writeln!(file, "{},{}", i.0, i.1).expect("Write to dump failed");
+                            }
+                        }
+                    }
+                    if let Ok(td) = env::var("CASE_DUMP") {
+                        println!("Dumping worst case to {:?}", td);
+                        let corpus = state.corpus();
+                        let mut worst = Duration::new(0,0);
+                        let mut worst_input = None;
+                        for i in 0..corpus.count() {
+                            let tc = corpus.get(i).expect("Could not get element from corpus").borrow();
+                            if worst < tc.exec_time().expect("Testcase missing duration") {
+                                worst_input = Some(tc.input().as_ref().unwrap().bytes().to_owned());
+                                worst = tc.exec_time().expect("Testcase missing duration");
+                            }
+                        }
+                        match worst_input {
+                            Some(wi) => {
+                                // let cd = format!("{}.case",&td);
+                                let cd = td.clone();
+                                fs::write(&cd,wi).expect("Failed to write worst corpus element");
+                            },
+                            None => (),
+                        }
+                        #[cfg(feature = "feed_systemgraph")]
+                        {
+                            let mut gd = String::from(&td);
+                            gd.push_str(".graph");
+                            if let Some(md) = state.named_metadata_mut().get_mut::<SysGraphFeedbackState>("SysMap") {
+                                fs::write(&gd,ron::to_string(&md).expect("Failed to serialize graph")).expect("Failed to write graph");
+                            }
+                        }
+                        {
+                            let mut gd = String::from(&td);
+                            if let Some(md) = state.metadata_mut().get_mut::<TopRatedsMetadata>() {
+                                let mut uniq: Vec<usize> = md.map.values().map(|x| x.clone()).collect();
+                                uniq.sort();
+                                uniq.dedup();
+                                gd.push_str(&format!(".{}.toprated", uniq.len()));
+                                fs::write(&gd,ron::to_string(&md.map).expect("Failed to serialize metadata")).expect("Failed to write graph");
+                            }
+                        }
+                    }
+                },
+            }
+        }
+        #[cfg(not(feature = "singlecore"))]
+        return Ok(());
+    };
+
+    // Special case where no fuzzing happens, but standard input is dumped
+    if let Ok(input_dump) = env::var("DUMP_SEED") {
+        // Initialize QEMU
+        let args: Vec<String> = env::args().collect();
+        let env: Vec<(String, String)> = env::vars().collect();
+        let emu = Emulator::new(&args, &env);
+
+        if let Some(main_addr) = main_addr {
+            unsafe { libafl_qemu_set_native_breakpoint(main_addr); }// BREAKPOINT 
+        }
+        unsafe {
+            emu.run();
+
+            let mut buf = [0u8].repeat(MAX_INPUT_SIZE);
+            emu.read_phys_mem(input_addr, buf.as_mut_slice());
+
+            let dir = env::var("SEED_DIR").map_or("./corpus".to_string(), |x| x);
+            let filename = if input_dump == "" {"input"} else {&input_dump};
+            println!("Dumping input to: {}/{}",&dir,filename);
+            fs::write(format!("{}/{}",&dir,filename), buf).expect("could not write input dump");
+        }
+        return
+}
+
+    #[cfg(feature = "singlecore")]
+    {
+        let monitor = SimpleMonitor::new(|s| println!("{}", s));
+        #[cfg(not(feature = "restarting"))]
+        {
+            let mgr = SimpleEventManager::new(monitor);
+            run_client(None, mgr, 0);
+        }
+
+        #[cfg(feature = "restarting")]
+        {
+            let mut shmem_provider = StdShMemProvider::new().unwrap();
+            let (state, mut mgr) = match SimpleRestartingEventManager::launch(monitor, &mut shmem_provider)
+            {
+                // The restarting state will spawn the same process again as child, then restarted it each time it crashes.
+                Ok(res) => res,
+                Err(err) => match err {
+                    Error::ShuttingDown => {
+                        return;
+                    }
+                    _ => {
+                        panic!("Failed to setup the restarter: {}", err);
+                    }
+                },
+            };
+            run_client(state, mgr, 0);
+        }
+    }
+    // else -> multicore
+    #[cfg(not(feature = "singlecore"))]
+    {
+        // The shared memory allocator
+        let shmem_provider = StdShMemProvider::new().expect("Failed to init shared memory");
+
+        // The stats reporter for the broker
+        let monitor = MultiMonitor::new(|s| println!("{}", s));
+
+        // Build and run a Launcher
+        match Launcher::builder()
+            .shmem_provider(shmem_provider)
+            .broker_port(broker_port)
+            .configuration(EventConfig::from_build_id())
+            .monitor(monitor)
+            .run_client(&mut run_client)
+            .cores(&cores)
+            // .stdout_file(Some("/dev/null"))
+            .build()
+            .launch()
+        {
+            Ok(()) => (),
+            Err(Error::ShuttingDown) => println!("Fuzzing stopped by user. Good bye."),
+            Err(err) => panic!("Failed to run launcher: {:?}", err),
+        }
+    }
+}

fuzzers/FRET/src/lib.rs

@@ -0,0 +1,13 @@
+#![feature(is_sorted)]
+#[cfg(target_os = "linux")]
+mod fuzzer;
+#[cfg(target_os = "linux")]
+mod clock;
+#[cfg(target_os = "linux")]
+mod qemustate;
+#[cfg(target_os = "linux")]
+pub mod systemstate;
+#[cfg(target_os = "linux")]
+mod mutational;
+#[cfg(target_os = "linux")]
+mod worst;

fuzzers/FRET/src/main.rs

@@ -0,0 +1,24 @@
+#![feature(is_sorted)]
+//! A libfuzzer-like fuzzer using qemu for binary-only coverage
+#[cfg(target_os = "linux")]
+mod fuzzer;
+#[cfg(target_os = "linux")]
+mod clock;
+#[cfg(target_os = "linux")]
+mod qemustate;
+#[cfg(target_os = "linux")]
+mod systemstate;
+#[cfg(target_os = "linux")]
+mod worst;
+#[cfg(target_os = "linux")]
+mod mutational;
+
+#[cfg(target_os = "linux")]
+pub fn main() {
+    fuzzer::fuzz();
+}
+
+#[cfg(not(target_os = "linux"))]
+pub fn main() {
+    panic!("qemu-user and libafl_qemu is only supported on linux!");
+}

fuzzers/FRET/src/mutational.rs

@@ -0,0 +1,240 @@
+//| The [`MutationalStage`] is the default stage used during fuzzing.
+//! For the current input, it will perform a range of random mutations, and then run them in the executor.
+
+use core::marker::PhantomData;
+use std::cmp::{max, min};
+
+use libafl::{
+    bolts::rands::Rand,
+    corpus::{Corpus, self},
+    fuzzer::Evaluator,
+    mark_feature_time,
+    stages::{Stage},
+    start_timer,
+    state::{HasClientPerfMonitor, HasCorpus, HasRand, UsesState, HasMetadata},
+    Error, prelude::{HasBytesVec, UsesInput, new_hash_feedback, StdRand, RandomSeed, MutationResult, Mutator},
+};
+use crate::{systemstate::{FreeRTOSSystemStateMetadata, RefinedFreeRTOSSystemState}, fuzzer::DO_NUM_INTERRUPT, clock::IcHist};
+
+pub const MINIMUM_INTER_ARRIVAL_TIME : u32 = 700 * 1000 * (1 << 4);
+
+//======================= Custom mutator
+
+/// The default mutational stage
+#[derive(Clone, Debug, Default)]
+pub struct MyStateStage<E, EM, Z> {
+    #[allow(clippy::type_complexity)]
+    phantom: PhantomData<(E, EM, Z)>,
+}
+
+impl<E, EM, Z> MyStateStage<E, EM, Z>
+where
+    E: UsesState<State = Z::State>,
+    EM: UsesState<State = Z::State>,
+    Z: Evaluator<E, EM>,
+    Z::State: HasClientPerfMonitor + HasCorpus + HasRand,
+{
+    pub fn new() -> Self {
+        Self { phantom: PhantomData }
+    }
+}
+
+impl<E, EM, Z> Stage<E, EM, Z> for MyStateStage<E, EM, Z>
+where
+    E: UsesState<State = Z::State>,
+    EM: UsesState<State = Z::State>,
+    Z: Evaluator<E, EM>,
+    Z::State: HasClientPerfMonitor + HasCorpus + HasRand + HasMetadata,
+    <Z::State as UsesInput>::Input: HasBytesVec
+{
+    fn perform(
+        &mut self,
+        fuzzer: &mut Z,
+        executor: &mut E,
+        state: &mut Self::State,
+        manager: &mut EM,
+        corpus_idx: usize,
+    ) -> Result<(), Error> {
+        let mut _input = state
+                .corpus()
+                .get(corpus_idx)?
+                .borrow_mut().clone();
+        let mut newinput = _input.input_mut().as_mut().unwrap().clone();
+        // let mut tmpinput = _input.input_mut().as_mut().unwrap().clone();
+        let mut do_rerun = false;
+        {
+            // need our own random generator, because borrowing rules
+            let mut myrand = StdRand::new();
+            let mut target_bytes : Vec<u8> = vec![];
+            {
+                let input = _input.input_mut().as_ref().unwrap();
+                let tmp = &mut state.rand_mut();
+                myrand.set_seed(tmp.next());   
+                target_bytes = input.bytes().to_vec();
+            }
+
+            // produce a slice of absolute interrupt times
+            let mut interrupt_offsets : [u32; 32] = [0u32; 32];
+            let mut num_interrupts : usize = 0;
+            {
+                let mut start_tick : u32 = 0;
+                for i in 0..DO_NUM_INTERRUPT {
+                    let mut t : [u8; 4] = [0,0,0,0];
+                    if target_bytes.len() > (i+1)*4 {
+                        for j in 0 as usize..4 as usize {
+                            t[j]=target_bytes[i*4+j];
+                        }
+                        if i == 0 || true {
+                            start_tick = u32::from_le_bytes(t);
+                        } else {
+                            start_tick = u32::saturating_add(start_tick,max(MINIMUM_INTER_ARRIVAL_TIME,u32::from_le_bytes(t)));
+                        }
+                        interrupt_offsets[i] = start_tick;
+                        num_interrupts = i+1;
+                    }
+                }
+            }
+            interrupt_offsets.sort();
+
+            // println!("Vor Mutator: {:?}", interrupt_offsets[0..num_interrupts].to_vec());
+            // let num_i = min(target_bytes.len() / 4, DO_NUM_INTERRUPT);
+            let mut suffix = target_bytes.split_off(4 * num_interrupts);
+            let mut prefix : Vec<[u8; 4]> = vec![];
+            // let mut suffix : Vec<u8> = vec![];
+            #[cfg(feature = "feed_systemtrace")]
+            {
+                let tmp = _input.metadata().get::<FreeRTOSSystemStateMetadata>();
+                if tmp.is_some() {
+                let trace = tmp.expect("FreeRTOSSystemStateMetadata not found");
+
+                // calculate hits and identify snippets
+                let mut last_m = false;
+                let mut marks : Vec<(&RefinedFreeRTOSSystemState, usize, usize)>= vec![]; // 1: got interrupted, 2: interrupt handler
+                for i in 0..trace.inner.len() {
+                    let curr = &trace.inner[i];
+                    let m = interrupt_offsets[0..num_interrupts].iter().any(|x| (curr.start_tick..curr.end_tick).contains(&(*x as u64)));
+                    if m {
+                        marks.push((curr, i, 1));
+                        // println!("1: {}",curr.current_task.task_name);
+                    } else if last_m {
+                        marks.push((curr, i, 2));
+                        // println!("2: {}",curr.current_task.task_name);
+                    } else {
+                        marks.push((curr, i, 0));
+                    }
+                    last_m = m;
+                }
+                for i in 0..num_interrupts {
+                    // bounds based on minimum inter-arrival time
+                    let mut lb = 0;
+                    let mut ub : u32 = marks[marks.len()-1].0.end_tick.try_into().expect("ticks > u32");
+                    if i > 0 {
+                        lb = u32::saturating_add(interrupt_offsets[i-1],MINIMUM_INTER_ARRIVAL_TIME);
+                    }
+                    if i < num_interrupts-1 {
+                        ub = u32::saturating_sub(interrupt_offsets[i+1],MINIMUM_INTER_ARRIVAL_TIME);
+                    }
+                    // get old hit and handler
+                    let old_hit = marks.iter().filter(
+                        |x| x.0.start_tick < (interrupt_offsets[i] as u64) && (interrupt_offsets[i] as u64) < x.0.end_tick
+                    ).next();
+                    let old_handler = match old_hit {
+                        Some(s) => if s.1 < num_interrupts-1 && s.1 < marks.len()-1 {
+                            Some(marks[s.1+1])
+                        } else {None},
+                        None => None
+                    };
+                    // find reachable alternatives
+                    let alternatives : Vec<_> = marks.iter().filter(|x|
+                        x.2 != 2 &&
+                        (
+                        x.0.start_tick < (lb as u64) && (lb as u64) < x.0.end_tick
+                        || x.0.start_tick < (ub as u64) && (ub as u64) < x.0.end_tick )
+                    ).collect();
+                    // in cases there are no alternatives
+                    if alternatives.len() == 0 {
+                        if old_hit.is_none() {
+                            // choose something random
+                            let untouched : Vec<_> = marks.iter().filter(
+                                |x| x.2 == 0
+                            ).collect();
+                            if untouched.len() > 0 {
+                                let tmp = interrupt_offsets[i];
+                                let choice = myrand.choose(untouched);
+                                interrupt_offsets[i] = myrand.between(choice.0.start_tick, choice.0.end_tick)
+                                    .try_into().expect("tick > u32");
+                                do_rerun = true;
+                            }
+                            // println!("no alternatives, choose random i: {} {} -> {}",i,tmp,interrupt_offsets[i]);
+                            continue;
+                        } else {
+                            // do nothing
+                            // println!("no alternatives, do nothing i: {} {}",i,interrupt_offsets[i]);
+                            continue;
+                        }
+                    }
+                    let replacement = myrand.choose(alternatives);
+                    if (old_hit.map_or(false, |x| x == replacement)) {
+                        // use the old value
+                        // println!("chose old value, do nothing i: {} {}",i,interrupt_offsets[i]);
+                        continue;
+                    } else {
+                        let extra = if (old_hit.map_or(false, |x| x.1 < replacement.1)) {
+                            // move futher back, respect old_handler
+                            old_handler.map_or(0, |x| x.0.end_tick - x.0.start_tick)
+                        } else { 0 };
+                        let tmp = interrupt_offsets[i];
+                        interrupt_offsets[i] = (myrand.between(replacement.0.start_tick,
+                            replacement.0.end_tick) + extra).try_into().expect("ticks > u32");
+                        // println!("chose new alternative, i: {} {} -> {}",i,tmp, interrupt_offsets[i]);
+                        do_rerun = true;
+                    }
+                }
+                let mut numbers : Vec<u32> = interrupt_offsets[0..num_interrupts].to_vec();
+                numbers.sort();
+                // println!("Mutator: {:?}", numbers);
+                let mut start : u32 = 0;
+                // for i in 0..numbers.len() {
+                //     let tmp = numbers[i];
+                //     numbers[i] = numbers[i]-start;
+                //     start = tmp;
+                // }
+                for i in 0..numbers.len() {
+                    prefix.push(u32::to_le_bytes(numbers[i]));
+                }
+                }
+            }
+            #[cfg(not(feature = "feed_systemtrace"))]
+            {
+                let metadata = state.metadata();
+                let hist = metadata.get::<IcHist>().unwrap();
+                let maxtick : u64 = hist.1.0;
+                // let maxtick : u64 = (_input.exec_time().expect("No duration found").as_nanos() >> 4).try_into().unwrap();
+                let mut numbers : Vec<u32> = vec![];
+                for i in 0..num_interrupts {
+                    prefix.push(u32::to_le_bytes(myrand.between(0, min(maxtick, u32::MAX as u64)).try_into().expect("ticks > u32")));
+                }
+            }
+
+            let mut n : Vec<u8> = vec![];
+            n = [prefix.concat(), suffix].concat();
+            newinput.bytes_mut().clear();
+            newinput.bytes_mut().append(&mut n);
+        }
+        // InterruptShifterMutator::mutate(&mut mymut, state, &mut input, 0)?;
+        if do_rerun {
+            let (_, corpus_idx) = fuzzer.evaluate_input(state, executor, manager, newinput)?;
+        }
+        Ok(())
+    }
+}
+
+impl<E, EM, Z> UsesState for MyStateStage<E, EM, Z>
+where
+    E: UsesState<State = Z::State>,
+    EM: UsesState<State = Z::State>,
+    Z: Evaluator<E, EM>,
+    Z::State: HasClientPerfMonitor + HasCorpus + HasRand,
+{
+    type State = Z::State;
+}

