Re: [PATCH 5.4 10/36] bpf: Fix tnum constraints for 32-bit comparisons

From: Daniel Borkmann <daniel@iogearbox.net>
To: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	linux-kernel@vger.kernel.org
Cc: stable@vger.kernel.org, Jann Horn <jannh@google.com>,
	Alexei Starovoitov <ast@kernel.org>,
	Sasha Levin <sashal@kernel.org>,
	bpf@vger.kernel.org
Subject: Re: [PATCH 5.4 10/36] bpf: Fix tnum constraints for 32-bit comparisons
Date: Tue, 7 Apr 2020 12:45:23 +0200	[thread overview]
Message-ID: <26e2a116-bc4c-59b2-7c54-6ebbfb140ea5@iogearbox.net> (raw)
In-Reply-To: <20200407101455.655552813@linuxfoundation.org>

Hey Sasha, hey Greg,

On 4/7/20 12:21 PM, Greg Kroah-Hartman wrote:
> From: Jann Horn <jannh@google.com>
> 
> [ Upstream commit 604dca5e3af1db98bd123b7bfc02b017af99e3a0 ]
> 
> The BPF verifier tried to track values based on 32-bit comparisons by
> (ab)using the tnum state via 581738a681b6 ("bpf: Provide better register
> bounds after jmp32 instructions"). The idea is that after a check like
> this:
> 
>      if ((u32)r0 > 3)
>        exit
> 
> We can't meaningfully constrain the arithmetic-range-based tracking, but
> we can update the tnum state to (value=0,mask=0xffff'ffff'0000'0003).
> However, the implementation from 581738a681b6 didn't compute the tnum
> constraint based on the fixed operand, but instead derives it from the
> arithmetic-range-based tracking. This means that after the following
> sequence of operations:
> 
>      if (r0 >= 0x1'0000'0001)
>        exit
>      if ((u32)r0 > 7)
>        exit
> 
> The verifier assumed that the lower half of r0 is in the range (0, 0)
> and apply the tnum constraint (value=0,mask=0xffff'ffff'0000'0000) thus
> causing the overall tnum to be (value=0,mask=0x1'0000'0000), which was
> incorrect. Provide a fixed implementation.
> 
> Fixes: 581738a681b6 ("bpf: Provide better register bounds after jmp32 instructions")
> Signed-off-by: Jann Horn <jannh@google.com>
> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
> Signed-off-by: Alexei Starovoitov <ast@kernel.org>
> Link: https://lore.kernel.org/bpf/20200330160324.15259-3-daniel@iogearbox.net
> Signed-off-by: Sasha Levin <sashal@kernel.org>

We've already addressed this issue (CVE-2020-8835) on 5.4/5.5/5.6 kernels through
the following backports:

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-5.4.y&id=8d62a8c7489a68b5738390b008134a644aa9b383
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-5.5.y&id=0ebc01466d98d016eb6a3780ec8edb0c86fa48bc
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-5.6.y&id=6797143df51c8ae259aa4bfe4e99c832b20bde8a

Given the severity of the issue, we concluded that revert-only is the best and
most straight forward way to address it for stable.

Was this selected via Sasha's ML mechanism? Should there be a commit tag to opt-out
for some commits being selected? E.g. this one 581738a681b6 ("bpf: Provide better
register bounds after jmp32 instructions") already fell through our radar and wrongly
made its way into 5.4 where it should have never landed. :/

Thanks,
Daniel