Skip to content

Latest commit

 

History

History
71 lines (51 loc) · 2.41 KB

README.md

File metadata and controls

71 lines (51 loc) · 2.41 KB
Raw

GRAN

Gran, a security analyser for Grsecurity RBAC policies

Requirements

  • Python 2.7 or greater (Python 3.X supported)
  • PLY module, available in most Linux distributions. It can also be installed manually by executing easy_install ply

Usage

usage: gran [-h] [-a] [-b] [-p PROCESSEDPOLICY] [-e ENTRYPOINTS] [-t TARGETS]
            [-l LEARNCONFIG] [-d] [-v]
            policy

a security analyser for Grsecurity RBAC policies.

positional arguments:
  policy                policy file to be analyzed

optional arguments:
  -h, --help            show this help message and exit
  -a, --admin           include administrative special roles in the analysis
  -b, --bestcase        assume there are not setuid/setgid files in the system
  -e ENTRYPOINTS, --entrypoints ENTRYPOINTS
			specify the entrypoints file
  -t TARGETS, --targets TARGETS
			specify a file with sensitive targets
  -l LEARNCONFIG, --learnconfig LEARNCONFIG
			specify the learn_config file to get sensitive targets
  -d, --debug           enable debugging mode
  -v, --version         show program's version number and exit

Targets file consists of sensitive path, one entry per line. In case of no targets specified, gran computes all the states and transitions and quits.

Each line of the entry-points file has the following syntax:

<stateA> [<stateB> <target>]

where stateA and stateB are triples of the form <role_name>:<type>:<subject> and target is a sensitive path that should not be leaked from stateA to stateB by an indirect flow.

If no entry-points file is provided, gran assumes all the roles in the policy file paired with the default subject '/' as entry points.

Examples of targets and entry-points files can be found under misc/.

Limitations

At the moment, the following Grsecurity RBAC features are supported:

  • role modes: u, g, s, A
  • role attributes: role_transitions
  • subject modes: o
  • subject attributes: user_transition_allow, user_transition_deny, group_transition_allow, group_transition_deny
  • object modes: r, w, a, c, x, d, h, none
  • capabilities: CAP_ALL, CAP_SETUID, CAP_SETGID
  • domains
  • include, define and replace rules

The tool provides a limited support for wild-carded objects. Nested subjects are not supported, since the learning system of Grsecurity does not account for them.