fuzzers/FRET/src/qemustate.rs

@@ -0,0 +1,96 @@
+use libafl::prelude::UsesInput;
+use libafl_qemu::CPUArchState;
+use libafl_qemu::Emulator;
+use libafl_qemu::FastSnapshot;
+use libafl_qemu::QemuExecutor;
+use libafl_qemu::QemuHelper;
+use libafl_qemu::QemuHelperTuple;
+use libafl::{executors::ExitKind, inputs::Input, observers::ObserversTuple, state::HasMetadata};
+use libafl_qemu::QemuHooks;
+
+use libafl_qemu::{
+    emu,
+};
+// TODO be thread-safe maybe with https://amanieu.github.io/thread_local-rs/thread_local/index.html
+#[derive(Debug)]
+pub struct QemuStateRestoreHelper {
+    has_snapshot: bool,
+    use_snapshot: bool,
+    saved_cpu_states: Vec<CPUArchState>,
+    fastsnap: Option<FastSnapshot>
+}
+
+impl QemuStateRestoreHelper {
+    #[must_use]
+    pub fn new() -> Self {
+        Self {
+            has_snapshot: false,
+            use_snapshot: true,
+            saved_cpu_states: vec![],
+            fastsnap: None
+        }
+    }
+}
+
+impl Default for QemuStateRestoreHelper {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+impl<S> QemuHelper<S> for QemuStateRestoreHelper
+where
+    S: UsesInput,
+{
+    const HOOKS_DO_SIDE_EFFECTS: bool = true;
+
+    fn init_hooks<QT>(&self, _hooks: &QemuHooks<'_, QT, S>)
+    where
+        QT: QemuHelperTuple<S>,
+    {
+    }
+
+    fn first_exec<QT>(&self, _hooks: &QemuHooks<'_, QT, S>)
+    where
+        QT: QemuHelperTuple<S>,
+    {
+    }
+
+    fn post_exec(&mut self, emulator: &Emulator, _input: &S::Input) {
+        // unsafe { println!("snapshot post {}",emu::icount_get_raw()) };
+    }
+
+    fn pre_exec(&mut self, emulator: &Emulator, _input: &S::Input) {
+        // only restore in pre-exec, to preserve the post-execution state for inspection
+        #[cfg(feature = "snapshot_restore")]
+        {
+            #[cfg(feature = "snapshot_fast")]
+            match self.fastsnap {
+                Some(s) => emulator.restore_fast_snapshot(s),
+                None => {self.fastsnap = Some(emulator.create_fast_snapshot(true));},
+            }
+            #[cfg(not(feature = "snapshot_fast"))]
+            if !self.has_snapshot {
+                emulator.save_snapshot("Start", true);
+                self.has_snapshot = true;
+            }
+            else
+            {
+                emulator.load_snapshot("Start", true);
+            }
+        }
+        #[cfg(not(feature = "snapshot_restore"))]
+        if !self.has_snapshot {
+            self.saved_cpu_states = (0..emulator.num_cpus())
+                .map(|i| emulator.cpu_from_index(i).save_state())
+                .collect();
+            self.has_snapshot = true;
+        } else {
+            for (i, s) in self.saved_cpu_states.iter().enumerate() {
+                emulator.cpu_from_index(i).restore_state(s);
+            }
+        }
+
+        // unsafe { println!("snapshot pre {}",emu::icount_get_raw()) };
+    }
+}

fuzzers/FRET/src/systemstate/feedbacks.rs

@@ -0,0 +1,299 @@
+use libafl::SerdeAny;
+use libafl::bolts::ownedref::OwnedSlice;
+use libafl::inputs::BytesInput;
+use libafl::prelude::UsesInput;
+use libafl::state::HasNamedMetadata;
+use std::path::PathBuf;
+use crate::clock::QemuClockObserver;
+use libafl::corpus::Testcase;
+use libafl::bolts::tuples::MatchName;
+use std::collections::hash_map::DefaultHasher;
+use std::hash::Hasher;
+use std::hash::Hash;
+use libafl::events::EventFirer;
+use libafl::state::HasClientPerfMonitor;
+use libafl::feedbacks::Feedback;
+use libafl::bolts::tuples::Named;
+use libafl::Error;
+use hashbrown::HashMap;
+use libafl::{executors::ExitKind, inputs::Input, observers::ObserversTuple, state::HasMetadata};
+use serde::{Deserialize, Serialize};
+
+use super::RefinedFreeRTOSSystemState;
+use super::FreeRTOSSystemStateMetadata;
+use super::observers::QemuSystemStateObserver;
+use petgraph::prelude::DiGraph;
+use petgraph::graph::NodeIndex;
+use petgraph::Direction;
+use std::cmp::Ordering;
+
+//============================= Feedback
+
+/// Shared Metadata for a systemstateFeedback
+#[derive(Debug, Serialize, Deserialize, SerdeAny, Clone, Default)]
+pub struct SystemStateFeedbackState
+{
+    known_traces: HashMap<u64,(u64,u64,usize)>, // encounters,ticks,length
+    longest: Vec<RefinedFreeRTOSSystemState>,
+}
+impl Named for SystemStateFeedbackState
+{
+    #[inline]
+    fn name(&self) -> &str {
+        "systemstate"
+    }
+}
+// impl FeedbackState for systemstateFeedbackState
+// {
+//     fn reset(&mut self) -> Result<(), Error> {
+//         self.longest.clear();
+//         self.known_traces.clear();
+//         Ok(())
+//     }
+// }
+
+/// A Feedback reporting novel System-State Transitions. Depends on [`QemuSystemStateObserver`]
+#[derive(Serialize, Deserialize, Clone, Debug, Default)]
+pub struct NovelSystemStateFeedback
+{
+    last_trace: Option<Vec<RefinedFreeRTOSSystemState>>,
+    // known_traces: HashMap<u64,(u64,usize)>,
+}
+
+impl<S> Feedback<S> for NovelSystemStateFeedback
+where
+    S: UsesInput + HasClientPerfMonitor + HasNamedMetadata,
+{
+    fn is_interesting<EM, OT>(
+        &mut self,
+        state: &mut S,
+        manager: &mut EM,
+        input: &S::Input,
+        observers: &OT,
+        exit_kind: &ExitKind,
+    ) -> Result<bool, Error>
+    where
+        EM: EventFirer<State = S>,
+        OT: ObserversTuple<S>
+    {
+        let observer = observers.match_name::<QemuSystemStateObserver>("systemstate")
+            .expect("QemuSystemStateObserver not found");
+        let clock_observer = observers.match_name::<QemuClockObserver>("clocktime")  //TODO not fixed
+            .expect("QemuClockObserver not found");
+        let feedbackstate = match state
+            .named_metadata_mut()
+            .get_mut::<SystemStateFeedbackState>("systemstate") {
+                Some(s) => s,
+                None => {
+                    let n=SystemStateFeedbackState::default();
+                    state.named_metadata_mut().insert(n, "systemstate");
+                    state.named_metadata_mut().get_mut::<SystemStateFeedbackState>("systemstate").unwrap()
+                }
+            };
+        // let feedbackstate = state
+        //     .feedback_states_mut()
+        //     .match_name_mut::<systemstateFeedbackState>("systemstate")
+        //     .unwrap();
+        // Do Stuff
+        let mut hasher = DefaultHasher::new();
+        observer.last_run.hash(&mut hasher);
+        let somehash = hasher.finish();
+        let mut is_novel = false;
+        let mut takes_longer = false;
+        match feedbackstate.known_traces.get_mut(&somehash) {
+            None => {
+                is_novel = true;
+                feedbackstate.known_traces.insert(somehash,(1,clock_observer.last_runtime(),observer.last_run.len()));
+            }
+            Some(s) => {
+                s.0+=1;
+                if s.1 < clock_observer.last_runtime() {
+                    s.1 = clock_observer.last_runtime();
+                    takes_longer = true;
+                }
+            }
+        }
+        if observer.last_run.len() > feedbackstate.longest.len() {
+            feedbackstate.longest=observer.last_run.clone();
+        }
+        self.last_trace = Some(observer.last_run.clone());
+        // if (!is_novel) { println!("not novel") };
+        Ok(is_novel | takes_longer)
+    }
+
+    /// Append to the testcase the generated metadata in case of a new corpus item
+    #[inline]
+    fn append_metadata(&mut self, _state: &mut S, testcase: &mut Testcase<S::Input>) -> Result<(), Error> {
+        let a = self.last_trace.take();
+        match a {
+            Some(s) => testcase.metadata_mut().insert(FreeRTOSSystemStateMetadata::new(s)),
+            None => (),
+        }
+        Ok(())
+    }
+
+    /// Discard the stored metadata in case that the testcase is not added to the corpus
+    #[inline]
+    fn discard_metadata(&mut self, _state: &mut S, _input: &S::Input) -> Result<(), Error> {
+        self.last_trace = None;
+        Ok(())
+    }
+}
+
+impl Named for NovelSystemStateFeedback
+{
+    #[inline]
+    fn name(&self) -> &str {
+        "systemstate"
+    }
+}
+
+//=============================
+
+pub fn match_traces(target: &Vec<RefinedFreeRTOSSystemState>, last: &Vec<RefinedFreeRTOSSystemState>) -> bool {
+    let mut ret = true;
+    if target.len() > last.len() {return false;}
+    for i in 0..target.len() {
+        ret &= target[i].current_task.task_name==last[i].current_task.task_name;
+    }
+    ret
+}
+pub fn match_traces_name(target: &Vec<String>, last: &Vec<RefinedFreeRTOSSystemState>) -> bool {
+    let mut ret = true;
+    if target.len() > last.len() {return false;}
+    for i in 0..target.len() {
+        ret &= target[i]==last[i].current_task.task_name;
+    }
+    ret
+}
+
+/// A Feedback reporting novel System-State Transitions. Depends on [`QemuSystemStateObserver`]
+#[derive(Serialize, Deserialize, Clone, Debug, Default)]
+pub struct HitSystemStateFeedback
+{
+    target: Option<Vec<String>>,
+}
+
+impl<S> Feedback<S> for HitSystemStateFeedback
+where
+    S: UsesInput + HasClientPerfMonitor,
+{
+    fn is_interesting<EM, OT>(
+        &mut self,
+        state: &mut S,
+        manager: &mut EM,
+        input: &S::Input,
+        observers: &OT,
+        exit_kind: &ExitKind,
+    ) -> Result<bool, Error>
+    where
+        EM: EventFirer<State = S>,
+        OT: ObserversTuple<S>
+    {
+        let observer = observers.match_name::<QemuSystemStateObserver>("systemstate")
+            .expect("QemuSystemStateObserver not found");
+        // Do Stuff
+        match &self.target {
+            Some(s) => {
+                // #[cfg(debug_assertions)] eprintln!("Hit systemstate Feedback trigger");
+                Ok(match_traces_name(s, &observer.last_run))
+            },
+            None => Ok(false),
+        }
+    }
+}
+
+impl Named for HitSystemStateFeedback
+{
+    #[inline]
+    fn name(&self) -> &str {
+        "hit_systemstate"
+    }
+}
+
+impl HitSystemStateFeedback {
+    pub fn new(target: Option<Vec<RefinedFreeRTOSSystemState>>) -> Self {
+        Self {target: target.map(|x| x.into_iter().map(|y| y.current_task.task_name).collect())}
+    }
+}
+//=========================== Debugging Feedback
+/// A [`Feedback`] meant to dump the system-traces for debugging. Depends on [`QemuSystemStateObserver`]
+#[derive(Debug)]
+pub struct DumpSystraceFeedback
+{
+    dumpfile: Option<PathBuf>,
+    dump_metadata: bool,
+    last_trace: Option<Vec<RefinedFreeRTOSSystemState>>,
+}
+
+impl<S> Feedback<S> for DumpSystraceFeedback
+where
+    S: UsesInput + HasClientPerfMonitor,
+{
+    fn is_interesting<EM, OT>(
+        &mut self,
+        state: &mut S,
+        manager: &mut EM,
+        input: &S::Input,
+        observers: &OT,
+        exit_kind: &ExitKind,
+    ) -> Result<bool, Error>
+    where
+        EM: EventFirer<State = S>,
+        OT: ObserversTuple<S>
+    {
+        let observer = observers.match_name::<QemuSystemStateObserver>("systemstate")
+            .expect("QemuSystemStateObserver not found");
+        let names : Vec<String> = observer.last_run.iter().map(|x| x.current_task.task_name.clone()).collect();
+        match &self.dumpfile {
+            Some(s) => {
+                    std::fs::write(s,ron::to_string(&observer.last_run).expect("Error serializing hashmap")).expect("Can not dump to file");
+                    self.dumpfile = None
+                },
+            None => if !self.dump_metadata {println!("{:?}\n{:?}",observer.last_run,names);}
+        };
+        if self.dump_metadata {self.last_trace=Some(observer.last_run.clone());}
+        Ok(!self.dump_metadata)
+    }
+    /// Append to the testcase the generated metadata in case of a new corpus item
+    #[inline]
+    fn append_metadata(&mut self, _state: &mut S, testcase: &mut Testcase<S::Input>) -> Result<(), Error> {
+        if !self.dump_metadata {return Ok(());}
+        let a = self.last_trace.take();
+        match a {
+            Some(s) => testcase.metadata_mut().insert(FreeRTOSSystemStateMetadata::new(s)),
+            None => (),
+        }
+        Ok(())
+    }
+
+    /// Discard the stored metadata in case that the testcase is not added to the corpus
+    #[inline]
+    fn discard_metadata(&mut self, _state: &mut S, _input: &S::Input) -> Result<(), Error> {
+        self.last_trace = None;
+        Ok(())
+    }
+}
+
+impl Named for DumpSystraceFeedback
+{
+    #[inline]
+    fn name(&self) -> &str {
+        "Dumpsystemstate"
+    }
+}
+
+impl DumpSystraceFeedback
+{
+    /// Creates a new [`DumpSystraceFeedback`]
+    #[must_use]
+    pub fn new() -> Self {
+        Self {dumpfile: None, dump_metadata: false, last_trace: None}
+    }
+    pub fn with_dump(dumpfile: Option<PathBuf>) -> Self {
+        Self {dumpfile: dumpfile, dump_metadata: false, last_trace: None}
+    }
+    pub fn metadata_only() -> Self {
+        Self {dumpfile: None, dump_metadata: true, last_trace: None}
+    }
+}

fuzzers/FRET/src/systemstate/freertos.rs

@@ -0,0 +1,122 @@
+#![allow(non_camel_case_types,non_snake_case,non_upper_case_globals,deref_nullptr)]
+use serde::{Deserialize, Serialize};
+// Manual Types
+use libafl_qemu::Emulator;
+
+/*========== Start of generated Code =============*/
+pub type char_ptr = ::std::os::raw::c_uint;
+pub type ListItem_t_ptr = ::std::os::raw::c_uint;
+pub type StackType_t_ptr = ::std::os::raw::c_uint;
+pub type void_ptr = ::std::os::raw::c_uint;
+pub type tskTaskControlBlock_ptr = ::std::os::raw::c_uint;
+pub type xLIST_ptr = ::std::os::raw::c_uint;
+pub type xLIST_ITEM_ptr = ::std::os::raw::c_uint;
+/* automatically generated by rust-bindgen 0.59.2 */
+
+pub type __uint8_t = ::std::os::raw::c_uchar;
+pub type __uint16_t = ::std::os::raw::c_ushort;
+pub type __uint32_t = ::std::os::raw::c_uint;
+pub type StackType_t = u32;
+pub type UBaseType_t = ::std::os::raw::c_uint;
+pub type TickType_t = u32;
+#[repr(C)]
+#[derive(Debug, Copy, Clone, Default, Serialize, Deserialize)]
+pub struct xLIST_ITEM {
+    pub xItemValue: TickType_t,
+    pub pxNext: xLIST_ITEM_ptr,
+    pub pxPrevious: xLIST_ITEM_ptr,
+    pub pvOwner: void_ptr,
+    pub pvContainer: xLIST_ptr,
+}
+pub type ListItem_t = xLIST_ITEM;
+#[repr(C)]
+#[derive(Debug, Copy, Clone, Default, Serialize, Deserialize)]
+pub struct xMINI_LIST_ITEM {
+    pub xItemValue: TickType_t,
+    pub pxNext: xLIST_ITEM_ptr,
+    pub pxPrevious: xLIST_ITEM_ptr,
+}
+pub type MiniListItem_t = xMINI_LIST_ITEM;
+#[repr(C)]
+#[derive(Debug, Copy, Clone, Default, Serialize, Deserialize)]
+pub struct xLIST {
+    pub uxNumberOfItems: UBaseType_t,
+    pub pxIndex: ListItem_t_ptr,
+    pub xListEnd: MiniListItem_t,
+}
+pub type List_t = xLIST;
+pub type TaskHandle_t = tskTaskControlBlock_ptr;
+pub const eTaskState_eRunning: eTaskState = 0;
+pub const eTaskState_eReady: eTaskState = 1;
+pub const eTaskState_eBlocked: eTaskState = 2;
+pub const eTaskState_eSuspended: eTaskState = 3;
+pub const eTaskState_eDeleted: eTaskState = 4;
+pub const eTaskState_eInvalid: eTaskState = 5;
+pub type eTaskState = ::std::os::raw::c_uint;
+#[repr(C)]
+#[derive(Debug, Copy, Clone, Default, Serialize, Deserialize)]
+pub struct xTASK_STATUS {
+    pub xHandle: TaskHandle_t,
+    pub pcTaskName: char_ptr,
+    pub xTaskNumber: UBaseType_t,
+    pub eCurrentState: eTaskState,
+    pub uxCurrentPriority: UBaseType_t,
+    pub uxBasePriority: UBaseType_t,
+    pub ulRunTimeCounter: u32,
+    pub pxStackBase: StackType_t_ptr,
+    pub usStackHighWaterMark: u16,
+}
+pub type TaskStatus_t = xTASK_STATUS;
+#[repr(C)]
+#[derive(Debug, Copy, Clone, Default, Serialize, Deserialize)]
+pub struct tskTaskControlBlock {
+    pub pxTopOfStack: StackType_t_ptr,
+    pub xStateListItem: ListItem_t,
+    pub xEventListItem: ListItem_t,
+    pub uxPriority: UBaseType_t,
+    pub pxStack: StackType_t_ptr,
+    pub pcTaskName: [::std::os::raw::c_char; 10usize],
+    pub uxBasePriority: UBaseType_t,
+    pub uxMutexesHeld: UBaseType_t,
+    pub ulNotifiedValue: [u32; 1usize],
+    pub ucNotifyState: [u8; 1usize],
+    pub ucStaticallyAllocated: u8,
+    pub ucDelayAborted: u8,
+}
+pub type tskTCB = tskTaskControlBlock;
+pub type TCB_t = tskTCB;
+/*========== End of generated Code =============*/
+
+pub trait emu_lookup {
+    fn lookup(emu: &Emulator, addr: ::std::os::raw::c_uint) -> Self;
+}
+
+
+#[derive(Debug, Copy, Clone, Serialize, Deserialize)]
+pub enum rtos_struct {
+    TCB_struct(TCB_t),
+    List_struct(List_t),
+    List_Item_struct(ListItem_t),
+    List_MiniItem_struct(MiniListItem_t),
+}
+
+#[macro_export]
+macro_rules! impl_emu_lookup {
+    ($struct_name:ident) => {
+        impl $crate::systemstate::freertos::emu_lookup for $struct_name {
+            fn lookup(emu: &Emulator, addr: ::std::os::raw::c_uint) -> $struct_name {
+                let mut tmp : [u8; std::mem::size_of::<$struct_name>()] = [0u8; std::mem::size_of::<$struct_name>()];
+                unsafe {
+                    emu.read_mem(addr.into(), &mut tmp);
+                    std::mem::transmute::<[u8; std::mem::size_of::<$struct_name>()], $struct_name>(tmp)
+                }
+            }
+        }
+    };
+}
+impl_emu_lookup!(TCB_t);
+impl_emu_lookup!(List_t);
+impl_emu_lookup!(ListItem_t);
+impl_emu_lookup!(MiniListItem_t);
+impl_emu_lookup!(void_ptr);
+impl_emu_lookup!(TaskStatus_t);

