Re: [PATCH v2] seq_file: Unconditionally use vmalloc for buffer

[Kees Cook <keescook@chromium.org>]

On Mon, Mar 15, 2021 at 06:33:10PM +0000, Al Viro wrote:
> On Mon, Mar 15, 2021 at 10:48:51AM -0700, Kees Cook wrote:
> > The sysfs interface to seq_file continues to be rather fragile, as seen
> > with some recent exploits[1]. Move the seq_file buffer to the vmap area
> > (while retaining the accounting flag), since it has guard pages that
> > will catch and stop linear overflows. This seems justified given that
> > seq_file already uses kvmalloc(), is almost always using a PAGE_SIZE or
> > larger allocation, has allocations are normally short lived, and is not
> > normally on a performance critical path.
> 
> You are attacking the wrong part of it.  Is there any reason for having
> seq_get_buf() public in the first place?

Completely agreed. seq_get_buf() should be totally ripped out.
Unfortunately, this is going to be a long road because of sysfs's ATTR
stuff, there are something like 5000 callers, and the entire API was
designed to avoid refactoring all those callers from
sysfs_kf_seq_show().

However, since I also need to entirely rewrite the sysfs vs kobj APIs[1]
for CFI, I'm working on a plan to fix it all at once, but based on my
experience refactoring the timer struct, it's going to be a very painful
and long road.

So, in the meantime, I'd like to make this change so we can get bounds
checking for free on seq_file (since it's almost always PAGE_SIZE
anyway).

-Kees

[1] https://lore.kernel.org/lkml/202006112217.2E6CE093@keescook/

-- 
Kees Cook