Fixes

README.md

@@ -32,7 +32,10 @@ installed. See "Artifact Evaluation" below.
 ```sh
 # install dependencies
 # on Fedora:
-sudo dnf install python3-pip python3-mypy python3-capstone glibc-static elfutils-libelf-devel dtc capstone-devel libdwarf-devel glib2-devel pixman-devel protobuf-devel protobuf-c-devel curl-devel jsoncpp-devel chrpath datamash telnet
+sudo dnf install python3-pip python3-mypy python3-capstone glibc-static \
+	elfutils-libelf-devel dtc capstone-devel libdwarf-devel glib2-devel \
+	pixman-devel protobuf-devel protobuf-c-devel curl-devel jsoncpp-devel \
+	chrpath datamash telnet
 pip install yq sortedcontainers
 
 # Rust via rustup (see https://rustup.rs)
@@ -50,7 +53,8 @@ fs-testing/linux/build-kernel.sh nova
 
 git clone https://github.com/linux-pmfs/pmfs fs-testing/linux/pmfs
 # Note: Building PMFS requires gcc4, build from a suitable container. For example:
-podman run --rm -v"$PWD/fs-testing/linux:/mnt" docker.io/library/gcc:4 sh -c 'apt-get update && apt-get install bc && /mnt/build-kernel.sh pmfs'
+podman run --rm -v"$PWD/fs-testing/linux:/mnt" docker.io/library/gcc:4 \
+	sh -c 'apt-get update && apt-get install bc && /mnt/build-kernel.sh pmfs'
 
 ```
 

artifact-evaluation/README.md

@@ -12,7 +12,8 @@ its analysis in parallel, so make sure to provide plenty of memory (`-m`) and
 vCPUs (`-smp`). As a rough guideline, provide 2 GB of memory per vCPU. For
 example with QEMU/KVM:
 ```
-qemu-kvm -m 16G -smp 8 -display none -serial mon:stdio -device e1000,netdev=net0 -netdev user,id=net0,hostfwd=tcp::2222-:22 vinter.qcow2
+qemu-kvm -m 16G -smp 8 -display none -serial mon:stdio -device e1000,netdev=net0 \
+    -netdev user,id=net0,hostfwd=tcp::2222-:22 vinter.qcow2
 ```
 
 Connect to the virtual machine via SSH. The password for the users vinter and
@@ -35,8 +36,10 @@ fs-testing/scripts/run_getting-started.sh
 The script will put results into the directory `results_getting-started`. View
 a short summary of these results with the following commands:
 ```
-vinter_python/report-results.py analyze results_getting-started/vinter_python/vm_nova/getting-started/test_hello-world
-vinter_python/report-results.py analyze results_getting-started/vinter_python/vm_pmfs/getting-started/test_hello-world
+vinter_python/report-results.py analyze \
+    results_getting-started/vinter_python/vm_nova/getting-started/test_hello-world
+vinter_python/report-results.py analyze \
+    results_getting-started/vinter_python/vm_pmfs/getting-started/test_hello-world
 ```
 
 You can see that Vinter reports a violation of *single final state* for the

artifact-evaluation/walkthrough_performance-benchmarks.md

@@ -12,7 +12,7 @@ fs-testing/scripts/script_run-paper-benchmark.sh
 The benchmark may take between 30 and 60 minutes to complete and writes its
 files to `results_paper-benchmark`.
 
-Its final output should be similar to the following:
+Its final output should be similar to the following (only placeholder values listed here):
 ```
 == Figure 8 (pmemtrace.py) results ==
                                                count  mean               sstdev
@@ -39,5 +39,7 @@ and result in 7 unique semantic crash states.
 success
 ```
 
+NOTE: The benchmark results listed in the paper PDF are not yet up to date with the latest artifact version. The differences are not big though.
+
 The tables should approximately reproduce the numbers in Figure 8 and Figure 9 (depending on the hardware used).
 The paragraph at the end of the output should further contain the same numbers as the corresponding sentence in §6.3.