fuzzers/FRET/src/systemstate/graph.rs

@@ -0,0 +1,604 @@
+
+use libafl::SerdeAny;
+/// Feedbacks organizing SystemStates as a graph
+use libafl::inputs::HasBytesVec;
+use libafl::bolts::rands::RandomSeed;
+use libafl::bolts::rands::StdRand;
+use libafl::mutators::Mutator;
+use libafl::mutators::MutationResult;
+use libafl::prelude::HasTargetBytes;
+use libafl::prelude::UsesInput;
+use libafl::state::HasNamedMetadata;
+use libafl::state::UsesState;
+use core::marker::PhantomData;
+use libafl::state::HasCorpus;
+use libafl::state::HasSolutions;
+use libafl::state::HasRand;
+use crate::worst::MaxExecsLenFavFactor;
+use libafl::schedulers::MinimizerScheduler;
+use libafl::bolts::HasRefCnt;
+use libafl::bolts::AsSlice;
+use libafl::bolts::ownedref::OwnedSlice;
+use libafl::inputs::BytesInput;
+use std::path::PathBuf;
+use crate::clock::QemuClockObserver;
+use libafl::corpus::Testcase;
+use libafl::bolts::tuples::MatchName;
+use std::collections::hash_map::DefaultHasher;
+use std::hash::Hasher;
+use std::hash::Hash;
+use libafl::events::EventFirer;
+use libafl::state::HasClientPerfMonitor;
+use libafl::feedbacks::Feedback;
+use libafl::bolts::tuples::Named;
+use libafl::Error;
+use hashbrown::HashMap;
+use libafl::{executors::ExitKind, inputs::Input, observers::ObserversTuple, state::HasMetadata};
+use serde::{Deserialize, Serialize};
+
+use super::RefinedFreeRTOSSystemState;
+use super::FreeRTOSSystemStateMetadata;
+use super::observers::QemuSystemStateObserver;
+use petgraph::prelude::DiGraph;
+use petgraph::graph::NodeIndex;
+use petgraph::Direction;
+use std::cmp::Ordering;
+
+use libafl::bolts::rands::Rand;
+
+//============================= Data Structures
+#[derive(Serialize, Deserialize, Clone, Debug, PartialEq, Default)]
+pub struct VariantTuple
+{
+    pub start_tick: u64,
+    pub end_tick: u64,
+    input_counter: u32,
+    pub input: Vec<u8>, // in the end any kind of input are bytes, regardless of type and lifetime
+}
+impl VariantTuple {
+    fn from(other: &RefinedFreeRTOSSystemState,input: Vec<u8>) -> Self {
+        VariantTuple{
+            start_tick: other.start_tick,
+            end_tick: other.end_tick,
+            input_counter: other.input_counter,
+            input: input,
+        }
+    }
+}
+
+#[derive(Serialize, Deserialize, Clone, Debug, Default)]
+pub struct SysGraphNode
+{
+    base: RefinedFreeRTOSSystemState,
+    pub variants: Vec<VariantTuple>,
+}
+impl SysGraphNode {
+    fn from(base: RefinedFreeRTOSSystemState, input: Vec<u8>) -> Self {
+        SysGraphNode{variants: vec![VariantTuple::from(&base, input)], base:base }
+    }
+    /// unites the variants of this value with another, draining the other if the bases are equal
+    fn unite(&mut self, other: &mut SysGraphNode) -> bool {
+        if self!=other {return false;}
+        self.variants.append(&mut other.variants);
+        self.variants.dedup();
+        return true;
+    }
+    /// add a Varint from a [`RefinedFreeRTOSSystemState`]
+    fn unite_raw(&mut self, other: &RefinedFreeRTOSSystemState, input: &Vec<u8>) -> bool {
+        if &self.base!=other {return false;}
+        self.variants.push(VariantTuple::from(other, input.clone()));
+        self.variants.dedup();
+        return true;
+    }
+    /// add a Varint from a [`RefinedFreeRTOSSystemState`], if it's interesting
+    fn unite_interesting(&mut self, other: &RefinedFreeRTOSSystemState, input: &Vec<u8>) -> bool {
+        if &self.base!=other {return false;}
+        let interesting =
+            self.variants.iter().all(|x| x.end_tick-x.start_tick<other.end_tick-other.start_tick) ||    // longest variant
+            self.variants.iter().all(|x| x.end_tick-x.start_tick>other.end_tick-other.start_tick) ||    // shortest variant
+            self.variants.iter().all(|x| x.input_counter>other.input_counter) ||                      // longest input
+            self.variants.iter().all(|x| x.input_counter<other.input_counter);                        // shortest input
+        if interesting {
+            let var = VariantTuple::from(other, input.clone());
+            self.variants.push(var);
+        }
+        return interesting;
+    }
+    pub fn get_taskname(&self) -> &str {
+        &self.base.current_task.task_name
+    }
+    pub fn get_input_counts(&self) -> Vec<u32> {
+        self.variants.iter().map(|x| x.input_counter).collect()
+    }
+}
+impl PartialEq for SysGraphNode {
+    fn eq(&self, other: &SysGraphNode) -> bool {
+        self.base==other.base
+    }
+}
+
+// Wrapper around Vec<RefinedFreeRTOSSystemState> to attach as Metadata
+#[derive(Debug, Default, Serialize, Deserialize, Clone)]
+pub struct SysGraphMetadata {
+    pub inner: Vec<NodeIndex>,
+    indices: Vec<usize>,
+    tcref: isize,
+}
+impl SysGraphMetadata {
+    pub fn new(inner: Vec<NodeIndex>) -> Self{
+        Self {indices: inner.iter().map(|x| x.index()).collect(), inner: inner, tcref: 0}
+    }
+}
+impl AsSlice for SysGraphMetadata {
+    /// Convert the slice of system-states to a slice of hashes over enumerated states
+    fn as_slice(&self) -> &[usize] {
+        self.indices.as_slice()
+    }
+
+    type Entry = usize;
+}
+
+impl HasRefCnt for SysGraphMetadata {
+    fn refcnt(&self) -> isize {
+        self.tcref
+    }
+
+    fn refcnt_mut(&mut self) -> &mut isize {
+        &mut self.tcref
+    }
+}
+
+libafl::impl_serdeany!(SysGraphMetadata);
+
+pub type GraphMaximizerCorpusScheduler<CS> =
+    MinimizerScheduler<CS, MaxExecsLenFavFactor<<CS as UsesState>::State>,SysGraphMetadata>;
+
+//============================= Graph Feedback
+
+/// Improved System State Graph
+#[derive(Serialize, Deserialize, Clone, Debug, Default, SerdeAny)]
+pub struct SysGraphFeedbackState
+{
+    pub graph: DiGraph<SysGraphNode, ()>,
+    entrypoint: NodeIndex,
+    exit: NodeIndex,
+    name: String,
+}
+impl SysGraphFeedbackState
+{
+    pub fn new() -> Self {
+        let mut graph = DiGraph::<SysGraphNode, ()>::new();
+        let mut entry = SysGraphNode::default();
+        entry.base.current_task.task_name="Start".to_string();
+        let mut exit = SysGraphNode::default();
+        exit.base.current_task.task_name="End".to_string();
+        let entry = graph.add_node(entry);
+        let exit = graph.add_node(exit);
+        Self {graph: graph, entrypoint: entry, exit: exit, name: String::from("SysMap")}
+    }
+    fn insert(&mut self, list: Vec<RefinedFreeRTOSSystemState>, input: &Vec<u8>) {
+        let mut current_index = self.entrypoint;
+        for n in list {
+            let mut done = false;
+            for i in self.graph.neighbors_directed(current_index, Direction::Outgoing) {
+                if n == self.graph[i].base {
+                    done = true;
+                    current_index = i;
+                    break;
+                }
+            }
+            if !done {
+                let j = self.graph.add_node(SysGraphNode::from(n,input.clone()));
+                self.graph.add_edge(current_index, j, ());
+                current_index = j;
+            }
+        }
+    }
+    /// Try adding a system state path from a [Vec<RefinedFreeRTOSSystemState>], return true if the path was interesting
+    fn update(&mut self, list: &Vec<RefinedFreeRTOSSystemState>, input: &Vec<u8>) -> (bool, Vec<NodeIndex>) {
+        let mut current_index = self.entrypoint;
+        let mut novel = false;
+        let mut trace : Vec<NodeIndex> = vec![current_index];
+        for n in list {
+            let mut matching : Option<NodeIndex> = None;
+            for i in self.graph.neighbors_directed(current_index, Direction::Outgoing) {
+                let tmp = &self.graph[i];
+                if n == &tmp.base {
+                    matching = Some(i);
+                    current_index = i;
+                    break;
+                }
+            }
+            match matching {
+                None => {
+                    novel = true;
+                    let j = self.graph.add_node(SysGraphNode::from(n.clone(),input.clone()));
+                    self.graph.add_edge(current_index, j, ());
+                    current_index = j;
+                },
+                Some(i) => {
+                    novel |= self.graph[i].unite_interesting(&n, input);
+                }
+            }
+            trace.push(current_index);
+        }
+        self.graph.update_edge(current_index, self.exit, ());  // every path ends in the exit noded
+        return (novel, trace);
+    }
+}
+impl Named for SysGraphFeedbackState
+{
+    #[inline]
+    fn name(&self) -> &str {
+        &self.name
+    }
+}
+impl SysGraphFeedbackState
+{
+    fn reset(&mut self) -> Result<(), Error> {
+        self.graph.clear();
+        let mut entry = SysGraphNode::default();
+        entry.base.current_task.task_name="Start".to_string();
+        let mut exit = SysGraphNode::default();
+        exit.base.current_task.task_name="End".to_string();
+        self.entrypoint = self.graph.add_node(entry);
+        self.exit = self.graph.add_node(exit);
+        Ok(())
+    }
+}
+
+/// A Feedback reporting novel System-State Transitions. Depends on [`QemuSystemStateObserver`]
+#[derive(Serialize, Deserialize, Clone, Debug, Default)]
+pub struct SysMapFeedback
+{
+    name: String,
+    last_trace: Option<Vec<NodeIndex>>,
+}
+impl SysMapFeedback {
+    pub fn new() -> Self {
+        Self {name: String::from("SysMapFeedback"), last_trace: None }
+    }
+}
+
+impl<S> Feedback<S> for SysMapFeedback
+where
+    S: UsesInput + HasClientPerfMonitor + HasNamedMetadata,
+    S::Input: HasTargetBytes,
+{
+    #[allow(clippy::wrong_self_convention)]
+    fn is_interesting<EM, OT>(
+        &mut self,
+        state: &mut S,
+        _manager: &mut EM,
+        _input: &S::Input,
+        observers: &OT,
+        _exit_kind: &ExitKind,
+    ) -> Result<bool, Error>
+    where
+        EM: EventFirer<State = S>,
+        OT: ObserversTuple<S>,
+    {
+        let observer = observers.match_name::<QemuSystemStateObserver>("systemstate")
+            .expect("QemuSystemStateObserver not found");
+        let feedbackstate = match state
+            .named_metadata_mut()
+            .get_mut::<SysGraphFeedbackState>("SysMap") {
+                Some(s) => s,
+                None => {
+                    let n=SysGraphFeedbackState::default();
+                    state.named_metadata_mut().insert(n, "SysMap");
+                    state.named_metadata_mut().get_mut::<SysGraphFeedbackState>("SysMap").unwrap()
+                }
+            };
+        let ret = feedbackstate.update(&observer.last_run, &observer.last_input);
+        self.last_trace = Some(ret.1);
+        Ok(ret.0)
+    }
+
+    /// Append to the testcase the generated metadata in case of a new corpus item
+    #[inline]
+    fn append_metadata(&mut self, _state: &mut S, testcase: &mut Testcase<S::Input>) -> Result<(), Error> {
+        let a = self.last_trace.take();
+        match a {
+            Some(s) => testcase.metadata_mut().insert(SysGraphMetadata::new(s)),
+            None => (),
+        }
+        Ok(())
+    }
+
+    /// Discard the stored metadata in case that the testcase is not added to the corpus
+    #[inline]
+    fn discard_metadata(&mut self, _state: &mut S, _input: &S::Input) -> Result<(), Error> {
+        self.last_trace = None;
+        Ok(())
+    }
+}
+impl Named for SysMapFeedback
+{
+    #[inline]
+    fn name(&self) -> &str {
+        &self.name
+    }
+}
+
+//============================= Mutators
+//=============================== Snippets
+// pub struct RandGraphSnippetMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     phantom: PhantomData<(I, S)>,
+// }
+// impl<I, S> RandGraphSnippetMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     pub fn new() -> Self {
+//         RandGraphSnippetMutator{phantom: PhantomData}
+//     }
+// }
+// impl<I, S> Mutator<I, S> for RandGraphSnippetMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     fn mutate(
+//         &mut self, 
+//         state: &mut S, 
+//         input: &mut I, 
+//         _stage_idx: i32
+//     ) -> Result<MutationResult, Error>
+//     {
+//         // need our own random generator, because borrowing rules
+//         let mut myrand = StdRand::new();
+//         let tmp = &mut state.rand_mut();
+//         myrand.set_seed(tmp.next());   
+//         drop(tmp);
+
+//         let feedbackstate = state
+//             .feedback_states()
+//             .match_name::<SysGraphFeedbackState>("SysMap")
+//             .unwrap();
+//         let g = &feedbackstate.graph;
+//         let tmp = state.metadata().get::<SysGraphMetadata>();
+//         if tmp.is_none() {  // if there are no metadata it was probably not interesting anyways
+//             return Ok(MutationResult::Skipped);
+//         }
+//         let trace =tmp.expect("SysGraphMetadata not found");
+//         // follow the path, extract snippets from last reads, find common snippets.
+//         // those are likley keys parts. choose random parts from other sibling traces
+//         let sibling_inputs : Vec<&Vec<u8>>= g[*trace.inner.last().unwrap()].variants.iter().map(|x| &x.input).collect();
+//         let mut snippet_collector = vec![];
+//         let mut per_input_counters = HashMap::<&Vec<u8>,usize>::new(); // ugly workaround to track multiple inputs
+//         for t in &trace.inner {
+//             let node = &g[*t];
+//             let mut per_node_snippets = HashMap::<&Vec<u8>,&[u8]>::new();
+//             for v in &node.variants {
+//                 match per_input_counters.get_mut(&v.input) {
+//                     None => {
+//                         if sibling_inputs.iter().any(|x| *x==&v.input) {    // only collect info about siblin inputs from target
+//                             per_input_counters.insert(&v.input, v.input_counter.try_into().unwrap());
+//                         }
+//                     },
+//                     Some(x) => {
+//                         let x_u = *x;
+//                         if x_u<v.input_counter as usize {
+//                             *x=v.input_counter as usize;
+//                             per_node_snippets.insert(&v.input,&v.input[x_u..v.input_counter as usize]);
+//                         }
+//                     }
+//                 }
+//             }
+//             snippet_collector.push(per_node_snippets);
+//         }
+//         let mut new_input : Vec<u8> = vec![];
+//         for c in snippet_collector {
+//             new_input.extend_from_slice(myrand.choose(c).1);
+//         }
+//         for i in new_input.iter().enumerate() {
+//             input.bytes_mut()[i.0]=*i.1;
+//         }
+
+//         Ok(MutationResult::Mutated)
+//     }
+
+//     fn post_exec(
+//         &mut self, 
+//         _state: &mut S, 
+//         _stage_idx: i32, 
+//         _corpus_idx: Option<usize>
+//     ) -> Result<(), Error> {
+//         Ok(())
+//     }
+// }
+
+// impl<I, S> Named for RandGraphSnippetMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     fn name(&self) -> &str {
+//         "RandGraphSnippetMutator"
+//     }
+// }
+// //=============================== Snippets
+// pub struct RandInputSnippetMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     phantom: PhantomData<(I, S)>,
+// }
+// impl<I, S> RandInputSnippetMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     pub fn new() -> Self {
+//         RandInputSnippetMutator{phantom: PhantomData}
+//     }
+// }
+// impl<I, S> Mutator<I, S> for RandInputSnippetMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     fn mutate(
+//         &mut self, 
+//         state: &mut S, 
+//         input: &mut I, 
+//         _stage_idx: i32
+//     ) -> Result<MutationResult, Error>
+//     {
+//         // need our own random generator, because borrowing rules
+//         let mut myrand = StdRand::new();
+//         let tmp = &mut state.rand_mut();
+//         myrand.set_seed(tmp.next());   
+//         drop(tmp);
+
+//         let feedbackstate = state
+//             .feedback_states()
+//             .match_name::<SysGraphFeedbackState>("SysMap")
+//             .unwrap();
+//         let g = &feedbackstate.graph;
+//         let tmp = state.metadata().get::<SysGraphMetadata>();
+//         if tmp.is_none() {  // if there are no metadata it was probably not interesting anyways
+//             return Ok(MutationResult::Skipped);
+//         }
+//         let trace = tmp.expect("SysGraphMetadata not found");
+
+//         let mut collection : Vec<Vec<u8>> = Vec::new();
+//         let mut current_pointer : usize = 0;
+//         for t in &trace.inner {
+//             let node = &g[*t];
+//             for v in &node.variants {
+//                 if v.input == input.bytes() {
+//                     if v.input_counter > current_pointer.try_into().unwrap() {
+//                         collection.push(v.input[current_pointer..v.input_counter as usize].to_owned());
+//                         current_pointer = v.input_counter as usize;
+//                     }
+//                     break;
+//                 }
+//             }
+//         }
+//         let index_to_mutate = myrand.below(collection.len() as u64) as usize;
+//         for i in 0..collection[index_to_mutate].len() {
+//             collection[index_to_mutate][i] = myrand.below(0xFF) as u8;
+//         }
+//         for i in collection.concat().iter().enumerate() {
+//             input.bytes_mut()[i.0]=*i.1;
+//         }
+
+//         Ok(MutationResult::Mutated)
+//     }
+
+//     fn post_exec(
+//         &mut self, 
+//         _state: &mut S, 
+//         _stage_idx: i32, 
+//         _corpus_idx: Option<usize>
+//     ) -> Result<(), Error> {
+//         Ok(())
+//     }
+// }
+
+// impl<I, S> Named for RandInputSnippetMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     fn name(&self) -> &str {
+//         "RandInputSnippetMutator"
+//     }
+// }
+// //=============================== Suffix
+// pub struct RandGraphSuffixMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     phantom: PhantomData<(I, S)>,
+// }
+// impl<I, S> RandGraphSuffixMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     pub fn new() -> Self {
+//         RandGraphSuffixMutator{phantom: PhantomData}
+//     }
+// }
+// impl<I, S> Mutator<I, S> for RandGraphSuffixMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     fn mutate(
+//         &mut self, 
+//         state: &mut S, 
+//         input: &mut I, 
+//         _stage_idx: i32
+//     ) -> Result<MutationResult, Error>
+//     {
+//         // need our own random generator, because borrowing rules
+//         let mut myrand = StdRand::new();
+//         let tmp = &mut state.rand_mut();
+//         myrand.set_seed(tmp.next());   
+//         drop(tmp);
+
+//         let feedbackstate = state
+//             .feedback_states()
+//             .match_name::<SysGraphFeedbackState>("SysMap")
+//             .unwrap();
+//         let g = &feedbackstate.graph;
+//         let tmp = state.metadata().get::<SysGraphMetadata>();
+//         if tmp.is_none() {  // if there are no metadata it was probably not interesting anyways
+//             return Ok(MutationResult::Skipped);
+//         }
+//         let trace =tmp.expect("SysGraphMetadata not found");
+//         // follow the path, extract snippets from last reads, find common snippets.
+//         // those are likley keys parts. choose random parts from other sibling traces
+//         let inp_c_end = g[*trace.inner.last().unwrap()].base.input_counter;
+//         let mut num_to_reverse = myrand.below(trace.inner.len().try_into().unwrap());
+//         for t in trace.inner.iter().rev() {
+//             let int_c_prefix = g[*t].base.input_counter;
+//             if int_c_prefix < inp_c_end {
+//                 num_to_reverse-=1;
+//                 if num_to_reverse<=0 {
+//                     let mut new_input=input.bytes()[..(int_c_prefix as usize)].to_vec();
+//                     let mut ext : Vec<u8> = (int_c_prefix..inp_c_end).map(|_| myrand.next().to_le_bytes()).flatten().collect();
+//                     new_input.append(&mut ext);
+//                     for i in new_input.iter().enumerate() {
+//                         if input.bytes_mut().len()>i.0 {
+//                             input.bytes_mut()[i.0]=*i.1;
+//                         }
+//                         else { break };
+//                     }
+//                     break;
+//                 }
+//             }
+//         }
+//         Ok(MutationResult::Mutated)
+//     }
+
+//     fn post_exec(
+//         &mut self, 
+//         _state: &mut S, 
+//         _stage_idx: i32, 
+//         _corpus_idx: Option<usize>
+//     ) -> Result<(), Error> {
+//         Ok(())
+//     }
+// }
+
+// impl<I, S> Named for RandGraphSuffixMutator<I, S>
+// where
+//     I: Input + HasBytesVec,
+//     S: HasRand + HasMetadata + HasCorpus<I> + HasSolutions<I>,
+// {
+//     fn name(&self) -> &str {
+//         "RandGraphSuffixMutator"
+//     }
+// }

