Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <CAEHB2488dNqBKcgWLSeq500JLC1+q6RV=ENcUPm=rN9bWf0QkQ@mail.gmail.com>
 <20220819123542.GA2461@breakpoint.cc>
In-Reply-To: <20220819123542.GA2461@breakpoint.cc>
From:   Gabriel Ryan <gabe@cs.columbia.edu>
Date:   Mon, 22 Aug 2022 16:29:01 -0400
Message-ID: <CALbthtdzW-_4TVngjt-VjCS6GqCEP967-UE7oEoDkBAVaRFOzw@mail.gmail.com>
Subject: Re: data-race in nf_tables_newtable / nf_tables_newtable
To:     Florian Westphal <fw@strlen.de>
Cc:     Abhishek Shah <abhishek.shah@columbia.edu>, coreteam@netfilter.org,
        davem@davemloft.net, edumazet@google.com, kadlec@netfilter.org,
        kuba@kernel.org, netdev@vger.kernel.org,
        netfilter-devel@vger.kernel.org, pabeni@redhat.com,
        pablo@netfilter.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Hi Florian,

I just looked at the lock event trace from our report and it looks
like two distinct commit mutexes were held when the race was
triggered. I think the race is probably on the table_handle variable
on net/netfilter/nf_tables_api.c:1221, and not the table->handle field
being written to.

Racing increments to table_handle could cause it to either overcount
or undercount. Could that be an issue?

Best,

Gabe

On Fri, Aug 19, 2022 at 8:35 AM Florian Westphal <fw@strlen.de> wrote:
>
> Abhishek Shah <abhishek.shah@columbia.edu> wrote:
> > Hi all,
> >
> > We found a race involving the table->handle variable here
> > <https://urldefense.proofpoint.com/v2/url?u=3Dhttps-3A__elixir.bootlin.=
com_linux_v5.18-2Drc5_source_net_netfilter_nf-5Ftables-5Fapi.c-23L1221&d=3D=
DwIBAg&c=3D009klHSCxuh5AI1vNQzSO0KGjl4nbi2Q0M1QLJX9BeE&r=3DEyAJYRJu01oaAhhV=
VY3o8zKgZvacDAXd_PNRtaqACCo&m=3DxlZC-wDg7fkTm6_4HfcoDqYfJx_OU2L5HHX2q_yTYZZ=
CEDCFAg-9I7T1gNmXPISg&s=3DJYkSOriQVx_3lJhAzBo7yqhe4bnf2Sy96cPL0L1NIn8&e=3D =
 >.
> > This race advances the pointer, which can cause out-of-bounds memory
> > accesses in the future. Please let us know what you think.
> >
> > Thanks!
> >
> >
> > *---------------------Report-----------------*
> > *read-write* to 0xffffffff883a01e8 of 8 bytes by task 6542 on cpu 0:
> >  nf_tables_newtable+0x6dc/0xc00 net/netfilter/nf_tables_api.c:1221
> >  nfnetlink_rcv_batch net/netfilter/nfnetlink.c:513 [inline]
>
> [..]
>
> > *read-write* to 0xffffffff883a01e8 of 8 bytes by task 6541 on cpu 1:
> >  nf_tables_newtable+0x6dc/0xc00 net/netfilter/nf_tables_api.c:1221
> >  nfnetlink_rcv_batch net/netfilter/nfnetlink.c:513 [inline]
>
> [..]
>
> I don't understand.  Like all batch operations, nf_tables_newtable is
> supposed to run with the transaction mutex held, i.e. parallel execution
> is not expected.
>
> There is a lockdep assertion at start of nf_tables_newtable(); I
> don't see how its possible that two threads can run this concurrently.

--=20
Gabriel Ryan
PhD Candidate at Columbia University