[kernel-hardening] Re: [RFC v2] implement SL*B and stack usercopy runtime checks

[Matt Mackall <mpm@selenic.com>]

On Mon, 2011-07-18 at 22:39 +0400, Vasiliy Kulikov wrote:
> This patch implements 2 additional checks for the data copied from
> kernelspace to userspace and vice versa (original PAX_USERCOPY from PaX
> patch).  Currently there are some very simple and cheap comparisons of
> supplied size and the size of a copied object known at the compile time
> in copy_* functions.  This patch enhances these checks to check against
> stack frame boundaries and against SL*B object sizes.
> 
> More precisely, it checks:
> 
> 1) if the data touches the stack, checks whether it fully fits in the stack
> and whether it fully fits in a single stack frame.  The latter is arch
> dependent, currently it is implemented for x86 with CONFIG_FRAME_POINTER=y
> only.  It limits infoleaks/overwrites to a single frame and local variables
> only, and prevents saved return instruction pointer overwriting.
> 
> 2) if the data is from the SL*B cache, checks whether it fully fits in a
> slab page and whether it overflows a slab object.  E.g. if the memory
> was allocated as kmalloc(64, GFP_KERNEL) and one tries to copy 150
> bytes, the copy would fail.

FYI, this should almost certainly be split into (at least) two patches:

- the stack check
- the SL*B check (probably one patch per allocator, preceded by one for
any shared infrastructure)

-- 
Mathematics is the supreme nostalgia of our time.