fuzzers/FRET/src/systemstate/helpers.rs

@@ -0,0 +1,209 @@
+use std::cell::UnsafeCell;
+use std::io::Write;
+use std::ops::Range;
+use libafl::prelude::UsesInput;
+use libafl_qemu::Emulator;
+use libafl_qemu::GuestAddr;
+use libafl_qemu::QemuHooks;
+use libafl_qemu::edges::QemuEdgesMapMetadata;
+use libafl_qemu::emu;
+use libafl_qemu::hooks;
+use crate::systemstate::RawFreeRTOSSystemState;
+use crate::systemstate::CURRENT_SYSTEMSTATE_VEC;
+use crate::systemstate::NUM_PRIOS;
+use super::freertos::TCB_t;
+use super::freertos::rtos_struct::List_Item_struct;
+use super::freertos::rtos_struct::*;
+use super::freertos;
+
+use libafl_qemu::{
+    helper::{QemuHelper, QemuHelperTuple},
+    // edges::SAVED_JUMP,
+};
+
+//============================= Struct definitions
+
+pub static mut INTR_OFFSET : Option<u64> = None;
+pub static mut INTR_DONE : bool = true;
+
+// only used when inputs are injected
+pub static mut NEXT_INPUT : Vec<u8> = Vec::new();
+
+//============================= Qemu Helper
+
+/// A Qemu Helper with reads FreeRTOS specific structs from Qemu whenever certain syscalls occur, also inject inputs
+#[derive(Debug)]
+pub struct QemuSystemStateHelper {
+    kerneladdr: u32,
+    tcb_addr: u32,
+    ready_queues: u32,
+    input_counter: Option<u64>,
+    app_range: Range<u32>,
+}
+
+impl QemuSystemStateHelper {
+    #[must_use]
+    pub fn new(
+        kerneladdr: u32,
+        tcb_addr: u32,
+        ready_queues: u32,
+        input_counter: Option<u64>,
+        app_range: Range<u32>,
+    ) -> Self {
+        QemuSystemStateHelper {
+            kerneladdr,
+            tcb_addr: tcb_addr,
+            ready_queues: ready_queues,
+            input_counter: input_counter,
+            app_range,
+        }
+    }
+}
+
+impl<S> QemuHelper<S> for QemuSystemStateHelper
+where
+    S: UsesInput,
+{
+    fn first_exec<QT>(&self, _hooks: &QemuHooks<'_, QT, S>)
+    where
+        QT: QemuHelperTuple<S>,
+    {
+        _hooks.instruction(self.kerneladdr, exec_syscall_hook::<QT, S>, false);
+        #[cfg(feature = "trace_abbs")]
+        _hooks.jmps(Some(gen_jmp_is_syscall::<QT, S>), Some(trace_api_call::<QT, S>));
+    }
+
+    // TODO: refactor duplicate code
+    fn pre_exec(&mut self, _emulator: &Emulator, _input: &S::Input) {
+        unsafe {
+            CURRENT_SYSTEMSTATE_VEC.clear();
+            let p = LAST_API_CALL.with(|x| x.get());
+            *p = None;
+        }
+    }
+
+    fn post_exec(&mut self, emulator: &Emulator, _input: &S::Input) {
+        trigger_collection(emulator, self)
+    }
+}
+
+#[inline]
+fn trigger_collection(emulator: &Emulator, h: &QemuSystemStateHelper) {
+    let listbytes : u32 = u32::try_from(std::mem::size_of::<freertos::List_t>()).unwrap();
+    let mut systemstate = RawFreeRTOSSystemState::default();
+    unsafe {
+        // TODO: investigate why can_do_io is not set sometimes, as this is just a workaround
+        let c = emulator.cpu_from_index(0);
+        let can_do_io = (*c.raw_ptr()).can_do_io;
+        (*c.raw_ptr()).can_do_io = 1;
+        systemstate.qemu_tick = emu::icount_get_raw();
+        (*c.raw_ptr()).can_do_io = can_do_io;
+    }
+    let mut buf : [u8; 4] = [0,0,0,0];
+    match h.input_counter {
+        Some(s) => unsafe { emulator.read_phys_mem(s, &mut buf); },
+        None => (),
+    };
+    systemstate.input_counter = u32::from_le_bytes(buf);
+
+    let curr_tcb_addr : freertos::void_ptr = freertos::emu_lookup::lookup(emulator, h.tcb_addr);
+    if curr_tcb_addr == 0 {
+        return;
+    };
+    systemstate.current_tcb = freertos::emu_lookup::lookup(emulator,curr_tcb_addr);
+
+    unsafe {
+        LAST_API_CALL.with(|x|
+            match *x.get() {
+                Some(s) => {
+                    systemstate.last_pc = Some(s.0 as u64);
+                },
+                None => (),
+            }
+        );
+    }
+    // println!("{:?}",std::str::from_utf8(&current_tcb.pcTaskName));
+
+    for i in 0..NUM_PRIOS {
+        let target : u32 = listbytes*u32::try_from(i).unwrap()+h.ready_queues;
+        systemstate.prio_ready_lists[i] = freertos::emu_lookup::lookup(emulator, target);
+        // println!("List at {}: {:?}",target, systemstate.prio_ready_lists[i]);
+        let mut next_index = systemstate.prio_ready_lists[i].pxIndex;
+        for _j in 0..systemstate.prio_ready_lists[i].uxNumberOfItems {
+            // always jump over the xListEnd marker
+            if (target..target+listbytes).contains(&next_index) {
+                let next_item : freertos::MiniListItem_t = freertos::emu_lookup::lookup(emulator, next_index);
+                let new_next_index=next_item.pxNext;
+                systemstate.dumping_ground.insert(next_index,List_MiniItem_struct(next_item));
+                next_index = new_next_index;
+            }
+            let next_item : freertos::ListItem_t = freertos::emu_lookup::lookup(emulator, next_index);
+            // println!("Item at {}: {:?}",next_index,next_item);
+            assert_eq!(next_item.pvContainer,target);
+            let new_next_index=next_item.pxNext;
+            let next_tcb : TCB_t= freertos::emu_lookup::lookup(emulator,next_item.pvOwner);
+            // println!("TCB at {}: {:?}",next_item.pvOwner,next_tcb);
+            systemstate.dumping_ground.insert(next_item.pvOwner,TCB_struct(next_tcb.clone()));
+            systemstate.dumping_ground.insert(next_index,List_Item_struct(next_item));
+            next_index=new_next_index;
+        }
+        // Handle edge case where the end marker was not included yet
+        if (target..target+listbytes).contains(&next_index) {
+            let next_item : freertos::MiniListItem_t = freertos::emu_lookup::lookup(emulator, next_index);
+            systemstate.dumping_ground.insert(next_index,List_MiniItem_struct(next_item));
+        }
+    }
+
+    unsafe { CURRENT_SYSTEMSTATE_VEC.push(systemstate); }
+}
+
+pub fn exec_syscall_hook<QT, S>(
+    hooks: &mut QemuHooks<'_, QT, S>,
+    _state: Option<&mut S>,
+    _pc: u32,
+)
+where
+    S: UsesInput,
+    QT: QemuHelperTuple<S>,
+{
+    let emulator = hooks.emulator();
+    let h = hooks.helpers().match_first_type::<QemuSystemStateHelper>().expect("QemuSystemHelper not found in helper tupel");
+    trigger_collection(emulator, h);
+}
+
+thread_local!(static LAST_API_CALL : UnsafeCell<Option<(GuestAddr,GuestAddr)>> = UnsafeCell::new(None));
+
+pub fn gen_jmp_is_syscall<QT, S>(
+    hooks: &mut QemuHooks<'_, QT, S>,
+    _state: Option<&mut S>,
+    src: GuestAddr,
+    dest: GuestAddr,
+) -> Option<u64>
+where
+    S: UsesInput,
+    QT: QemuHelperTuple<S>,
+{
+    if let Some(h) = hooks.helpers().match_first_type::<QemuSystemStateHelper>() {
+        if h.app_range.contains(&src) && !h.app_range.contains(&dest) {
+            // println!("New jmp {:x} {:x}", src, dest);
+            return Some(1);
+        }
+    }
+    return None;
+}
+
+pub fn trace_api_call<QT, S>(
+    _hooks: &mut QemuHooks<'_, QT, S>,
+    _state: Option<&mut S>,
+    src: GuestAddr, dest: GuestAddr, id: u64
+)
+where
+    S: UsesInput,
+    QT: QemuHelperTuple<S>,
+{
+    unsafe {
+        let p = LAST_API_CALL.with(|x| x.get());
+        *p = Some((src,dest));
+        // print!("*");
+    }
+}

