Model checking the BPF verifier
a project by shunghsiyu
Updated about 1 year ago. 9 hacker ♥️. 3 followers.

Project Description

BPF verifier plays a crucial role in securing the system (though less so now that unprivileged BPF is disabled by default in both upstream and SLES), and bugs in the verifier has lead to privilege escalation vulnerabilities in the past (e.g. CVE-2021-3490).

One way to check whether the verifer has bugs to use model checking (a formal verification technique), in other words, build a abstract model of how the verifier operates, and then see if certain condition can occur (e.g. incorrect calculation during value tracking of registers) by giving both the model and condition to a solver.

For the solver I will be using the Z3 SMT solver to do the checking since it provide a Python binding that's relatively easy to use.

Goal for this Hackweek

Learn how to use the Z3 Python binding (i.e. Z3Py) to build a model of (part of) the BPF verifier, probably the part that's related to value tracking using tristate numbers (aka tnum), and then check that the algorithm work as intended.

Resources

Join this project

Looking for hackers with the skills:

bpf ebpf formalverification modelchecking kernel security

This project is part of:

Hack Week 21 Hack Week 23 Hack Week 24

Activity

  • about 1 month ago: janvhs liked this project.
  • about 1 month ago: flonnegren liked this project.
  • about 1 month ago: vliaskovitis liked this project.
  • about 2 months ago: r1chard-lyu liked this project.
  • about 1 year ago: moio liked this project.
  • over 2 years ago: dsterba liked this project.
  • over 2 years ago: mbrugger liked this project.
  • over 2 years ago: jzerebecki left this project.
  • over 2 years ago: jzerebecki added keyword "security" to this project.
  • over 2 years ago: jzerebecki joined this project.
  • over 2 years ago: jzerebecki liked this project.
  • over 2 years ago: ailiopoulos liked this project.
  • over 2 years ago: shunghsiyu started this project.
  • over 2 years ago: shunghsiyu added keyword "kernel" to this project.
  • over 2 years ago: shunghsiyu added keyword "bpf" to this project.
  • over 2 years ago: shunghsiyu added keyword "ebpf" to this project.
  • over 2 years ago: shunghsiyu added keyword "formalverification" to this project.
  • over 2 years ago: shunghsiyu added keyword "modelchecking" to this project.
  • over 2 years ago: shunghsiyu originated this project.


    • Comments

    • shunghsiyu
      over 2 years ago by shunghsiyu | Reply

      I've uploaded the jupyter notebook on GitHub that contains a minimal model of tnum along with tnum_add(), as well as the prove that it works.

    • shunghsiyu
      over 2 years ago by shunghsiyu | Reply

      The slides used for lightning talk can be found here

      While I'd like to achieve much more, I think what I've done during hack week is suffice to be called a complete project, so I'm marking this as complete

    • shunghsiyu
      about 1 year ago by shunghsiyu | Reply

      I'm restarting this project to model check the range tracking (minimal and maximal value possible in s32/u32/u64/s64 range) done in BPF verifier.

      In addition to that I hope to unify signed and unsigned range tracking based on previously posted idea and the "Interval Analysis and Machine Arithmetic: Why Signedness Ignorance Is Bliss" paper for simpler range tracking code.

    Similar Projects

    Tracing system calls with eBPF by doreilly

    Description

    Many security tools need to record system calls like execve. Using the Linux audit system for this can have a detrimental performance impact in some cases.

    Goals

    The goal is to investigate eBPF as an alternative and do some benchmarking to see the impact and how it compares to using the audit subsystem.

    Progress

    BPF done - traceexec

    Benchmark report

    Resources

    eBPF doc

    libbpf

    libMicro benchmark tool

    Tracing system calls with eBPF by doreilly

    Description

    Many security tools need to record system calls like execve. Using the Linux audit system for this can have a detrimental performance impact in some cases.

    Goals

    The goal is to investigate eBPF as an alternative and do some benchmarking to see the impact and how it compares to using the audit subsystem.

    Progress

    BPF done - traceexec

    Benchmark report

    Resources

    eBPF doc

    libbpf

    libMicro benchmark tool

    Officially Become a Kernel Hacker! by m.crivellari

    Description

    My studies as well my spare time are dedicated to the Linux Kernel. Currently I'm focusing on interrupts on x86_64, but my interests are not restricted to one specific topic, for now.

    I also "played" a little bit with kernel modules (ie lantern, a toy packet analyzer) and I've added a new syscall in order read from a task A, the memory of a task B.

    Maybe this will be a good chance to...

