Skip to main content

12 Principles of Secure Linux Programming

1. Operate with least privilege

  • Hold privileges only while they're required.
  • Drop privileges permanently when they'll never be used again.
  • A privileged program should never exec a shell unless the environment can silently ignore set-user-ID and set-group-ID permissions.
  • Close all unnecessary file descriptors before an exec.

2. Avoid exposing sensitive information

  • Use mlock to lock access to the virtual memory page in use.
  • Prevent core dumps by using setrlimit.

3. Confine the process

  • Use capabilities and securebit flags whenever possible.
  • Establish a chroot jail to limit the set of directories and files that a program may access.
  • Use a virtual server: UML, Xen, KVM, etc.

4. Beware of signals and race conditions

  • Signals should be caught, blocked, or ignored to prevent possible security problems.

5. Pitfalls of file operations and file I/O

  • Umask should be set to a value that ensures the process never creates publicly writable files.
  • Use seteuid and setreuid to temporarily change process credentials to ensure new files do not belong to the wrong user.
  • Never create a file owned by the program owner.
  • Never allow other users to write to files that the process uses.
  • Use mkstemp to create files with unpredictable names.

6. Don't trust inputs or the environment

  • Do not assume values of environment variables are reliable.
  • Validate all inputs from untrusted sources.
  • Avoid unreliable assumptions about the process's run-time environment.

7. Beware of injection attacks

  • Use regex pattern matching to filter user inputs.
  • Encode user input into Base-64 encoded strings.
  • Never compare raw strings when checking credentials.

8. Beware of buffer overruns

  • Do not allow an input value or a copied string to exceed allocated buffer space.
  • Use if statements that prevent buffer overruns.

9. Beware of denial-of-service attacks

  • Minimize the risk and consequences of overload attacks.
    • Perform load throttling.
    • Use resource limits and disk quotas.
  • Employ timeouts for communication with clients.
  • Perform log throttling.
  • Perform bounds checking on data structures.
  • Design data structures that avoid algorithmic-complexity attacks.

10. Beware of database attacks

  • Always use a mutex to ensure the right process has access at the right time.
  • Use databases that enforce host-based security policies.
  • Use databases that enforce separation of database ownership and table ownership.

11. Check return statuses and fail safely

  • Always check the program's return values.
  • Store the value then check it before returning it.
  • Return it then ask for verification.
  • Unexpected situations must cause the program to terminate or drop the client request.

12. Beware of reverse-engineering attacks

  • Avoid hardcoding sensitive information in string literals.
  • Generate one-time IDs for runtime execution.
  • Salt IDs with unique hardware device addresses.

Do you have a suggestion about how to improve this blog? Let's talk about it. Contact me at David.Brenner.Jr@Gmail.com or 720-584-5229.

Comments

Popular posts from this blog

The meaning of time in reinforcement learning

Reinforcement learning (RL) is one of three basic machine learning paradigms, alongside supervised learning and unsupervised learning. Reinforcement learning is concerned with how software agents ought to take actions in an environment in order to maximize the notion of cumulative reward through the process of trial and error. In reinforcement learning an agent starts at an empty state then analyzes the available datasets according to a policy of positive states and negative states. Rather than being explicitly taught as in supervised learning the correct set of actions for performing a task, reinforcement learning uses rewards as signals for positive states and punishments as signals for negative states. The agent obtains the best path to a desirable reward as a cumulation of positive states and negative states. As compared to unsupervised learning, reinforcement learning is different in terms of goals. While the goal in unsupervised learning is to find similarities and differences...

Old idea of encrypted, anonymous group chats

Encrypted, Anonymous Group Chats An owner of the chat connects through multiple VPNs, like NordVPN and SurfShark which are the most popular. Then the owner obtains access to an email provider hosted in a country outside the United States. Once the new email account has been setup and ready to use, the owner shares the login username and login password of that email account with the participants. The idea is to never send/receive any emails, only exchange messages saved as drafts in the email account. The drafts don't get sent/received anywhere. It's also important to note that messages are saved as new drafts without subject-line and recipient-info. Ideally there will be at most two drafts in the account at a time. When the chat is finished the email account is deleted. Whenever the participants are ready to chat, the participants login to that email account and compose a new email. They will write a message and encrypt it with PGP, then save it as a new draft without subjec...

Uploading files through Secure WebDAV using DAVfs

WebDAV is a protocol that facilitates uploading and downloading files through HTTP (port 80) and HTTPS (port 443). Whenever a WebDAV service is being ran over SSL it is called Secure WebDAV. DAVfs is a file system interface to the WebDAV protocol, it works with WebDAV and Secure WebDAV. The command mount uses DAVfs to recognize a WebDAV share as a regular file system so that other tools, scripts, services, and users can access the share's contents (as a file system with actual directories). Here's an easy solution for uploading files to your WebDAV account. These instructions work on Linux, FreeBSD, Solaris, and probably other distributions too. 1. Make a local directory for transferring files. mkdir <your directory>; 2. Stop other processes and users from interfering with your transfers. chown root:root <your directory> && chmod 770 <your directory>; 3. Mount your online cloud share using davfs. Enter your password when the prompt appears askin...