fuzzers/FRET/src/systemstate/mod.rs

@@ -0,0 +1,167 @@
+//! systemstate referes to the State of a FreeRTOS fuzzing target
+use std::collections::hash_map::DefaultHasher;
+use libafl::bolts::HasRefCnt;
+use libafl::bolts::AsSlice;
+use std::hash::Hasher;
+use std::hash::Hash;
+use hashbrown::HashMap;
+use serde::{Deserialize, Serialize};
+
+use freertos::TCB_t;
+
+pub mod freertos;
+pub mod helpers;
+pub mod observers;
+pub mod feedbacks;
+pub mod graph;
+pub mod schedulers;
+
+// #[cfg(feature = "fuzz_interrupt")]
+// pub const IRQ_INPUT_BYTES_NUMBER : u32 = 2; // Offset for interrupt bytes
+// #[cfg(not(feature = "fuzz_interrupt"))]
+// pub const IRQ_INPUT_BYTES_NUMBER : u32 = 0; // Offset for interrupt bytes
+// pub const IRQ_INPUT_OFFSET : u32 = 347780; // Tick offset for app code start
+
+// Constants
+const NUM_PRIOS: usize = 5;
+
+//============================= Struct definitions
+/// Raw info Dump from Qemu
+#[derive(Debug, Default, Serialize, Deserialize)]
+pub struct RawFreeRTOSSystemState {
+    qemu_tick: u64,
+    current_tcb: TCB_t,
+    prio_ready_lists: [freertos::List_t; NUM_PRIOS],
+    dumping_ground: HashMap<u32,freertos::rtos_struct>,
+    input_counter: u32,
+    last_pc: Option<u64>,
+}
+/// List of system state dumps from QemuHelpers
+static mut CURRENT_SYSTEMSTATE_VEC: Vec<RawFreeRTOSSystemState> = vec![];
+
+/// A reduced version of freertos::TCB_t
+#[derive(Debug, Default, Serialize, Deserialize, Clone, PartialEq)]
+pub struct RefinedTCB {
+    pub task_name: String,
+    pub priority: u32,
+    pub base_priority: u32,
+    mutexes_held: u32,
+    notify_value: u32,
+    notify_state: u8,
+}
+
+impl Hash for RefinedTCB {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        self.task_name.hash(state);
+        self.priority.hash(state);
+        self.mutexes_held.hash(state);
+        #[cfg(not(feature = "no_hash_state"))]
+        self.notify_state.hash(state);
+        // self.notify_value.hash(state);
+    }
+}
+
+impl RefinedTCB {
+    pub fn from_tcb(input: &TCB_t) -> Self {
+        unsafe {
+            let tmp = std::mem::transmute::<[i8; 10],[u8; 10]>(input.pcTaskName);
+            let name : String = std::str::from_utf8(&tmp).expect("TCB name was not utf8").chars().filter(|x| *x != '\0').collect::<String>();
+            Self {
+                task_name: name,
+                priority: input.uxPriority,
+                base_priority: input.uxBasePriority,
+                mutexes_held: input.uxMutexesHeld,
+                notify_value: input.ulNotifiedValue[0],
+                notify_state: input.ucNotifyState[0],
+            }
+        }
+    }
+    pub fn from_tcb_owned(input: TCB_t) -> Self {
+        unsafe {
+            let tmp = std::mem::transmute::<[i8; 10],[u8; 10]>(input.pcTaskName);
+            let name : String = std::str::from_utf8(&tmp).expect("TCB name was not utf8").chars().filter(|x| *x != '\0').collect::<String>();
+            Self {
+                task_name: name,
+                priority: input.uxPriority,
+                base_priority: input.uxBasePriority,
+                mutexes_held: input.uxMutexesHeld,
+                notify_value: input.ulNotifiedValue[0],
+                notify_state: input.ucNotifyState[0],
+            }
+        }
+    }
+}
+
+/// Refined information about the states an execution transitioned between
+#[derive(Debug, Default, Serialize, Deserialize, Clone)]
+pub struct RefinedFreeRTOSSystemState {
+    pub start_tick: u64,
+    pub end_tick: u64,
+    last_pc: Option<u64>,
+    input_counter: u32,
+    pub current_task: RefinedTCB,
+    ready_list_after: Vec<RefinedTCB>,
+}
+impl PartialEq for RefinedFreeRTOSSystemState {
+    fn eq(&self, other: &Self) -> bool {
+        self.current_task == other.current_task && self.ready_list_after == other.ready_list_after &&
+            self.last_pc == other.last_pc
+    }
+}
+
+impl Hash for RefinedFreeRTOSSystemState {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        self.current_task.hash(state);
+        self.ready_list_after.hash(state);
+        // self.last_pc.hash(state);
+    }
+}
+impl RefinedFreeRTOSSystemState {
+    fn get_time(&self) -> u64 {
+        self.end_tick-self.start_tick
+    }
+}
+
+// Wrapper around Vec<RefinedFreeRTOSSystemState> to attach as Metadata
+#[derive(Debug, Default, Serialize, Deserialize, Clone)]
+pub struct FreeRTOSSystemStateMetadata {
+    pub inner: Vec<RefinedFreeRTOSSystemState>,
+    trace_length: usize,
+    indices: Vec<usize>,    // Hashed enumeration of States
+    tcref: isize,
+}
+impl FreeRTOSSystemStateMetadata {
+    pub fn new(inner: Vec<RefinedFreeRTOSSystemState>) -> Self{
+        let tmp = inner.iter().enumerate().map(|x| compute_hash(x) as usize).collect();
+        Self {trace_length: inner.len(), inner: inner, indices: tmp, tcref: 0}
+    }
+}
+pub fn compute_hash<T>(obj: T) -> u64 
+where
+    T: Hash
+{
+    let mut s = DefaultHasher::new();
+    obj.hash(&mut s);
+    s.finish()
+}
+
+impl AsSlice for FreeRTOSSystemStateMetadata {
+    /// Convert the slice of system-states to a slice of hashes over enumerated states
+    fn as_slice(&self) -> &[usize] {
+        self.indices.as_slice()
+    }
+
+    type Entry = usize;
+}
+
+impl HasRefCnt for FreeRTOSSystemStateMetadata {
+    fn refcnt(&self) -> isize {
+        self.tcref
+    }
+
+    fn refcnt_mut(&mut self) -> &mut isize {
+        &mut self.tcref
+    }
+}
+
+libafl::impl_serdeany!(FreeRTOSSystemStateMetadata);

fuzzers/FRET/src/systemstate/observers.rs

@@ -0,0 +1,133 @@
+// use crate::systemstate::IRQ_INPUT_BYTES_NUMBER;
+use libafl::prelude::{ExitKind, AsSlice};
+use libafl::{inputs::HasTargetBytes, prelude::UsesInput};
+use libafl::bolts::HasLen;
+use libafl::bolts::tuples::Named;
+use libafl::Error;
+use libafl::observers::Observer;
+use hashbrown::HashMap;
+use serde::{Deserialize, Serialize};
+
+use super::{
+    CURRENT_SYSTEMSTATE_VEC,
+    RawFreeRTOSSystemState,
+    RefinedTCB,
+    RefinedFreeRTOSSystemState,
+    freertos::{List_t, TCB_t, rtos_struct, rtos_struct::*},
+};
+
+//============================= Observer
+
+/// The Qemusystemstate Observer retrieves the systemstate
+/// that will get updated by the target.
+#[derive(Serialize, Deserialize, Debug, Default)]
+#[allow(clippy::unsafe_derive_deserialize)]
+pub struct QemuSystemStateObserver
+{
+    pub last_run: Vec<RefinedFreeRTOSSystemState>,
+    pub last_input: Vec<u8>,
+    name: String,
+}
+
+impl<S> Observer<S> for QemuSystemStateObserver
+where
+    S: UsesInput,
+    S::Input : HasTargetBytes,
+{
+    #[inline]
+    fn pre_exec(&mut self, _state: &mut S, _input: &S::Input) -> Result<(), Error> {
+        unsafe {CURRENT_SYSTEMSTATE_VEC.clear(); }
+        Ok(())
+    }
+
+    #[inline]
+    fn post_exec(&mut self, _state: &mut S, _input: &S::Input, _exit_kind: &ExitKind) -> Result<(), Error> {
+        unsafe {self.last_run = refine_system_states(&mut CURRENT_SYSTEMSTATE_VEC);}
+        self.last_input=_input.target_bytes().as_slice().to_owned();
+        Ok(())
+    }
+}
+
+impl Named for QemuSystemStateObserver
+{
+    #[inline]
+    fn name(&self) -> &str {
+        self.name.as_str()
+    }
+}
+
+impl HasLen for QemuSystemStateObserver
+{
+    #[inline]
+    fn len(&self) -> usize {
+        self.last_run.len()
+    }
+}
+
+impl QemuSystemStateObserver {
+    pub fn new() -> Self {
+        Self{last_run: vec![], last_input: vec![], name: "systemstate".to_string()}
+    }
+
+}
+
+//============================= Parsing helpers
+
+/// Parse a List_t containing TCB_t into Vec<TCB_t> from cache. Consumes the elements from cache
+fn tcb_list_to_vec_cached(list: List_t, dump: &mut HashMap<u32,rtos_struct>) -> Vec<TCB_t>
+{
+    let mut ret : Vec<TCB_t> = Vec::new();
+    if list.uxNumberOfItems == 0 {return ret;}
+    let last_list_item = match dump.remove(&list.pxIndex).expect("List_t entry was not in Hashmap") {
+        List_Item_struct(li) => li,
+        List_MiniItem_struct(mli) => match dump.remove(&mli.pxNext).expect("MiniListItem pointer invaild") {
+                List_Item_struct(li) => li,
+                _ => panic!("MiniListItem of a non empty List does not point to ListItem"),
+            },
+        _ => panic!("List_t entry was not a ListItem"),
+    };
+    let mut next_index = last_list_item.pxNext;
+    let last_tcb = match dump.remove(&last_list_item.pvOwner).expect("ListItem Owner not in Hashmap") {
+        TCB_struct(t) => t,
+        _ => panic!("List content does not equal type"),
+    };
+    for _ in 0..list.uxNumberOfItems-1 {
+        let next_list_item = match dump.remove(&next_index).expect("List_t entry was not in Hashmap") {
+            List_Item_struct(li) => li,
+            List_MiniItem_struct(mli) => match dump.remove(&mli.pxNext).expect("MiniListItem pointer invaild") {
+                    List_Item_struct(li) => li,
+                    _ => panic!("MiniListItem of a non empty List does not point to ListItem"),
+                },
+            _ => panic!("List_t entry was not a ListItem"),
+        };
+        match dump.remove(&next_list_item.pvOwner).expect("ListItem Owner not in Hashmap") {
+            TCB_struct(t) => {ret.push(t)},
+            _ => panic!("List content does not equal type"),
+        }
+        next_index=next_list_item.pxNext;
+    }
+    ret.push(last_tcb);
+    ret
+}
+/// Drains a List of raw SystemStates to produce a refined trace
+fn refine_system_states(input: &mut Vec<RawFreeRTOSSystemState>) -> Vec<RefinedFreeRTOSSystemState> {
+let mut ret = Vec::<RefinedFreeRTOSSystemState>::new();
+let mut start_tick : u64 = 0;
+for mut i in input.drain(..) {
+    let mut collector = Vec::<RefinedTCB>::new();
+    for j in i.prio_ready_lists.into_iter().rev() {
+        let mut tmp = tcb_list_to_vec_cached(j,&mut i.dumping_ground).iter().map(|x| RefinedTCB::from_tcb(x)).collect();
+        collector.append(&mut tmp);
+    }
+    ret.push(RefinedFreeRTOSSystemState {
+        current_task: RefinedTCB::from_tcb_owned(i.current_tcb),
+        start_tick: start_tick,
+        end_tick: i.qemu_tick,
+        ready_list_after: collector,
+        input_counter: i.input_counter,//+IRQ_INPUT_BYTES_NUMBER,
+        last_pc: i.last_pc,
+    });
+    start_tick=i.qemu_tick;
+}
+return ret;
+}

fuzzers/FRET/src/systemstate/schedulers.rs

@@ -0,0 +1,267 @@
+//! The Minimizer schedulers are a family of corpus schedulers that feed the fuzzer
+//! with testcases only from a subset of the total corpus.
+
+use core::{marker::PhantomData};
+use std::{cmp::{max, min}, mem::swap, borrow::BorrowMut};
+
+use serde::{Deserialize, Serialize};
+
+use libafl::{
+    bolts::{rands::Rand, serdeany::SerdeAny, AsSlice, HasRefCnt},
+    corpus::{Corpus, Testcase},
+    inputs::UsesInput,
+    schedulers::{Scheduler, TestcaseScore, minimizer::DEFAULT_SKIP_NON_FAVORED_PROB },
+    state::{HasCorpus, HasMetadata, HasRand, UsesState, State},
+    Error, SerdeAny, prelude::HasLen,
+    
+};
+
+use crate::worst::MaxTimeFavFactor;
+
+use super::FreeRTOSSystemStateMetadata;
+
+/// A state metadata holding a map of favoreds testcases for each map entry
+#[derive(Debug, Serialize, Deserialize, SerdeAny, Default)]
+pub struct LongestTracesMetadata {
+    /// map index -> corpus index
+    pub max_trace_length: usize,
+}
+
+impl LongestTracesMetadata {
+    fn new(l : usize) -> Self {
+        Self {max_trace_length: l}
+    }
+}
+
+/// The [`MinimizerScheduler`] employs a genetic algorithm to compute a subset of the
+/// corpus that exercise all the requested features (e.g. all the coverage seen so far)
+/// prioritizing [`Testcase`]`s` using [`TestcaseScore`]
+#[derive(Debug, Clone)]
+pub struct LongestTraceScheduler<CS> {
+    base: CS,
+    skip_non_favored_prob: u64,
+}
+
+impl<CS> UsesState for LongestTraceScheduler<CS>
+where
+    CS: UsesState,
+{
+    type State = CS::State;
+}
+
+impl<CS> Scheduler for LongestTraceScheduler<CS>
+where
+    CS: Scheduler,
+    CS::State: HasCorpus + HasMetadata + HasRand,
+{
+    /// Add an entry to the corpus and return its index
+    fn on_add(&self, state: &mut CS::State, idx: usize) -> Result<(), Error> {
+        let l = state.corpus()
+                .get(idx)?
+                .borrow()
+                .metadata()
+                .get::<FreeRTOSSystemStateMetadata>().map_or(0, |x| x.trace_length);
+        self.get_update_trace_length(state,l);
+        self.base.on_add(state, idx)
+    }
+
+    /// Replaces the testcase at the given idx
+    fn on_replace(
+        &self,
+        state: &mut CS::State,
+        idx: usize,
+        testcase: &Testcase<<CS::State as UsesInput>::Input>,
+    ) -> Result<(), Error> {
+        let l = state.corpus()
+                .get(idx)?
+                .borrow()
+                .metadata()
+                .get::<FreeRTOSSystemStateMetadata>().map_or(0, |x| x.trace_length);
+        self.get_update_trace_length(state, l);
+        self.base.on_replace(state, idx, testcase)
+    }
+
+    /// Removes an entry from the corpus, returning M if M was present.
+    fn on_remove(
+        &self,
+        state: &mut CS::State,
+        idx: usize,
+        testcase: &Option<Testcase<<CS::State as UsesInput>::Input>>,
+    ) -> Result<(), Error> {
+        self.base.on_remove(state, idx, testcase)?;
+        Ok(())
+    }
+
+    /// Gets the next entry
+    fn next(&self, state: &mut CS::State) -> Result<usize, Error> {
+        let mut idx = self.base.next(state)?;
+        while {
+            let l = state.corpus()
+                    .get(idx)?
+                    .borrow()
+                    .metadata()
+                    .get::<FreeRTOSSystemStateMetadata>().map_or(0, |x| x.trace_length);
+            let m = self.get_update_trace_length(state,l);
+            state.rand_mut().below(m) > l as u64 
+        } && state.rand_mut().below(100) < self.skip_non_favored_prob
+        {
+            idx = self.base.next(state)?;
+        }
+        Ok(idx)
+    }
+}
+
+impl<CS> LongestTraceScheduler<CS>
+where
+    CS: Scheduler,
+    CS::State: HasCorpus + HasMetadata + HasRand,
+{
+    pub fn get_update_trace_length(&self, state: &mut CS::State, par: usize) -> u64 {
+        // Create a new top rated meta if not existing
+        if let Some(td) = state.metadata_mut().get_mut::<LongestTracesMetadata>() {
+            let m = max(td.max_trace_length, par);
+            td.max_trace_length = m;
+            m as u64
+        } else {
+            state.add_metadata(LongestTracesMetadata::new(par));
+            par as u64
+        }
+    }
+    pub fn new(base: CS) -> Self {
+        Self {
+            base,
+            skip_non_favored_prob: DEFAULT_SKIP_NON_FAVORED_PROB,
+        }
+    }
+}
+
+//==========================================================================================
+
+/// A state metadata holding a map of favoreds testcases for each map entry
+#[derive(Debug, Serialize, Deserialize, SerdeAny, Default)]
+pub struct GeneticMetadata {
+    pub current_gen: Vec<(usize, f64)>,
+    pub current_cursor: usize,
+    pub next_gen: Vec<(usize, f64)>,
+    pub gen: usize
+}
+
+impl GeneticMetadata {
+    fn new(current_gen: Vec<(usize, f64)>, next_gen: Vec<(usize, f64)>) -> Self {
+        Self {current_gen, current_cursor: 0, next_gen, gen: 0}
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct GenerationScheduler<S> {
+    phantom: PhantomData<S>,
+    gen_size: usize,
+}
+
+impl<S> UsesState for GenerationScheduler<S>
+where
+    S: UsesInput,
+{
+    type State = S;
+}
+
+impl<S> Scheduler for GenerationScheduler<S>
+where
+    S: HasCorpus + HasMetadata,
+    S::Input: HasLen,
+{
+    /// get first element in current gen,
+    /// if current_gen is empty, swap lists, sort by FavFactor, take top k and return first
+    fn next(&self, state: &mut Self::State) -> Result<usize, Error> {
+        let mut to_remove : Vec<(usize, f64)> = vec![];
+        let mut to_return : usize = 0;
+        let c = state.corpus().count();
+        let gm = state.metadata_mut().get_mut::<GeneticMetadata>().expect("Corpus Scheduler empty");
+        // println!("index: {} curr: {:?} next: {:?} gen: {} corp: {}", gm.current_cursor, gm.current_gen.len(), gm.next_gen.len(), gm.gen,
+        // c);
+        match gm.current_gen.get(gm.current_cursor) {
+            Some(c) => {
+                gm.current_cursor+=1;
+                // println!("normal next: {}", (*c).0);
+                return Ok((*c).0)
+            },
+            None => {
+                swap(&mut to_remove, &mut gm.current_gen);
+                swap(&mut gm.next_gen, &mut gm.current_gen);
+                gm.current_gen.sort_by(|a, b| a.1.partial_cmp(&b.1).unwrap());
+                // gm.current_gen.reverse();
+                if gm.current_gen.len() == 0 {panic!("Corpus is empty");}
+                let d : Vec<(usize, f64)> = gm.current_gen.drain(min(gm.current_gen.len(), self.gen_size)..).collect();
+                to_remove.extend(d);
+                // move all indices to the left, since all other indices will be deleted
+                gm.current_gen.sort_by(|a,b| a.0.cmp(&(*b).0)); // in order of the corpus index
+                for i in 0..gm.current_gen.len() {
+                    gm.current_gen[i] = (i, gm.current_gen[i].1);
+                }
+                to_return = gm.current_gen.get(0).unwrap().0;
+                gm.current_cursor=1;
+                gm.gen+=1;
+            }
+        };
+        // removing these elements will move all indices left by to_remove.len()
+        to_remove.sort_by(|x,y| x.0.cmp(&(*y).0));
+        to_remove.reverse();
+        for i in to_remove {
+            state.corpus_mut().remove(i.0).unwrap();
+        }
+        // println!("switch next: {to_return}");
+        return Ok(to_return);
+    }
+
+    /// Add the new input to the next generation
+    fn on_add(
+        &self,
+        state: &mut Self::State,
+        idx: usize
+    ) -> Result<(), Error> {
+        // println!("On Add {idx}");
+        let mut tc = state.corpus_mut().get(idx).unwrap().borrow_mut().clone();
+        let ff = MaxTimeFavFactor::compute(&mut tc, state).unwrap();
+        if let Some(gm) = state.metadata_mut().get_mut::<GeneticMetadata>() {
+            gm.next_gen.push((idx,ff));
+        } else {
+            state.add_metadata(GeneticMetadata::new(vec![], vec![(idx,ff)]));
+        }
+        Ok(())
+    }
+    fn on_replace(
+        &self,
+        _state: &mut Self::State,
+        _idx: usize,
+        _prev: &Testcase<<Self::State as UsesInput>::Input>
+    ) -> Result<(), Error> {
+        // println!("On Replace {_idx}");
+        Ok(())
+    }
+
+    fn on_remove(
+        &self,
+        state: &mut Self::State,
+        idx: usize,
+        _testcase: &Option<Testcase<<Self::State as UsesInput>::Input>>
+    ) -> Result<(), Error> {
+        // println!("On Remove {idx}");
+        if let Some(gm) = state.metadata_mut().get_mut::<GeneticMetadata>() {
+            gm.next_gen = gm.next_gen.drain(..).into_iter().filter(|x| (*x).0 != idx).collect::<Vec<(usize, f64)>>();
+            gm.current_gen = gm.current_gen.drain(..).into_iter().filter(|x| (*x).0 != idx).collect::<Vec<(usize, f64)>>();
+        } else {
+            state.add_metadata(GeneticMetadata::new(vec![], vec![]));
+        }
+        Ok(())
+    }
+}
+
+impl<S> GenerationScheduler<S>
+{
+    pub fn new() -> Self {
+        Self {
+            phantom: PhantomData,
+            gen_size: 100,
+        }
+    }
+}

