TCP_ULP setsockopt cannot be used for mptcp because its already
used internally to plumb subflow (tcp) sockets to the mptcp layer.
syzbot managed to trigger a crash for mptcp connections that are
in fallback mode:
KASAN: null-ptr-deref in range [0x0000000000000020-0x0000000000000027]
CPU: 1 PID: 1083 Comm: syz-executor.3 Not tainted 5.16.0-rc2-syzkaller #0
RIP: 0010:tls_build_proto net/tls/tls_main.c:776 [inline]
[..]
__tcp_set_ulp net/ipv4/tcp_ulp.c:139 [inline]
tcp_set_ulp+0x428/0x4c0 net/ipv4/tcp_ulp.c:160
do_tcp_setsockopt+0x455/0x37c0 net/ipv4/tcp.c:3391
mptcp_setsockopt+0x1b47/0x2400 net/mptcp/sockopt.c:638
Remove support for TCP_ULP setsockopt.
Reported-by: syzbot+1fd9b69cde42967d1add@syzkaller.appspotmail.com
Signed-off-by: Florian Westphal <fw@strlen.de>
---
diff --git a/net/mptcp/sockopt.c b/net/mptcp/sockopt.c
index 3c3db22fd36a..aa3fcd86dbe2 100644
--- a/net/mptcp/sockopt.c
+++ b/net/mptcp/sockopt.c
@@ -543,7 +543,6 @@ static bool mptcp_supported_sockopt(int level, int optname)
case TCP_NODELAY:
case TCP_THIN_LINEAR_TIMEOUTS:
case TCP_CONGESTION:
- case TCP_ULP:
case TCP_CORK:
case TCP_KEEPIDLE:
case TCP_KEEPINTVL:
--
2.32.0
Hi Florian,
(without netdev and syzbot ML)
On 05/12/2021 20:27, Florian Westphal wrote:
> TCP_ULP setsockopt cannot be used for mptcp because its already
> used internally to plumb subflow (tcp) sockets to the mptcp layer.
>
> syzbot managed to trigger a crash for mptcp connections that are
> in fallback mode:
>
> KASAN: null-ptr-deref in range [0x0000000000000020-0x0000000000000027]
> CPU: 1 PID: 1083 Comm: syz-executor.3 Not tainted 5.16.0-rc2-syzkaller #0
> RIP: 0010:tls_build_proto net/tls/tls_main.c:776 [inline]
> [..]
> __tcp_set_ulp net/ipv4/tcp_ulp.c:139 [inline]
> tcp_set_ulp+0x428/0x4c0 net/ipv4/tcp_ulp.c:160
> do_tcp_setsockopt+0x455/0x37c0 net/ipv4/tcp.c:3391
> mptcp_setsockopt+0x1b47/0x2400 net/mptcp/sockopt.c:638
>
> Remove support for TCP_ULP setsockopt.
Good catch!
Indeed, it doesn't make sense to support TCP_ULP.
> Reported-by: syzbot+1fd9b69cde42967d1add@syzkaller.appspotmail.com
> Signed-off-by: Florian Westphal <fw@strlen.de>
I guess we need a Fixes tag here:
Fixes: d9e4c1291810 ("mptcp: only admit explicitly supported sockopt")
BTW, did you send this patch to netdev ML on purpose?
I guess this is fine but they will probably ask for the Fixes tag as it
looks like a bug-fix. Do you want me to reply to this email on netdev
and include this "Fixes" tag in my email?
Cheers,
Matt
--
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net
Matthieu Baerts <matthieu.baerts@tessares.net> wrote: > BTW, did you send this patch to netdev ML on purpose? Yes, it was a reply to syzbot report. > I guess this is fine but they will probably ask for the Fixes tag as it > looks like a bug-fix. Do you want me to reply to this email on netdev > and include this "Fixes" tag in my email? I think it can go in via mptcp tree. This patch is fine but there is still another ULP bug left, will send a patch soon.
On Mon, 6 Dec 2021, Matthieu Baerts wrote:
> Hi Florian,
>
> (without netdev and syzbot ML)
>
> On 05/12/2021 20:27, Florian Westphal wrote:
>> TCP_ULP setsockopt cannot be used for mptcp because its already
>> used internally to plumb subflow (tcp) sockets to the mptcp layer.
