Re: [kernel-hardening] [PATCH] printk: hash addresses printed with %p

[Tobin C. Harding]

Removing kvm@xxxxxxxxxxxxxxx from the CC list.

On Wed, Oct 11, 2017 at 01:48:58PM -0400, Theodore Ts'o wrote:
> On Wed, Oct 11, 2017 at 02:48:16PM +1100, Tobin C. Harding wrote:
> > +/*
> > + * Obfuscates pointer (algorithm taken from kptr_obfuscate(). See kernel/kcmp.c)
> > + * v is the pointer value, randval is some random value, oddval is some random
> > + * odd value.
> > + *
> > + * The obfuscation is done in two steps. First we xor the kernel pointer with
> > + * a random value, which puts pointer into a new position in a reordered space.
> > + * Secondly we multiply the xor production with a large odd random number to
> > + * permute its bits even more (the odd multiplier guarantees that the product
> > + * is unique ever after the high bits are truncated, since any odd number is
> > + * relative prime to 2^n).
> > + */
> 
> Why not just expose kptr_obfusecate() and use it, instead of copying
> code?
> 
> Also, I'm nervous about the obfuscation.  If the attacker can get a
> handful of known "real kernel pointer" and "obfuscated kernel pointer"
> values, it wouldn't be that hard for them to be able to reverse
> engineer the two secret values.
> 
> Perhaps the argument is "if the attacker can get a _single_ real
> kernel address, it's all over anyway", which is probably true for
> KASLR, but which might not be true for all attacks.
> 
> Anyway, if you use kptr_obfuscate in kernel/kcmp.c, then if we later
> decide that we should change the obfuscation algorithm to something
> stronger, we only need to do it in one place.
> 
> 					- Ted

Thanks Ted, others have misgivings about this method also. The email threads are all a bit mixed up
(thansk to my ineptness at posting patches :).

FYI, in the other threads Jason A. Donenfel and Linus Torvalds have discussed SipHash as a solution.

thanks,
Tobin.