fuzzers/FRET/src/worst.rs

@@ -0,0 +1,381 @@
+use core::fmt::Debug;
+use core::cmp::Ordering::{Greater,Less,Equal};
+use libafl::inputs::BytesInput;
+use libafl::inputs::HasTargetBytes;
+use libafl::feedbacks::MapIndexesMetadata;
+use libafl::corpus::Testcase;
+use libafl::prelude::{UsesInput, AsSlice};
+use core::marker::PhantomData;
+use libafl::schedulers::{MinimizerScheduler, TestcaseScore};
+use std::path::PathBuf;
+use std::fs;
+use hashbrown::{HashMap};
+use libafl::observers::ObserversTuple;
+use libafl::executors::ExitKind;
+use libafl::events::EventFirer;
+use libafl::state::{HasClientPerfMonitor, HasCorpus, UsesState};
+use libafl::inputs::Input;
+use libafl::feedbacks::Feedback;
+use libafl::state::HasMetadata;
+use libafl_qemu::edges::QemuEdgesMapMetadata;
+use libafl::observers::MapObserver;
+use serde::{Deserialize, Serialize};
+use std::cmp;
+
+use libafl::{
+    bolts::{
+        tuples::Named,
+        HasLen,
+    },
+    observers::Observer,
+    Error,
+};
+
+use crate::clock::QemuClockObserver;
+use crate::systemstate::FreeRTOSSystemStateMetadata;
+//=========================== Scheduler
+
+pub type TimeMaximizerCorpusScheduler<CS> =
+    MinimizerScheduler<CS, MaxTimeFavFactor<<CS as UsesState>::State>, MapIndexesMetadata>;
+
+/// Multiply the testcase size with the execution time.
+/// This favors small and quick testcases.
+#[derive(Debug, Clone)]
+pub struct MaxTimeFavFactor<S>
+where
+    S: HasCorpus + HasMetadata,
+    S::Input: HasLen,
+{
+    phantom: PhantomData<S>,
+}
+
+impl<S> TestcaseScore<S> for MaxTimeFavFactor<S>
+where
+    S: HasCorpus + HasMetadata,
+    S::Input: HasLen,
+{
+    fn compute(entry: &mut Testcase<<S as UsesInput>::Input>, state: &S) -> Result<f64, Error> {
+        // TODO maybe enforce entry.exec_time().is_some()
+        let et = entry.exec_time().expect("testcase.exec_time is needed for scheduler");
+        let tns : i64 = et.as_nanos().try_into().expect("failed to convert time");
+        Ok(-tns as f64)
+    }
+}
+
+pub type LenTimeMaximizerCorpusScheduler<CS> =
+    MinimizerScheduler<CS, MaxExecsLenFavFactor<<CS as UsesState>::State>, MapIndexesMetadata>;
+
+pub type TimeStateMaximizerCorpusScheduler<CS> =
+    MinimizerScheduler<CS, MaxTimeFavFactor<<CS as UsesState>::State>, FreeRTOSSystemStateMetadata>;
+
+/// Multiply the testcase size with the execution time.
+/// This favors small and quick testcases.
+#[derive(Debug, Clone)]
+pub struct MaxExecsLenFavFactor<S>
+where
+    S: HasCorpus + HasMetadata,
+    S::Input: HasLen,
+{
+    phantom: PhantomData<S>,
+}
+
+impl<S> TestcaseScore<S> for MaxExecsLenFavFactor<S>
+where
+    S: HasCorpus + HasMetadata,
+    S::Input: HasLen,
+{
+    fn compute(entry: &mut Testcase<S::Input>, state: &S) -> Result<f64, Error> {
+        let execs_per_hour = (3600.0/entry.exec_time().expect("testcase.exec_time is needed for scheduler").as_secs_f64());
+        let execs_times_length_per_hour = execs_per_hour*entry.cached_len()? as f64;
+        Ok(execs_times_length_per_hour)
+    }
+}
+
+//===================================================================
+
+/// A Feedback reporting if the Input consists of strictly decreasing bytes.
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct SortedFeedback {
+}
+
+impl<S> Feedback<S> for SortedFeedback
+where
+    S: UsesInput + HasClientPerfMonitor,
+    S::Input: HasTargetBytes,
+{
+    #[allow(clippy::wrong_self_convention)]
+    fn is_interesting<EM, OT>(
+        &mut self,
+        _state: &mut S,
+        _manager: &mut EM,
+        _input: &S::Input,
+        observers: &OT,
+        _exit_kind: &ExitKind,
+    ) -> Result<bool, Error>
+    where
+        EM: EventFirer<State = S>,
+        OT: ObserversTuple<S>,
+    {
+        let t = _input.target_bytes();
+        let tmp = t.as_slice();
+        if tmp.len()<32 {return Ok(false);}
+        let tmp = Vec::<u8>::from(&tmp[0..32]);
+        // tmp.reverse();
+        if tmp.is_sorted_by(|a,b| match a.partial_cmp(b).unwrap_or(Less) {
+            Less => Some(Greater),
+            Equal => Some(Greater),
+            Greater => Some(Less),
+        }) {return Ok(true)};
+        return Ok(false);
+    }
+}
+
+impl Named for SortedFeedback {
+    #[inline]
+    fn name(&self) -> &str {
+        "Sorted"
+    }
+}
+
+impl SortedFeedback {
+    /// Creates a new [`HitFeedback`]
+    #[must_use]
+    pub fn new() -> Self {
+        Self {}
+    }
+}
+
+impl Default for SortedFeedback {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+//===================================================================
+/// A Feedback which expects a certain minimum execution time
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct ExecTimeReachedFeedback
+{
+    target_time: u64,
+}
+
+impl<S> Feedback<S> for ExecTimeReachedFeedback
+where
+    S: UsesInput + HasClientPerfMonitor,
+{
+    #[allow(clippy::wrong_self_convention)]
+    fn is_interesting<EM, OT>(
+        &mut self,
+        _state: &mut S,
+        _manager: &mut EM,
+        _input: &S::Input,
+        observers: &OT,
+        _exit_kind: &ExitKind,
+    ) -> Result<bool, Error>
+    where
+        EM: EventFirer<State = S>,
+        OT: ObserversTuple<S>,
+    {
+        let observer = observers.match_name::<QemuClockObserver>("clock")
+            .expect("QemuClockObserver not found");
+        Ok(observer.last_runtime() >= self.target_time)
+    }
+}
+
+impl Named for ExecTimeReachedFeedback
+{
+    #[inline]
+    fn name(&self) -> &str {
+        "ExecTimeReachedFeedback"
+    }
+}
+
+impl ExecTimeReachedFeedback
+where
+{
+    /// Creates a new [`ExecTimeReachedFeedback`]
+    #[must_use]
+    pub fn new(target_time : u64) -> Self {
+        Self {target_time: target_time}
+    }
+}
+
+pub static mut EXEC_TIME_COLLECTION : Vec<u32> = Vec::new();
+
+/// A Noop Feedback which records a list of all execution times
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct ExecTimeCollectorFeedback
+{
+}
+
+impl<S> Feedback<S> for ExecTimeCollectorFeedback
+where
+    S: UsesInput + HasClientPerfMonitor,
+{
+    #[allow(clippy::wrong_self_convention)]
+    fn is_interesting<EM, OT>(
+        &mut self,
+        _state: &mut S,
+        _manager: &mut EM,
+        _input: &S::Input,
+        observers: &OT,
+        _exit_kind: &ExitKind,
+    ) -> Result<bool, Error>
+    where
+        EM: EventFirer<State = S>,
+        OT: ObserversTuple<S>,
+    {
+        let observer = observers.match_name::<QemuClockObserver>("clock")
+            .expect("QemuClockObserver not found");
+        unsafe { EXEC_TIME_COLLECTION.push(observer.last_runtime().try_into().unwrap()); }
+        Ok(false)
+    }
+}
+
+impl Named for ExecTimeCollectorFeedback
+{
+    #[inline]
+    fn name(&self) -> &str {
+        "ExecTimeCollectorFeedback"
+    }
+}
+
+impl ExecTimeCollectorFeedback
+where
+{
+    /// Creates a new [`ExecTimeCollectorFeedback`]
+    #[must_use]
+    pub fn new() -> Self {
+        Self {}
+    }
+}
+
+/// Shared Metadata for a SysStateFeedback
+#[derive(Serialize, Deserialize, Clone, Debug, Default)]
+pub struct ExecTimeCollectorFeedbackState
+{
+    collection: Vec<u32>,
+}
+impl Named for ExecTimeCollectorFeedbackState
+{
+    #[inline]
+    fn name(&self) -> &str {
+        "ExecTimeCollectorFeedbackState"
+    }
+}
+
+//===================================================================
+/// A Feedback which expects a certain minimum execution time
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct ExecTimeIncFeedback
+{
+    longest_time: u64,
+    last_is_longest: bool
+}
+
+impl<S> Feedback<S> for ExecTimeIncFeedback
+where
+    S: UsesInput + HasClientPerfMonitor,
+{
+    #[allow(clippy::wrong_self_convention)]
+    fn is_interesting<EM, OT>(
+        &mut self,
+        _state: &mut S,
+        _manager: &mut EM,
+        _input: &S::Input,
+        observers: &OT,
+        _exit_kind: &ExitKind,
+    ) -> Result<bool, Error>
+    where
+        EM: EventFirer<State = S>,
+        OT: ObserversTuple<S>,
+    {
+        let observer = observers.match_name::<QemuClockObserver>("clocktime")
+            .expect("QemuClockObserver not found");
+        if observer.last_runtime() > self.longest_time {
+            self.longest_time = observer.last_runtime();
+            self.last_is_longest = true;
+            Ok(true)
+        } else {
+            self.last_is_longest = false;
+            Ok(false)
+        }
+    }
+    fn append_metadata(
+        &mut self,
+        _state: &mut S,
+        testcase: &mut Testcase<<S as UsesInput>::Input>,
+    ) -> Result<(), Error> {
+        #[cfg(feature = "feed_afl")]
+        if self.last_is_longest {
+            let mim : Option<&mut MapIndexesMetadata>= testcase.metadata_mut().get_mut();
+            // pretend that the longest input alone excercises some non-existing edge, to keep it relevant
+            mim.unwrap().list.push(usize::MAX);
+        };
+        Ok(())
+    }
+}
+
+impl Named for ExecTimeIncFeedback
+{
+    #[inline]
+    fn name(&self) -> &str {
+        "ExecTimeReachedFeedback"
+    }
+}
+
+impl ExecTimeIncFeedback
+where
+{
+    /// Creates a new [`ExecTimeReachedFeedback`]
+    #[must_use]
+    pub fn new() -> Self {
+        Self {longest_time: 0, last_is_longest: false}
+    }
+}
+
+/// A Noop Feedback which records a list of all execution times
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct AlwaysTrueFeedback
+{
+}
+
+impl<S> Feedback<S> for AlwaysTrueFeedback
+where
+    S: UsesInput + HasClientPerfMonitor,
+{
+    #[allow(clippy::wrong_self_convention)]
+    fn is_interesting<EM, OT>(
+        &mut self,
+        _state: &mut S,
+        _manager: &mut EM,
+        _input: &S::Input,
+        _observers: &OT,
+        _exit_kind: &ExitKind,
+    ) -> Result<bool, Error>
+    where
+        EM: EventFirer<State = S>,
+        OT: ObserversTuple<S>,
+    {
+        Ok(true)
+    }
+}
+
+impl Named for AlwaysTrueFeedback
+{
+    #[inline]
+    fn name(&self) -> &str {
+        "AlwaysTrueFeedback"
+    }
+}
+
+impl AlwaysTrueFeedback
+where
+{
+    /// Creates a new [`ExecTimeCollectorFeedback`]
+    #[must_use]
+    pub fn new() -> Self {
+        Self {
+        }
+    }
+}

fuzzers/baby_fuzzer/src/main.rs

@@ -1,5 +1,4 @@
 use std::path::PathBuf;
-
 #[cfg(windows)]
 use std::ptr::write_volatile;
 

fuzzers/baby_fuzzer_gramatron/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "baby_fuzzer_gramatron"
-version = "0.8.0"
+version = "0.8.2"
 authors = ["Andrea Fioraldi <andreafioraldi@gmail.com>", "Dominik Maier <domenukk@gmail.com>"]
 edition = "2021"
 

fuzzers/baby_fuzzer_gramatron/src/main.rs

@@ -1,12 +1,10 @@
-use std::io::Read;
-use std::{
-    fs,
-    io::BufReader,
-    path::{Path, PathBuf},
-};
-
 #[cfg(windows)]
 use std::ptr::write_volatile;
+use std::{
+    fs,
+    io::{BufReader, Read},
+    path::{Path, PathBuf},
+};
 
 use libafl::{
     bolts::{current_nanos, rands::StdRand, tuples::tuple_list},

fuzzers/baby_fuzzer_grimoire/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "baby_fuzzer_grimoire"
-version = "0.8.0"
+version = "0.8.2"
 authors = ["Andrea Fioraldi <andreafioraldi@gmail.com>", "Dominik Maier <domenukk@gmail.com>"]
 edition = "2021"
 

fuzzers/baby_fuzzer_grimoire/src/main.rs

@@ -1,8 +1,6 @@
-use std::io::Read;
-use std::{fs, path::PathBuf};
-
 #[cfg(windows)]
 use std::ptr::write_volatile;
+use std::{fs, io::Read, path::PathBuf};
 
 use libafl::{
     bolts::{current_nanos, rands::StdRand, tuples::tuple_list, AsSlice},

fuzzers/baby_fuzzer_minimizing/.gitignore

@@ -0,0 +1,3 @@
+corpus
+minimized
+solutions

fuzzers/baby_fuzzer_minimizing/Cargo.toml

@@ -0,0 +1,23 @@
+[package]
+name = "baby_fuzzer_minimizing"
+version = "0.8.2"
+authors = ["Andrea Fioraldi <andreafioraldi@gmail.com>", "Dominik Maier <domenukk@gmail.com>", "Addison Crump <research@addisoncrump.info>"]
+edition = "2021"
+
+[features]
+default = ["std"]
+tui = []
+std = []
+
+[profile.dev]
+panic = "abort"
+
+[profile.release]
+panic = "abort"
+lto = true
+codegen-units = 1
+opt-level = 3
+debug = true
+
+[dependencies]
+libafl = { path = "../../libafl/", features = ["prelude"] }

fuzzers/baby_fuzzer_minimizing/README.md

@@ -0,0 +1,9 @@
+# Baby fuzzer
+
+This is a minimalistic example about how to create a libafl based fuzzer which leverages minimisation.
+
+The fuzzer steps until a crash occurs, minimising each corpus entry as it is discovered. Then, once a
+solution is found, it attempts to minimise that as well.
+
+The tested program is a simple Rust function without any instrumentation.
+For real fuzzing, you will want to add some sort to add coverage or other feedback.