>>
>> syzbot managed to trigger a crash for mptcp connections that are
>> in fallback mode:
>>
>> KASAN: null-ptr-deref in range [0x0000000000000020-0x0000000000000027]
>> CPU: 1 PID: 1083 Comm: syz-executor.3 Not tainted 5.16.0-rc2-syzkaller #0
>> RIP: 0010:tls_build_proto net/tls/tls_main.c:776 [inline]
>> [..]
>> __tcp_set_ulp net/ipv4/tcp_ulp.c:139 [inline]
>> tcp_set_ulp+0x428/0x4c0 net/ipv4/tcp_ulp.c:160
>> do_tcp_setsockopt+0x455/0x37c0 net/ipv4/tcp.c:3391
>> mptcp_setsockopt+0x1b47/0x2400 net/mptcp/sockopt.c:638
>>
>> Remove support for TCP_ULP setsockopt.
>
> Good catch!
> Indeed, it doesn't make sense to support TCP_ULP.
>
>> Reported-by: syzbot+1fd9b69cde42967d1add@syzkaller.appspotmail.com
>> Signed-off-by: Florian Westphal <fw@strlen.de>
>
> I guess we need a Fixes tag here:
>
> Fixes: d9e4c1291810 ("mptcp: only admit explicitly supported sockopt")
>
Thanks Florian.
Matthieu, could you apply this to the export branch with your suggested
Fixes tag?
Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
--
Mat Martineau
Intel
Hi Mat,
On 06/12/2021 19:47, Mat Martineau wrote:
> On Mon, 6 Dec 2021, Matthieu Baerts wrote:
>
>> Hi Florian,
>>
>> (without netdev and syzbot ML)
>>
>> On 05/12/2021 20:27, Florian Westphal wrote:
>>> TCP_ULP setsockopt cannot be used for mptcp because its already
>>> used internally to plumb subflow (tcp) sockets to the mptcp layer.
>>>
>>> syzbot managed to trigger a crash for mptcp connections that are
>>> in fallback mode:
>>>
>>> KASAN: null-ptr-deref in range [0x0000000000000020-0x0000000000000027]
>>> CPU: 1 PID: 1083 Comm: syz-executor.3 Not tainted
>>> 5.16.0-rc2-syzkaller #0
>>> RIP: 0010:tls_build_proto net/tls/tls_main.c:776 [inline]
>>> [..]
>>> __tcp_set_ulp net/ipv4/tcp_ulp.c:139 [inline]
>>> tcp_set_ulp+0x428/0x4c0 net/ipv4/tcp_ulp.c:160
>>> do_tcp_setsockopt+0x455/0x37c0 net/ipv4/tcp.c:3391
>>> mptcp_setsockopt+0x1b47/0x2400 net/mptcp/sockopt.c:638
>>>
>>> Remove support for TCP_ULP setsockopt.
>>
>> Good catch!
>> Indeed, it doesn't make sense to support TCP_ULP.
>>
>>> Reported-by: syzbot+1fd9b69cde42967d1add@syzkaller.appspotmail.com
>>> Signed-off-by: Florian Westphal <fw@strlen.de>
>>
>> I guess we need a Fixes tag here:
>>
>> Fixes: d9e4c1291810 ("mptcp: only admit explicitly supported sockopt")
>>
>
> Thanks Florian.
>
> Matthieu, could you apply this to the export branch with your suggested
> Fixes tag?