    Goals

    • create my first kernel patch

    Resources

    Achivements

    Improve various phones kernel mainline support (Qualcomm, Exynos, MediaTek) by pvorel

    Similar to previous hackweeks ( https://hackweek.opensuse.org/projects/improve-qualcomm-soc-msm8994-slash-msm8992-kernel-mainline-support, https://hackweek.opensuse.org/projects/test-mainline-kernel-on-an-older-qualcomm-soc-msm89xx-explore-mainline-kernel-qualcomm-mainlining) try to improve kernel mainline support of various phones.

    Modernize ocfs2 by goldwynr

    Ocfs2 has gone into a stage of neglect and disrepair. Modernize the code to generate enough interest.

    Goals: * Change the mount sequence to use fscontext * Move from using bufferhead to bio/folios * Use iomap * Run it through xfstests

    Create DRM drivers for VESA and EFI framebuffers by tdz

    Description

    We already have simpledrm for firmware framebuffers. But the driver is originally for ARM boards, not PCs. It is already overloaded with code to support both use cases. At the same time it is missing possible features for VESA and EFI, such as palette modes or EDID support. We should have DRM drivers for VESA and EFI interfaces. The infrastructure exists already and initial drivers can be forked from simpledrm.

    Goals

    • Initially, a bare driver for VESA or EFI should be created. It can take functionality from simpledrm.
    • Then we can begin to add additional features. The boot loader can provide EDID data. With VGA hardware, VESA can support paletted modes or color management. Example code exists in vesafb.

    early stage kdump support by mbrugger

    Project Description

    When we experience a early boot crash, we are not able to analyze the kernel dump, as user-space wasn't able to load the crash system. The idea is to make the crash system compiled into the host kernel (think of initramfs) so that we can create a kernel dump really early in the boot process.

    Goal for the Hackweeks

    1. Investigate if this is possible and the implications it would have (done in HW21)
    2. Hack up a PoC (done in HW22 and HW23)
    3. Prepare RFC series (giving it's only one week, we are entering wishful thinking territory here).

    update HW23

    • I was able to include the crash kernel into the kernel Image.
    • I'll need to find a way to load that from init/main.c:start_kernel() probably after kcsan_init()
    • I workaround for a smoke test was to hack kexec_file_load() systemcall which has two problems:
      1. My initramfs in the porduction kernel does not have a new enough kexec version, that's not a blocker but where the week ended
      2. As the crash kernel is part of init.data it will be already stale once I can call kexec_file_load() from user-space.

    The solution is probably to rewrite the POC so that the invocation can be done from init.text (that's my theory) but I'm not sure if I can reuse the kexec infrastructure in the kernel from there, which I rely on heavily.

    update HW24

    • Day1
      • rebased on v6.12 with no problems others then me breaking the config
      • setting up a new compilation and qemu/virtme env
      • getting desperate as nothing works that used to work
    • Day 2
      • getting to call the invocation of loading the early kernel from __init after kcsan_init()

    • Day 3

      • fix problem of memdup not being able to alloc so much memory... use 64K page sizes for now
      • code refactoring
      • I'm now able to load the crash kernel
      • When using virtme I can boot into the crash kernel, also it doesn't boot completely (major milestone!), crash in elfcorehdr_read_notes()

    • Day 4

      • crash systems crashes (no pun intended) in copy_old_mempage() link; will need to understand elfcorehdr...
      • call path vmcore_init() -> parse_crash_elf_headers() -> elfcorehdr_read() -> read_from_oldmem() -> copy_oldmem_page() -> copy_to_iter()

    • Day 5

      • hacking arch/arm64/kernel/crash_dump.c:copy_old_mempage() to see if crash system really starts. It does.
      • fun fact: retested with more reserved memory and with UEFI FW, host kernel crashes in init but directly starts the crash kernel, so it works (somehow) \o/

    • TODOs

      • fix elfcorehdr so that we actually can make use of all this...
      • test where in the boot __init() chain we can/should call kexec_early_dump()

    VulnHeap by r1chard-lyu

    Description

    The VulnHeap project is dedicated to the in-depth analysis and exploitation of vulnerabilities within heap memory management. It focuses on understanding the intricate workflow of heap allocation, chunk structures, and bin management, which are essential to identifying and mitigating security risks.

    Goals

    • Familiarize with heap
      • Heap workflow
      • Chunk and bin structure
      • Vulnerabilities
    • Vulnerability
      • Use after free (UAF)
      • Heap overflow
      • Double free
    • Use Docker to create a vulnerable environment and apply techniques to exploit it

    Resources

    • https://heap-exploitation.dhavalkapil.com/divingintoglibc_heap
    • https://raw.githubusercontent.com/cloudburst/libheap/master/heap.png
    • https://github.com/shellphish/how2heap?tab=readme-ov-file

    OIDC Loginproxy by toe

    Description

    Reverse proxies can be a useful option to separate authentication logic from application logic. SUSE and openSUSE use "loginproxies" as an authentication layer in front of several services.

