User Controls

niceness and limits

  1. #1
    maddie Houston
    Was heading home yesterday and was thinking..

    How much damage could i cause if i write a small bash script that manipulates niceness, and also change the stack size and processes limits for the kernel. That way it will to prevent any user/group or root that i choose from executing certain functions to their system..

    It would basically be like backdooring the GNU coreutils with rust or C, but different?

    Obviously the obfuscation would be seriously lacking in this department, but i actually never seen it done, before indulging in this small projects, i was curious if anyone has seen this done with documentation, or in the wild before?
  2. #2
    Sudo Dark Matter [my hereto riemannian peach]
    You can literally cause more damage with a 2liter of gas and a book of matches. Aim higher
  3. #3
    maddie Houston
    Originally posted by Sudo You can literally cause more damage with a 2liter of gas and a book of matches. Aim higher

    its not about aiming higher, its about new techniques, and trying new things.
    actually i was just messing around and completely broke a system and it went into a kernel panic..
    so try again
  4. #4
    Op, you have penis don't you.
  5. #5
    Originally posted by Jiggaboo_Johnson Op, you have penis don't you.

    more like penises
  6. #6
    Sophie Pedophile Tech Support
    Buffers, and why they overflow. A tentative step towards your goals.

    Alright, ok.

    Any particular procs particularly sensitive to having their NICE values changed? I'd imagine anything userland shouldn't have that much of an impact.
  7. #7
    maddie Houston
    i just set a few random processes to -20, changed limits of a few groups. created a forkbomb and down it does.

    random variables for these dont seem to work everytime, but every couple of dozen runs, it seems to get janky and likes to fall on its face.
    ill start hard coding values and see if i can get a more accurate measure on what is being changed.
  8. #8
    Sophie Pedophile Tech Support
    Originally posted by maddie i just set a few random processes to -20, changed limits of a few groups. created a forkbomb and down it does.

    random variables for these dont seem to work everytime, but every couple of dozen runs, it seems to get janky and likes to fall on its face.
    ill start hard coding values and see if i can get a more accurate measure on what is being changed.

    You probably want to automate the runs and save the numbers that get the result you are hoping to achieve, then when you have a good sample size run the numbers that seem to be working more often let's call it Pool X and as a control run a Pool Y with the same sample size but randomized numbers. Then you can see if the values that are coming up as 'successful' have any statistical significance.

    Might wanna try to see if it also works without the forkbomb.
Jump to Top