fuzzers/baby_fuzzer_minimizing/src/main.rs

@@ -0,0 +1,142 @@
+use std::path::PathBuf;
+#[cfg(windows)]
+use std::ptr::write_volatile;
+
+use libafl::prelude::*;
+
+/// Coverage map with explicit assignments due to the lack of instrumentation
+static mut SIGNALS: [u8; 16] = [0; 16];
+
+/// Assign a signal to the signals map
+fn signals_set(idx: usize) {
+    unsafe { SIGNALS[idx] = 1 };
+}
+
+#[allow(clippy::similar_names)]
+pub fn main() -> Result<(), Error> {
+    // The closure that we want to fuzz
+    let mut harness = |input: &BytesInput| {
+        let target = input.target_bytes();
+        let buf = target.as_slice();
+        signals_set(0);
+        if !buf.is_empty() && buf[0] == b'a' {
+            signals_set(1);
+            if buf.len() > 1 && buf[1] == b'b' {
+                signals_set(2);
+                if buf.len() > 2 && buf[2] == b'c' {
+                    return ExitKind::Crash;
+                }
+            }
+        }
+        ExitKind::Ok
+    };
+
+    // Create an observation channel using the signals map
+    let observer =
+        unsafe { StdMapObserver::new_from_ptr("signals", SIGNALS.as_mut_ptr(), SIGNALS.len()) };
+
+    let factory = MapEqualityFactory::new_from_observer(&observer);
+
+    // Feedback to rate the interestingness of an input
+    let mut feedback = MaxMapFeedback::new(&observer);
+
+    // A feedback to choose if an input is a solution or not
+    let mut objective = CrashFeedback::new();
+
+    // The Monitor trait define how the fuzzer stats are displayed to the user
+    let mon = SimpleMonitor::new(|s| println!("{}", s));
+
+    let mut mgr = SimpleEventManager::new(mon);
+
+    let corpus_dir = PathBuf::from("./corpus");
+    let solution_dir = PathBuf::from("./solutions");
+
+    // create a State from scratch
+    let mut state = StdState::new(
+        // RNG
+        StdRand::with_seed(current_nanos()),
+        // Corpus that will be evolved, we keep it in memory for performance
+        OnDiskCorpus::new(&corpus_dir).unwrap(),
+        // Corpus in which we store solutions (crashes in this example),
+        // on disk so the user can get them after stopping the fuzzer
+        OnDiskCorpus::new(&solution_dir).unwrap(),
+        // States of the feedbacks.
+        // The feedbacks can report the data that should persist in the State.
+        &mut feedback,
+        // Same for objective feedbacks
+        &mut objective,
+    )
+    .unwrap();
+
+    // A queue policy to get testcasess from the corpus
+    let scheduler = QueueScheduler::new();
+
+    // A fuzzer with feedbacks and a corpus scheduler
+    let mut fuzzer = StdFuzzer::new(scheduler, feedback, objective);
+
+    // Create the executor for an in-process function with just one observer
+    let mut executor = InProcessExecutor::new(
+        &mut harness,
+        tuple_list!(observer),
+        &mut fuzzer,
+        &mut state,
+        &mut mgr,
+    )
+    .expect("Failed to create the Executor");
+
+    // Generator of printable bytearrays of max size 32
+    let mut generator = RandPrintablesGenerator::new(32);
+
+    // Generate 8 initial inputs
+    state
+        .generate_initial_inputs(&mut fuzzer, &mut executor, &mut generator, &mut mgr, 8)
+        .expect("Failed to generate the initial corpus");
+
+    // Setup a mutational stage with a basic bytes mutator
+    let mutator = StdScheduledMutator::new(havoc_mutations());
+    let minimizer = StdScheduledMutator::new(havoc_mutations());
+    let mut stages = tuple_list!(
+        StdMutationalStage::new(mutator),
+        StdTMinMutationalStage::new(minimizer, factory, 128)
+    );
+
+    while state.solutions().is_empty() {
+        fuzzer.fuzz_one(&mut stages, &mut executor, &mut state, &mut mgr)?;
+    }
+
+    let minimized_dir = PathBuf::from("./minimized");
+
+    let mut state = StdState::new(
+        StdRand::with_seed(current_nanos()),
+        OnDiskCorpus::new(&minimized_dir).unwrap(),
+        InMemoryCorpus::new(),
+        &mut (),
+        &mut (),
+    )
+    .unwrap();
+
+    // The Monitor trait define how the fuzzer stats are displayed to the user
+    let mon = SimpleMonitor::new(|s| println!("{}", s));
+
+    let mut mgr = SimpleEventManager::new(mon);
+
+    let minimizer = StdScheduledMutator::new(havoc_mutations());
+    let mut stages = tuple_list!(StdTMinMutationalStage::new(
+        minimizer,
+        CrashFeedbackFactory::default(),
+        1 << 10
+    ));
+
+    let scheduler = QueueScheduler::new();
+
+    // A fuzzer with feedbacks and a corpus scheduler
+    let mut fuzzer = StdFuzzer::new(scheduler, (), ());
+
+    // Create the executor for an in-process function with just one observer
+    let mut executor = InProcessExecutor::new(&mut harness, (), &mut fuzzer, &mut state, &mut mgr)?;
+
+    state.load_initial_inputs_forced(&mut fuzzer, &mut executor, &mut mgr, &[solution_dir])?;
+    stages.perform_all(&mut fuzzer, &mut executor, &mut state, &mut mgr, 0)?;
+
+    Ok(())
+}

fuzzers/baby_fuzzer_nautilus/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "baby_fuzzer_nautilus"
-version = "0.8.0"
+version = "0.8.2"
 authors = ["Andrea Fioraldi <andreafioraldi@gmail.com>", "Dominik Maier <domenukk@gmail.com>"]
 edition = "2018"
 

fuzzers/baby_fuzzer_nautilus/src/main.rs

@@ -1,5 +1,4 @@
 use std::path::PathBuf;
-
 #[cfg(windows)]
 use std::ptr::write_volatile;
 

fuzzers/baby_fuzzer_swap_differential/.gitignore

@@ -0,0 +1 @@
+libpng-*

fuzzers/baby_fuzzer_swap_differential/Cargo.toml

@@ -0,0 +1,40 @@
+[package]
+name = "baby_fuzzer_swap_differential"
+version = "0.7.1"
+authors = ["Addison Crump <research@addisoncrump.info>"]
+edition = "2021"
+default-run = "fuzzer_sd"
+
+[features]
+tui = []
+multimap = []
+
+[profile.dev]
+panic = "abort"
+
+[profile.release]
+panic = "abort"
+lto = true
+codegen-units = 1
+opt-level = 3
+debug = true
+
+[build-dependencies]
+anyhow = "1"
+bindgen = "0.61"
+cc = "1.0"
+
+[dependencies]
+libafl = { path = "../../libafl" }
+libafl_targets = { path = "../../libafl_targets", features = ["sancov_pcguard_hitcounts", "libfuzzer", "sancov_cmplog", "pointer_maps"] }
+mimalloc = { version = "*", default-features = false }
+
+libafl_cc = { path = "../../libafl_cc/" }
+
+[[bin]]
+name = "fuzzer_sd"
+path = "src/main.rs"
+
+[[bin]]
+name = "libafl_cc"
+path = "src/bin/libafl_cc.rs"

fuzzers/baby_fuzzer_swap_differential/Makefile.toml

@@ -0,0 +1,45 @@
+# Variables
+[env]
+FUZZER_NAME='fuzzer_sd'
+CARGO_TARGET_DIR = { value = "target", condition = { env_not_set = ["CARGO_TARGET_DIR"] } }
+LIBAFL_CC = '${CARGO_TARGET_DIR}/release/libafl_cc'
+FUZZER = '${CARGO_TARGET_DIR}/release/${FUZZER_NAME}'
+PROJECT_DIR = { script = ["pwd"] }
+
+# Compilers
+[tasks.cc]
+command = "cargo"
+args = ["build" , "--release", "--bin", "libafl_cc"]
+
+# Harness
+[tasks.fuzzer]
+command = "cargo"
+args = ["build" , "--release", "--bin", "${FUZZER_NAME}"]
+dependencies = [ "cc" ]
+
+# Run the fuzzer
+[tasks.run]
+command = "${CARGO_TARGET_DIR}/release/${FUZZER_NAME}"
+dependencies = [ "fuzzer" ]
+
+# Test
+[tasks.test]
+linux_alias = "test_unix"
+mac_alias = "test_unix"
+windows_alias = "unsupported"
+
+[tasks.test_unix]
+script_runner = "@shell"
+script='''
+timeout 10s ${CARGO_TARGET_DIR}/release/${FUZZER_NAME}
+'''
+dependencies = [ "fuzzer" ]
+
+# Clean up
+[tasks.clean]
+# Disable default `clean` definition
+clear = true
+script_runner="@shell"
+script='''
+cargo clean
+'''

fuzzers/baby_fuzzer_swap_differential/README.md

@@ -0,0 +1,11 @@
+# Baby fuzzer (swap differential)
+
+This is a minimalistic example about how to create a libafl-based differential fuzzer which swaps out the AFL map during
+execution so that both maps may be measured.
+
+It runs on a single core until an input is discovered which both inputs accept.
+
+The tested programs are provided in `first.c` and `second.c`.
+
+You may execute this fuzzer with `cargo make run`. If you prefer to do so manually, you may also simply use
+`cargo build --release --bin libafl_cc` followed by `cargo run --release --bin fuzzer_sd`

fuzzers/baby_fuzzer_swap_differential/build.rs

@@ -0,0 +1,45 @@
+use std::{env, path::PathBuf, str::FromStr};
+
+fn main() -> anyhow::Result<()> {
+    if env::var("CARGO_BIN_NAME").map_or(true, |v| v != "libafl_cc") {
+        println!("cargo:rerun-if-changed=./first.h");
+        println!("cargo:rerun-if-changed=./first.c");
+        println!("cargo:rerun-if-changed=./second.h");
+        println!("cargo:rerun-if-changed=./second.c");
+        println!("cargo:rerun-if-changed=./common.c");
+
+        // Configure and generate bindings.
+        let bindings = bindgen::builder()
+            .header("first.h")
+            .header("second.h")
+            .parse_callbacks(Box::new(bindgen::CargoCallbacks))
+            .generate()?;
+
+        // Write the generated bindings to an output file.
+        let out_path = PathBuf::from(env::var("OUT_DIR").unwrap());
+        bindings
+            .write_to_file(out_path.join("bindings.rs"))
+            .expect("Couldn't write bindings!");
+
+        let compiler = env::var("CARGO_TARGET_DIR")
+            .map_or(PathBuf::from_str("target").unwrap(), |v| {
+                PathBuf::from_str(&v).unwrap()
+            })
+            .join("release/libafl_cc");
+        println!("cargo:rerun-if-changed={}", compiler.to_str().unwrap());
+        if !compiler.try_exists().unwrap_or(false) {
+            println!("cargo:warning=Can't find libafl_cc; assuming that we're building it.");
+        } else {
+            cc::Build::new()
+                .compiler(compiler)
+                .file("first.c")
+                .file("second.c")
+                .file("common.c")
+                .compile("diff-target");
+
+            println!("cargo:rustc-link-lib=diff-target");
+        }
+    }
+
+    Ok(())
+}

fuzzers/baby_fuzzer_swap_differential/common.c

@@ -0,0 +1,12 @@
+#include "common.h"
+
+bool both_require(const uint8_t *bytes, size_t len) {
+  if (len >= 1 && bytes[0] == 'a') {
+    if (len >= 2 && bytes[1] == 'b') {
+      if (len >= 3 && bytes[2] == 'c') {
+        return ACCEPT;
+      }
+    }
+  }
+  return REJECT;
+}

fuzzers/baby_fuzzer_swap_differential/common.h

@@ -0,0 +1,13 @@
+#ifndef LIBAFL_COMMON_H
+#define LIBAFL_COMMON_H
+
+#include <stdbool.h>
+#include <stddef.h>
+#include <stdint.h>
+
+#define ACCEPT true
+#define REJECT false
+
+bool both_require(const uint8_t *bytes, size_t len);
+
+#endif  // LIBAFL_COMMON_H

fuzzers/baby_fuzzer_swap_differential/first.c

@@ -0,0 +1,10 @@
+#include "first.h"
+
+bool inspect_first(const uint8_t *bytes, size_t len) {
+  if (both_require(bytes, len)) {
+    if (len >= 4 && bytes[3] == 'd') {
+      return ACCEPT;
+    }
+  }
+  return REJECT;
+}

fuzzers/baby_fuzzer_swap_differential/first.h

@@ -0,0 +1,8 @@
+#ifndef LIBAFL_FIRST_H
+#define LIBAFL_FIRST_H
+
+#include "common.h"
+
+bool inspect_first(const uint8_t *bytes, size_t len);
+
+#endif  // LIBAFL_FIRST_H

fuzzers/baby_fuzzer_swap_differential/second.c

@@ -0,0 +1,10 @@
+#include "second.h"
+
+bool inspect_second(const uint8_t *bytes, size_t len) {
+  if (both_require(bytes, len)) {
+    if (len >= 5 && bytes[4] == 'e') {
+      return ACCEPT;
+    }
+  }
+  return REJECT;
+}

fuzzers/baby_fuzzer_swap_differential/second.h

@@ -0,0 +1,8 @@
+#ifndef LIBAFL_SECOND_H
+#define LIBAFL_SECOND_H
+
+#include "common.h"
+
+bool inspect_second(const uint8_t *bytes, size_t len);
+
+#endif  // LIBAFL_SECOND_H

fuzzers/baby_fuzzer_swap_differential/src/bin/libafl_cc.rs

@@ -0,0 +1,35 @@
+use std::env;
+
+use libafl_cc::{ClangWrapper, CompilerWrapper};
+
+pub fn main() {
+    let args: Vec<String> = env::args().collect();
+    if args.len() > 1 {
+        let mut dir = env::current_exe().unwrap();
+        let wrapper_name = dir.file_name().unwrap().to_str().unwrap();
+
+        let is_cpp = match wrapper_name[wrapper_name.len()-2..].to_lowercase().as_str() {
+            "cc" => false,
+            "++" | "pp" | "xx" => true,
+            _ => panic!("Could not figure out if c or c++ wrapper was called. Expected {:?} to end with c or cxx", dir),
+        };
+
+        dir.pop();
+
+        let mut cc = ClangWrapper::new();
+        if let Some(code) = cc
+            .cpp(is_cpp)
+            // silence the compiler wrapper output, needed for some configure scripts.
+            .silence(true)
+            .parse_args(&args)
+            .expect("Failed to parse the command line")
+            .add_arg("-fsanitize-coverage=trace-pc-guard")
+            .run()
+            .expect("Failed to run the wrapped compiler")
+        {
+            std::process::exit(code);
+        }
+    } else {
+        panic!("LibAFL CC: No Arguments given");
+    }
+}