    Currently, loginproxies exist which support LDAP authentication or SAML authentication.

    Goals

    The goal of this Hack Week project is, to create another loginproxy which supports OpenID Connect authentication which can then act as a drop-in replacement for the existing LDAP or SAML loginproxies.

    Testing is intended to focus on the integration with OIDC IDPs from Okta, KanIDM and Authentik.

    Resources

    Kanidm: A safe and modern IDM system by firstyear

    Kanidm is an IDM system written in Rust for modern systems authentication. The github repo has a detailed "getting started" on the readme.

    Kanidm Github

    In addition Kanidm has spawn a number of adjacent projects in the Rust ecosystem such as LDAP, Kerberos, Webauthn, and cryptography libraries.

    In this hack week, we'll be working on Quokca, a certificate authority that supports PKCS11/TPM storage of keys, issuance of PIV certificates, and ACME without the feature gatekeeping implemented by other CA's like smallstep.

    For anyone who wants to participate in Kanidm, we have documentation and developer guides which can help.

    I'm happy to help and share more, so please get in touch!

    CVE portal for SUSE Rancher products by gmacedo

    Description

    Currently it's a bit difficult for users to quickly see the list of CVEs affecting images in Rancher, RKE2, Harvester and Longhorn releases. Users need to individually look for each CVE in the SUSE CVE database page - https://www.suse.com/security/cve/ . This is not optimal, because those CVE pages are a bit hard to read and contain data for all SLE and BCI products too, making it difficult to easily see only the CVEs affecting the latest release of Rancher, for example. We understand that certain costumers are only looking for CVE data for Rancher and not SLE or BCI.

    Goals

    The objective is to create a simple to read and navigate page that contains only CVE data related to Rancher, RKE2, Harvester and Longhorn, where it's easy to search by a CVE ID, an image name or a release version. The page should also provide the raw data as an exportable CSV file.

    It must be an MVP with the minimal amount of effort/time invested, but still providing great value to our users and saving the wasted time that the Rancher Security team needs to spend by manually sharing such data. It might not be long lived, as it can be replaced in 2-3 years with a better SUSE wide solution.

    Resources

    • The page must be simple and easy to read.
    • The UI/UX must be as straightforward as possible with minimal visual noise.
    • The content must be created automatically from the raw data that we already have internally.
    • It must be updated automatically on a daily basis and on ad-hoc runs (when needed).
    • The CVE status must be aligned with VEX.
    • The raw data must be exportable as CSV file.
    • Ideally it will be written in Go or pure Shell script with basic HTML and no external dependencies in CSS or JS.

    Contributing to Linux Kernel security by pperego

    Description

    A couple of weeks ago, I found this blog post by Gustavo Silva, a Linux Kernel contributor.

    I always strived to start again into hacking the Linux Kernel, so I asked Coverity scan dashboard access and I want to contribute to Linux Kernel by fixing some minor issues.

    I want also to create a Linux Kernel fuzzing lab using qemu and syzkaller

    Goals

    1. Fix at least 2 security bugs
    2. Create the fuzzing lab and having it running

    The story so far

    • Day 1: setting up a virtual machine for kernel development using Tumbleweed. Reading a lot of documentation, taking confidence with Coverity dashboard and with procedures to submit a kernel patch
    • Day 2: I read really a lot of documentation and I triaged some findings on Coverity SAST dashboard. I have to confirm that SAST tool are great false positives generator, even for low hanging fruits.
    • Day 3: Working on trivial changes after I read this blog post: https://www.toblux.com/posts/2024/02/linux-kernel-patches.html. I have to take confidence with the patch preparation and submit process yet.
      • First trivial patch sent: using strtruefalse() macro instead of hard-coded strings in a staging driver for a lcd display
      • Fix for a dereference before null check issue discovered by Coverity (CID 1601566) https://scan7.scan.coverity.com/#/project-view/52110/11354?selectedIssue=1601566
    • Day 4: Triaging more issues found by Coverity.
      • The patch for CID 1601566 was refused. The check against the NULL pointer was pointless so I prepared a version 2 of the patch removing the check.
      • Fixed another dereference before NULL check in iwlmvmparsewowlaninfo_notif() routine (CID 1601547). This one was already submitted by another kernel hacker :(
    • Day 5: Wrapping up. I had to do some minor rework on patch for CID 1601566. I found a stalker bothering me in private emails and people I interacted with me, advised he is a well known bothering person. Markus Elfring for the record.

    • Wrapping up: being back doing kernel hacking is amazing and I don't want to stop it. My battery pack is completely drained but changing the scope gave me a great twist and I really want to feel this energy not doing a single task for months.

      I failed in setting up a fuzzing lab but I was too optimistic for the patch submission process.

    The patches

    1