Re: [RFC PATCH 0/1] Integer overflows while scanning for integers

From: Petr Mladek <pmladek@suse.com>
To: Kees Cook <keescook@chromium.org>
Cc: "Richard Weinberger" <richard@nod.at>,
	linux-hardening@vger.kernel.org, netdev@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	"Steven Rostedt" <rostedt@goodmis.org>,
	"Sergey Senozhatsky" <senozhatsky@chromium.org>,
	"Andy Shevchenko" <andriy.shevchenko@linux.intel.com>,
	"Rasmus Villemoes" <linux@rasmusvillemoes.dk>,
	"David S. Miller" <davem@davemloft.net>,
	"Eric Dumazet" <edumazet@google.com>,
	"Jakub Kicinski" <kuba@kernel.org>,
	"Paolo Abeni" <pabeni@redhat.com>,
	"Miguel Ojeda" <ojeda@kernel.org>,
	"Alex Gaynor" <alex.gaynor@gmail.com>,
	"Wedson Almeida Filho" <wedsonaf@gmail.com>,
	"Boqun Feng" <boqun.feng@gmail.com>,
	"Gary Guo" <gary@garyguo.net>,
	"Björn Roy Baron" <bjorn3_gh@protonmail.com>,
	"Benno Lossin" <benno.lossin@proton.me>,
	"Alexei Starovoitov" <ast@kernel.org>,
	"Daniel Borkmann" <daniel@iogearbox.net>,
	"Jesper Dangaard Brouer" <hawk@kernel.org>,
	"John Fastabend" <john.fastabend@gmail.com>
Subject: Re: [RFC PATCH 0/1] Integer overflows while scanning for integers
Date: Thu, 8 Jun 2023 17:27:40 +0200	[thread overview]
Message-ID: <ZIHzbBXlxEz6As9N@alley> (raw)
In-Reply-To: <202306071634.51BBAFD14@keescook>

On Wed 2023-06-07 16:36:12, Kees Cook wrote:
> On Thu, Jun 08, 2023 at 12:37:54AM +0200, Richard Weinberger wrote:
> > Hi!
> > 
> > Lately I wondered whether users of integer scanning functions check
> > for overflows.
> > To detect such overflows around scanf I came up with the following
> > patch. It simply triggers a WARN_ON_ONCE() upon an overflow.
> > 
> > After digging into various scanf users I found that the network device
> > naming code can trigger an overflow.
> > 
> > e.g:
> > $ ip link add 1 type veth peer name 9999999999
> > $ ip link set name "%d" dev 1
> > 
> > It will trigger the following WARN_ON_ONCE():
> > ------------[ cut here ]------------
> > WARNING: CPU: 2 PID: 433 at lib/vsprintf.c:3701 vsscanf+0x6ce/0x990
> 
> Hm, it's considered a bug if a WARN or BUG can be reached from
> userspace,

Good point. WARN() does not look like the right way in this case.

Another problem is that some users use panic_on_warn. In this case,
the above "ip" command calls would trigger panic(). And it does not
look like an optimal behavior.

I know there already are some WARN_ONs for similar situations, e.g.
set_field_width() or set_precision(). But these do not get random
values. And it would actually be nice to introduce something like
INFO() that would be usable for these less serious problems where
the backtrace is useful but they should never trigger panic().

> so this probably needs to be rearranged (or callers fixed).
> Do we need to change the scanf API for sane use inside the kernel?

It seems that userspace implementation of sscanf() and vsscanf()
returns -ERANGE in this case. It might be a reasonable solution.

Well, there is a risk of introducing security problems. The error
value might cause an underflow/overflow when the caller does not expect
a negative value.

Alternative solution would be to update the "ip" code so that it
reads the number separately and treat zero return value as
-EINVAL.

Best Regards,
Petr