Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Suggestions for kernel-hardening-checker #113

Closed
asarubbo opened this issue Mar 7, 2024 · 4 comments
Closed

Suggestions for kernel-hardening-checker #113

asarubbo opened this issue Mar 7, 2024 · 4 comments
Labels
question Further information is requested

Comments

@asarubbo
Copy link

asarubbo commented Mar 7, 2024

Hello @a13xp0p0v

I have two suggestions for kernel-hardening-checker

  1. It's a matter of fact that enable all suggested security features impact on perfomance and I have verified this by myself, e.g. a 64 threads modern server takes 10 more minutes to compile chromium.
    Would be great have a column that gives a rank (from 1-10) maybe about how much a CONFIG_* impacts on performance.
    To give an idea about what I'm talking, with a general example with C and stack protection:
    FORTIFY_SOURCE has impact of 1;
    -fstack-protector-strong has an impact of 5;
    -fstack-protector-all has an impact of 8;

Rank number can be on your judge with will give the idea to the user.

  1. I noticed that the option do not follow the order from menuconfig. For example for enable the first suggestion from kernel-hardening-checker I need to go in a section of the kernel, then for the second suggestion I need to go to another section, and for the third I maybe come back to the section of the first suggestion. That takes a lot of time for navigating into the menuconfig sections, while group CONFIG_* based on the menuconfig order will save a lot of time.

Thanks

@migrgh
Copy link

migrgh commented Mar 16, 2024

I had similar thoughts, the performance rating sounds sensible
but is probably difficult to implement because you always have
to ask yourself in which scenario you achieve a plus or not.
I would say that someone who uses the suggestions does not
use them from a performance point of view but from a security
point of view.

@a13xp0p0v
Copy link
Owner

Hello @asarubbo and @migrgh!

Would be great have a column that gives a rank (from 1-10) maybe about how much a CONFIG_* impacts on performance.

@asarubbo, that's an interesting idea. Could you please describe it in the issue #66 as well?

However, creating such a rating would not be easy because some kernel security features have different performance penalty depending on the type of system workload (a number and type of system calls, for example).

Do you have an idea which particular kernel option makes your system run slow on compiling chromium?

First of all, I would recommend comparing performance of the default configuration and hardened configuration without mitigations=auto,nosmt and nosmt boot options (they may have the biggest performance penalty). I guess, in that comparison, you will not see a big difference in the chromium compilation time.

@a13xp0p0v
Copy link
Owner

I noticed that the option do not follow the order from menuconfig.

@asarubbo, yes, that's true. Currently, the options are sorted by the complexity of the checking rule. It's easier for maintenance.

You have multiple options to avoid exhausting navigation in menuconfig.

  • Try using search in menuconfig: press '/', enter the option name, hit enter, and then choose the number (1, 2, 3, ...) of the option that you want to see. I like it.
  • Try automatic merging of the Kconfig fragment with options that you want to change. See the example in the README.

@a13xp0p0v
Copy link
Owner

Closing for now.
Feel free to reopen this issue, if you have new info.
Thanks!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
question Further information is requested
Projects
None yet
Development

No branches or pull requests

3 participants