fuzzers/baby_fuzzer_swap_differential/src/main.rs

@@ -0,0 +1,246 @@
+#[cfg(windows)]
+use std::ptr::write_volatile;
+use std::{
+    alloc::{alloc_zeroed, Layout},
+    path::PathBuf,
+};
+
+#[cfg(feature = "tui")]
+use libafl::monitors::tui::TuiMonitor;
+#[cfg(not(feature = "tui"))]
+use libafl::monitors::SimpleMonitor;
+use libafl::{
+    bolts::{current_nanos, rands::StdRand, tuples::tuple_list, AsSlice},
+    corpus::{Corpus, InMemoryCorpus, OnDiskCorpus},
+    events::SimpleEventManager,
+    executors::{inprocess::InProcessExecutor, DiffExecutor, ExitKind},
+    feedbacks::{CrashFeedback, MaxMapFeedback},
+    fuzzer::{Fuzzer, StdFuzzer},
+    generators::RandPrintablesGenerator,
+    inputs::{BytesInput, HasTargetBytes},
+    mutators::scheduled::{havoc_mutations, StdScheduledMutator},
+    observers::StdMapObserver,
+    schedulers::QueueScheduler,
+    stages::mutational::StdMutationalStage,
+    state::{HasSolutions, StdState},
+};
+use libafl_targets::{DifferentialAFLMapSwapObserver, MAX_EDGES_NUM};
+use mimalloc::MiMalloc;
+
+#[global_allocator]
+static GLOBAL: MiMalloc = MiMalloc;
+
+// bindings to the functions defined in the target
+mod bindings {
+    #![allow(non_camel_case_types)]
+    #![allow(non_upper_case_globals)]
+    #![allow(unused)]
+
+    include!(concat!(env!("OUT_DIR"), "/bindings.rs"));
+}
+
+use bindings::{inspect_first, inspect_second};
+
+#[cfg(feature = "multimap")]
+mod multimap {
+    pub use libafl::observers::{HitcountsIterableMapObserver, MultiMapObserver};
+
+    pub static mut FIRST_EDGES: &'static mut [u8] = &mut [];
+    pub static mut SECOND_EDGES: &'static mut [u8] = &mut [];
+    pub static mut COMBINED_EDGES: [&'static mut [u8]; 2] = [&mut [], &mut []];
+}
+#[cfg(feature = "multimap")]
+use multimap::*;
+
+#[cfg(not(feature = "multimap"))]
+mod slicemap {
+    pub use libafl::observers::HitcountsMapObserver;
+
+    pub static mut EDGES: &'static mut [u8] = &mut [];
+}
+#[cfg(not(feature = "multimap"))]
+use slicemap::*;
+
+#[allow(clippy::similar_names)]
+pub fn main() {
+    // The closure that we want to fuzz
+    let mut first_harness = |input: &BytesInput| {
+        let target = input.target_bytes();
+        let buf = target.as_slice();
+        if unsafe { inspect_first(buf.as_ptr(), buf.len()) } {
+            ExitKind::Crash
+        } else {
+            ExitKind::Ok
+        }
+    };
+    let mut second_harness = |input: &BytesInput| {
+        let target = input.target_bytes();
+        let buf = target.as_slice();
+        if unsafe { inspect_second(buf.as_ptr(), buf.len()) } {
+            ExitKind::Crash
+        } else {
+            ExitKind::Ok
+        }
+    };
+
+    #[cfg(feature = "multimap")]
+    let (first_map_observer, second_map_observer, map_swapper, map_observer) = {
+        // initialize the maps
+        unsafe {
+            let layout = Layout::from_size_align(MAX_EDGES_NUM, 64).unwrap();
+            FIRST_EDGES = core::slice::from_raw_parts_mut(alloc_zeroed(layout), MAX_EDGES_NUM);
+            SECOND_EDGES = core::slice::from_raw_parts_mut(alloc_zeroed(layout), MAX_EDGES_NUM);
+            COMBINED_EDGES = [&mut FIRST_EDGES, &mut SECOND_EDGES];
+        }
+
+        // create the base maps used to observe the different executors from two independent maps
+        let mut first_map_observer = StdMapObserver::new("first-edges", unsafe { FIRST_EDGES });
+        let mut second_map_observer = StdMapObserver::new("second-edges", unsafe { SECOND_EDGES });
+
+        // create a map swapper so that we can replace the coverage map pointer (requires feature pointer_maps!)
+        let map_swapper =
+            DifferentialAFLMapSwapObserver::new(&mut first_map_observer, &mut second_map_observer);
+
+        // create a combined map observer, e.g. for calibration
+        // we use MultiMapObserver::differential to indicate that we want to use the observer in
+        // differential mode
+        let map_observer = HitcountsIterableMapObserver::new(MultiMapObserver::differential(
+            "combined-edges",
+            unsafe { &mut COMBINED_EDGES },
+        ));
+        (
+            first_map_observer,
+            second_map_observer,
+            map_swapper,
+            map_observer,
+        )
+    };
+    #[cfg(not(feature = "multimap"))]
+    let (first_map_observer, second_map_observer, map_swapper, map_observer) = {
+        // initialize the map
+        unsafe {
+            let layout = Layout::from_size_align(MAX_EDGES_NUM * 2, 64).unwrap();
+            EDGES = core::slice::from_raw_parts_mut(alloc_zeroed(layout), MAX_EDGES_NUM * 2);
+        }
+
+        // create the base maps used to observe the different executors by splitting a slice
+        let mut first_map_observer =
+            StdMapObserver::new("first-edges", unsafe { &mut EDGES[..MAX_EDGES_NUM] });
+        let mut second_map_observer =
+            StdMapObserver::new("second-edges", unsafe { &mut EDGES[MAX_EDGES_NUM..] });
+
+        // create a map swapper so that we can replace the coverage map pointer (requires feature pointer_maps!)
+        let map_swapper =
+            DifferentialAFLMapSwapObserver::new(&mut first_map_observer, &mut second_map_observer);
+
+        // create a combined map observer, e.g. for calibration
+        // we use StdMapObserver::differential to indicate that we want to use the observer in
+        // differential mode
+        let map_observer =
+            HitcountsMapObserver::new(StdMapObserver::differential("combined-edges", unsafe {
+                EDGES
+            }));
+        (
+            first_map_observer,
+            second_map_observer,
+            map_swapper,
+            map_observer,
+        )
+    };
+
+    // Feedback to rate the interestingness of an input
+    let mut feedback = MaxMapFeedback::new(&map_observer);
+
+    // A feedback to choose if an input is a solution or not
+    // Crash here means "both crashed", which is our objective
+    let mut objective = CrashFeedback::new();
+
+    // create a State from scratch
+    let mut state = StdState::new(
+        // RNG
+        StdRand::with_seed(current_nanos()),
+        // Corpus that will be evolved, we keep it in memory for performance
+        InMemoryCorpus::new(),
+        // Corpus in which we store solutions (crashes in this example),
+        // on disk so the user can get them after stopping the fuzzer
+        OnDiskCorpus::new(PathBuf::from("./crashes")).unwrap(),
+        // States of the feedbacks.
+        // The feedbacks can report the data that should persist in the State.
+        &mut feedback,
+        // Same for objective feedbacks
+        &mut objective,
+    )
+    .unwrap();
+
+    // The Monitor trait define how the fuzzer stats are displayed to the user
+    #[cfg(not(feature = "tui"))]
+    let mon = SimpleMonitor::new(|s| println!("{}", s));
+    #[cfg(feature = "tui")]
+    let mon = TuiMonitor::new(String::from("Baby Fuzzer"), false);
+
+    // The event manager handle the various events generated during the fuzzing loop
+    // such as the notification of the addition of a new item to the corpus
+    let mut mgr = SimpleEventManager::new(mon);
+
+    // A queue policy to get testcases from the corpus
+    let scheduler = QueueScheduler::new();
+
+    // A fuzzer with feedbacks and a corpus scheduler
+    let mut fuzzer = StdFuzzer::new(scheduler, feedback, objective);
+
+    // Create the executor for an in-process function with just one observer
+    let first_executor = InProcessExecutor::new(
+        &mut first_harness,
+        tuple_list!(first_map_observer),
+        &mut fuzzer,
+        &mut state,
+        &mut mgr,
+    )
+    .expect("Failed to create the first executor");
+    let second_executor = InProcessExecutor::new(
+        &mut second_harness,
+        tuple_list!(second_map_observer),
+        &mut fuzzer,
+        &mut state,
+        &mut mgr,
+    )
+    .expect("Failed to create the second executor");
+
+    // create the differential executor, providing both the map swapper (which will ensure the
+    // instrumentation picks the correct map to write to) and the map observer (which provides the
+    // combined feedback)
+    let mut differential_executor = DiffExecutor::new(
+        first_executor,
+        second_executor,
+        tuple_list!(map_swapper, map_observer),
+    );
+
+    // Generator of printable bytearrays of max size 32
+    let mut generator = RandPrintablesGenerator::new(32);
+
+    // Generate 8 initial inputs
+    state
+        .generate_initial_inputs(
+            &mut fuzzer,
+            &mut differential_executor,
+            &mut generator,
+            &mut mgr,
+            8,
+        )
+        .expect("Failed to generate the initial corpus");
+
+    // Setup a mutational stage with a basic bytes mutator
+    let mutator = StdScheduledMutator::new(havoc_mutations());
+    let mut stages = tuple_list!(StdMutationalStage::new(mutator));
+
+    while state.solutions().is_empty() {
+        fuzzer
+            .fuzz_one(
+                &mut stages,
+                &mut differential_executor,
+                &mut state,
+                &mut mgr,
+            )
+            .expect("Error in the fuzzing loop");
+    }
+}

fuzzers/baby_fuzzer_tokens/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "baby_fuzzer_tokens"
-version = "0.8.0"
+version = "0.8.2"
 authors = ["Andrea Fioraldi <andreafioraldi@gmail.com>", "Dominik Maier <domenukk@gmail.com>"]
 edition = "2021"
 

fuzzers/baby_fuzzer_tokens/src/main.rs

@@ -1,8 +1,6 @@
-use std::io::Read;
-use std::{fs, path::PathBuf};
-
 #[cfg(windows)]
 use std::ptr::write_volatile;
+use std::{fs, io::Read, path::PathBuf};
 
 use libafl::{
     bolts::{current_nanos, rands::StdRand, tuples::tuple_list},

fuzzers/baby_fuzzer_with_forkexecutor/src/main.rs

@@ -1,5 +1,4 @@
 use std::path::PathBuf;
-
 #[cfg(windows)]
 use std::ptr::write_volatile;
 

fuzzers/baby_no_std/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "baby_no_std"
-version = "0.8.0"
+version = "0.8.2"
 authors = ["Andrea Fioraldi <andreafioraldi@gmail.com>", "Dominik Maier <domenukk@gmail.com>"]
 edition = "2021"
 
@@ -20,5 +20,4 @@ static-alloc = "0.2.3"
 
 [target.'cfg(unix)'.dependencies]
 libc = "0.2"
-cstr_core = "0.2.3"
 

fuzzers/baby_no_std/Makefile.toml

@@ -0,0 +1,34 @@
+[env]
+FUZZER_NAME="fuzzer"
+PROJECT_DIR = { script = ["pwd"] }
+
+[tasks.unsupported]
+script_runner="@shell"
+script='''
+echo "Cargo-make not integrated yet on this"
+'''
+
+# Fuzzer
+[tasks.build]
+command = "cargo"
+args = ["build", "--release", "-Zbuild-std=core,alloc", "--target", "x86_64-unknown-linux-gnu"]
+
+# Test
+[tasks.test]
+linux_alias = "test_unix"
+mac_alias = "unsupported"
+windows_alias = "unsupported"
+
+[tasks.test_unix]
+script='''
+cargo run -Zbuild-std=core,alloc --target x86_64-unknown-linux-gnu || true
+'''
+dependencies = ["build"]
+
+[tasks.build_aarch]
+script = "cargo +nightly build -Zbuild-std=core,alloc --target aarch64-unknown-none -v --release"
+
+# Clean
+[tasks.clean]
+command = "cargo"
+args = ["clean"]

fuzzers/baby_no_std/build.rs

@@ -0,0 +1,5 @@
+fn main() {
+    if std::env::var("CARGO_CFG_TARGET_FAMILY").unwrap_or_default() == "unix" {
+        println!("cargo:rustc-link-lib=c");
+    };
+}

fuzzers/baby_no_std/rust-toolchain

@@ -0,0 +1 @@
+nightly

fuzzers/baby_no_std/src/main.rs

@@ -1,8 +1,15 @@
 #![no_std]
 // Embedded targets: build with no_main
-#![cfg_attr(not(any(windows, unix)), no_main)]
+#![cfg_attr(not(any(windows)), no_main)]
 // Embedded needs alloc error handlers which only work on nightly right now...
-#![cfg_attr(not(any(windows, unix)), feature(default_alloc_error_handler))]
+#![cfg_attr(not(any(windows)), feature(default_alloc_error_handler))]
+
+#[cfg(any(windows, unix))]
+extern crate alloc;
+#[cfg(any(windows, unix))]
+use alloc::ffi::CString;
+#[cfg(not(any(windows)))]
+use core::panic::PanicInfo;
 
 use libafl::{
     bolts::{current_nanos, rands::StdRand, tuples::tuple_list, AsSlice},
@@ -20,23 +27,23 @@ use libafl::{
     stages::mutational::StdMutationalStage,
     state::StdState,
 };
-
 #[cfg(any(windows, unix))]
-use cstr_core::CString;
+use libc::{abort, printf};
-#[cfg(any(windows, unix))]
-use libc::{c_char, printf};
-
-#[cfg(not(any(windows, unix)))]
-use core::panic::PanicInfo;
 use static_alloc::Bump;
 
 #[global_allocator]
 static A: Bump<[u8; 512 * 1024 * 1024]> = Bump::uninit();
 
-#[cfg(not(any(windows, unix)))]
 #[panic_handler]
 fn panic(_info: &PanicInfo) -> ! {
-    loop {}
+    #[cfg(unix)]
+    unsafe {
+        abort();
+    }
+    #[cfg(not(unix))]
+    loop {
+        // On embedded, there's not much left to do.
+    }
 }
 
 /// Coverage map with explicit assignments due to the lack of instrumentation
@@ -47,7 +54,7 @@ fn signals_set(idx: usize) {
     unsafe { SIGNALS[idx] = 1 };
 }
 
-/// Provide custom time in no_std environment
+/// Provide custom time in `no_std` environment
 /// Use a time provider of your choice
 #[no_mangle]
 pub extern "C" fn external_current_millis() -> u64 {
@@ -55,8 +62,12 @@ pub extern "C" fn external_current_millis() -> u64 {
     1000
 }
 
+/// The main of this program.
+/// # Panics
+/// Will panic once the fuzzer finds the correct conditions.
 #[allow(clippy::similar_names)]
-pub fn main() {
+#[no_mangle]
+pub extern "C" fn main(_argc: isize, _argv: *const *const u8) -> isize {
     // The closure that we want to fuzz
     let mut harness = |input: &BytesInput| {
         let target = input.target_bytes();
@@ -66,6 +77,7 @@ pub fn main() {
             signals_set(1);
             if buf.len() > 1 && buf[1] == b'b' {
                 signals_set(2);
+                #[allow(clippy::manual_assert)]
                 if buf.len() > 2 && buf[2] == b'c' {
                     panic!("=)");
                 }
@@ -105,10 +117,8 @@ pub fn main() {
         // TODO: Print `s` here, if your target permits it.
         #[cfg(any(windows, unix))]
         unsafe {
-            printf(
+            let s = CString::new(s).unwrap();
-                b"%s\n\0".as_ptr() as *const c_char,
+            printf(b"%s\n\0".as_ptr().cast(), s.as_ptr());
-                CString::new(s).unwrap().as_ptr() as *const c_char,
-            );
         }
     });
 
@@ -147,4 +157,6 @@ pub fn main() {
     fuzzer
         .fuzz_loop(&mut stages, &mut executor, &mut state, &mut mgr)
         .expect("Error in the fuzzing loop");
+
+    0
 }

fuzzers/backtrace_baby_fuzzers/README.md

@@ -1,6 +1,6 @@
 # Backtrace baby fuzzers
 
-The projects contained in this directory are simple fuzzers derived from the original baby_fuzzer examples, whose perpose is to show how to use a `BacktraceObserver` or an `ASANObserver` to dedupe crashes and other necessary component for this feature.
+The projects contained in this directory are simple fuzzers derived from the original baby_fuzzer examples, whose purpose is to show how to use a `BacktraceObserver` or an `ASANObserver` to dedupe crashes and other necessary components for this feature.
 
 The examples cover:
 

fuzzers/backtrace_baby_fuzzers/c_code_with_fork_executor/src/main.rs

@@ -1,14 +1,12 @@
 use std::path::PathBuf;
 
-use libafl::bolts::shmem::ShMemProvider;
-use libafl::bolts::AsSlice;
-use libafl::observers::ConstMapObserver;
 use libafl::{
     bolts::{
         current_nanos,
         rands::StdRand,
-        shmem::{ShMem, StdShMemProvider},
+        shmem::{ShMem, ShMemProvider, StdShMemProvider},
         tuples::tuple_list,
+        AsSlice,
     },
     corpus::{InMemoryCorpus, OnDiskCorpus},
     events::SimpleEventManager,
@@ -20,13 +18,12 @@ use libafl::{
     inputs::{BytesInput, HasTargetBytes},
     monitors::SimpleMonitor,
     mutators::scheduled::{havoc_mutations, StdScheduledMutator},
-    observers::BacktraceObserver,
+    observers::{BacktraceObserver, ConstMapObserver},
     schedulers::QueueScheduler,
     stages::mutational::StdMutationalStage,
     state::StdState,
 };
-use libc::c_int;
+use libc::{c_int, c_uchar};
-use libc::c_uchar;
 extern crate libc;
 
 extern "C" {
@@ -61,10 +58,7 @@ pub fn main() {
     let mut feedback = MaxMapFeedback::new(&observer);
 
     // A feedback to choose if an input is a solution or not
-    let mut objective = feedback_and!(
+    let mut objective = feedback_and!(CrashFeedback::new(), NewHashFeedback::new(&bt_observer));
-        CrashFeedback::new(),
-        NewHashFeedback::<BacktraceObserver>::new(&bt_observer)
-    );
 
     // create a State from scratch
     let mut state = StdState::new(

fuzzers/backtrace_baby_fuzzers/c_code_with_inprocess_executor/src/main.rs

@@ -1,12 +1,10 @@
 use std::path::PathBuf;
 
-use libafl::bolts::AsSlice;
-use libafl::executors::InProcessExecutor;
-use libafl::observers::ConstMapObserver;
 use libafl::{
-    bolts::{current_nanos, rands::StdRand, tuples::tuple_list},
+    bolts::{current_nanos, rands::StdRand, tuples::tuple_list, AsSlice},
     corpus::{InMemoryCorpus, OnDiskCorpus},
     events::SimpleEventManager,
+    executors::InProcessExecutor,
     feedback_and,
     feedbacks::{CrashFeedback, MaxMapFeedback, NewHashFeedback},
     fuzzer::{Fuzzer, StdFuzzer},
@@ -14,7 +12,7 @@ use libafl::{
     inputs::{BytesInput, HasTargetBytes},
     monitors::SimpleMonitor,
     mutators::scheduled::{havoc_mutations, StdScheduledMutator},
-    observers::BacktraceObserver,
+    observers::{BacktraceObserver, ConstMapObserver},
     schedulers::QueueScheduler,
     stages::mutational::StdMutationalStage,
     state::StdState,
@@ -50,10 +48,7 @@ pub fn main() {
     let mut feedback = MaxMapFeedback::new(&observer);
 
     // A feedback to choose if an input is a solution or not
-    let mut objective = feedback_and!(
+    let mut objective = feedback_and!(CrashFeedback::new(), NewHashFeedback::new(&bt_observer));
-        CrashFeedback::new(),
-        NewHashFeedback::<BacktraceObserver>::new(&bt_observer)
-    );
 
     // create a State from scratch
     let mut state = StdState::new(

fuzzers/backtrace_baby_fuzzers/command_executor/build.rs

@@ -3,7 +3,7 @@ use std::env;
 fn main() {
     let cwd = env::current_dir().unwrap().to_string_lossy().to_string();
     let mut cmd = cc::Build::new().get_compiler().to_command();
-    cmd.args(&["src/test_command.c", "-o"])
+    cmd.args(["src/test_command.c", "-o"])
         .arg(&format!("{}/test_command", &cwd))
         .arg("-fsanitize=address")
         .status()