>
> Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Sure, just did:
- 13dc812544f7: mptcp: remove tcp ulp setsockopt support
- Results: 5fd3630ba52a..4be3d8d5b45c
Builds and tests are now in progress:
https://cirrus-ci.com/github/multipath-tcp/mptcp_net-next/export/20211207T132312
https://github.com/multipath-tcp/mptcp_net-next/actions/workflows/build-validation.yml?query=branch:export
Cheers,
Matt
--
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net
On Sun, 5 Dec 2021 20:27:00 +0100 Florian Westphal wrote: > TCP_ULP setsockopt cannot be used for mptcp because its already > used internally to plumb subflow (tcp) sockets to the mptcp layer. > > syzbot managed to trigger a crash for mptcp connections that are > in fallback mode: Fallback mode meaning ops are NULL? I'm slightly confused by this report. > KASAN: null-ptr-deref in range [0x0000000000000020-0x0000000000000027] > CPU: 1 PID: 1083 Comm: syz-executor.3 Not tainted 5.16.0-rc2-syzkaller #0 > RIP: 0010:tls_build_proto net/tls/tls_main.c:776 [inline] > [..] > __tcp_set_ulp net/ipv4/tcp_ulp.c:139 [inline] > tcp_set_ulp+0x428/0x4c0 net/ipv4/tcp_ulp.c:160 > do_tcp_setsockopt+0x455/0x37c0 net/ipv4/tcp.c:3391 > mptcp_setsockopt+0x1b47/0x2400 net/mptcp/sockopt.c:638 > > Remove support for TCP_ULP setsockopt. > > Reported-by: syzbot+1fd9b69cde42967d1add@syzkaller.appspotmail.com > Signed-off-by: Florian Westphal <fw@strlen.de>
On Mon, 6 Dec 2021 07:53:26 -0800 Jakub Kicinski wrote: > On Sun, 5 Dec 2021 20:27:00 +0100 Florian Westphal wrote: > > TCP_ULP setsockopt cannot be used for mptcp because its already > > used internally to plumb subflow (tcp) sockets to the mptcp layer. > > > > syzbot managed to trigger a crash for mptcp connections that are > > in fallback mode: > > Fallback mode meaning ops are NULL? I'm slightly confused by this > report. Ah, it's the socket not the ops. > > KASAN: null-ptr-deref in range [0x0000000000000020-0x0000000000000027] > > CPU: 1 PID: 1083 Comm: syz-executor.3 Not tainted 5.16.0-rc2-syzkaller #0 > > RIP: 0010:tls_build_proto net/tls/tls_main.c:776 [inline] > > [..] > > __tcp_set_ulp net/ipv4/tcp_ulp.c:139 [inline] > > tcp_set_ulp+0x428/0x4c0 net/ipv4/tcp_ulp.c:160 > > do_tcp_setsockopt+0x455/0x37c0 net/ipv4/tcp.c:3391 > > mptcp_setsockopt+0x1b47/0x2400 net/mptcp/sockopt.c:638 > > > > Remove support for TCP_ULP setsockopt. > > > > Reported-by: syzbot+1fd9b69cde42967d1add@syzkaller.appspotmail.com > > Signed-off-by: Florian Westphal <fw@strlen.de>
On Mon, 6 Dec 2021, Jakub Kicinski wrote: > On Mon, 6 Dec 2021 07:53:26 -0800 Jakub Kicinski wrote: >> On Sun, 5 Dec 2021 20:27:00 +0100 Florian Westphal wrote: >>> TCP_ULP setsockopt cannot be used for mptcp because its already >>> used internally to plumb subflow (tcp) sockets to the mptcp layer. >>> >>> syzbot managed to trigger a crash for mptcp connections that are >>> in fallback mode: >> >> Fallback mode meaning ops are NULL? I'm slightly confused by this >> report. > > Ah, it's the socket not the ops. > >>> KASAN: null-ptr-deref in range [0x0000000000000020-0x0000000000000027] >>> CPU: 1 PID: 1083 Comm: syz-executor.3 Not tainted 5.16.0-rc2-syzkaller #0 >>> RIP: 0010:tls_build_proto net/tls/tls_main.c:776 [inline] >>> [..] >>> __tcp_set_ulp net/ipv4/tcp_ulp.c:139 [inline] >>> tcp_set_ulp+0x428/0x4c0 net/ipv4/tcp_ulp.c:160 >>> do_tcp_setsockopt+0x455/0x37c0 net/ipv4/tcp.c:3391 >>> mptcp_setsockopt+0x1b47/0x2400 net/mptcp/sockopt.c:638 >>> >>> Remove support for TCP_ULP setsockopt. >>> >>> Reported-by: syzbot+1fd9b69cde42967d1add@syzkaller.appspotmail.com >>> Signed-off-by: Florian Westphal <fw@strlen.de> Jakub - If you could mark this as "Not Applicable" in netdevbpf patchwork, we'll apply it to the mptcp tree and resubmit to netdev with some related patches. Thanks, -- Mat Martineau Intel
On Mon, 6 Dec 2021 10:44:16 -0800 (PST) Mat Martineau wrote: > If you could mark this as "Not Applicable" in netdevbpf patchwork, we'll > apply it to the mptcp tree and resubmit to netdev with some related > patches. For sure, I was asking because technically I added the code that hit the oops to tls a week or so ago, so I wanted to make sure it's not my fault ;) Marked in pw appropriately now.
© 2016 - 2026 Red Hat, Inc.