Re: [kernel-hardening] [RFC V2 4/6] lib: vsprintf: default kptr_restrict to the maximum value

From: Linus Torvalds
Date: Wed Oct 04 2017 - 13:28:20 EST

On Wed, Oct 4, 2017 at 10:08 AM, Linus Torvalds
<torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
> So I honestly doubt the value of kptr_restrict. Any *sane* policy
> pretty much has to be in the caller, and by thinking about what you
> print out. IOW, things like proc_pid_wchan().

Looking at /proc/kallsyms is actually a prime example of this.

IOW, the old "open /proc/kallsyms as a normal user, then make it stdin
for some suid-root program that can be fooled to output it probably
works on it.

So kptr_restrict ends up being entirely the wrong thing to do there.

The only value in kptr_restrict ends up being as a complete hack,
where you say "I trust nobody" and make %p almost entirely useless.

And as mentioned, that will just make people use %x instead, or
randomly sprinkle the new "I didn't really mean this" modifiers like
the already discussed pr_debug() case.

So even when kptr_restrict "works", it ends up just fighting itself.
And most of the time it just doesn't work.