Re: [PATCH net v2 1/2] net: Fix UDP length overflow with PMTU discover and big MTU

On Thu, Aug 13, 2026, at 23:11, Willem de Bruijn wrote:
> Alice Mikityanska wrote:
>> From: Alice Mikityanska <[email protected]>
>> 
>> This commit bounds cork->base.fragsize to IP(6)_MAX_MTU to avoid a
>> possible overflow of UDP length that triggers a WARN in
>> udp_set_len_short when setsockopt IP(V6)_MTU_DISCOVER is set to
>> IPV6_PMTUDISC_DO or IP(V6)_PMTUDISC_PROBE, and a large packet is sent
>> over a netdev with an unusually large MTU.
>> 
>> Steps to reproduce (included in the new selftest):
>> 
>> 1. Set device MTU bigger than IP6_MAX_MTU (or IP_MAX_MTU + 20).
>>    cork->base.fragsize will be set to that MTU in ip(6)_setup_cork.
>> 2. Set IP(V6)_MTU_DISCOVER to IP(V6)_PMTUDISC_PROBE or IPV6_PMTUDISC_DO.
>>    It lets maxnonfragsize be set to device MTU (cork->fragsize) in
>>    __ip(6)_append_data, rather than to IP(6)_MAX_MTU.
>> 3. Send 65528 bytes of payload (+8 bytes of UDP header, +20/40 bytes of
>>    IPv4/IPv6 header). Device MTU allows it (it's only one byte bigger
>>    than IP6_MAX_MTU or IP_MAX_MTU + IPv4 header, and the device MTU is
>>    bigger than that).
>> 4. The UDP length in the built packet is 65536, which overflows the
>>    16-bit length field and triggers the WARN in udp_set_len_short.
>> 
>> Note: IP_PMTUDISC_DO with IPv4 is safe, because ip_dst_mtu_maybe_forward
>> always clamps at IP_MAX_MTU, unlike ip6_dst_mtu_maybe_forward.
>> 
>> The Fixes tag points at the first commit where I could reproduce the
>> overflow with IPv4 and IP_PMTUDISC_PROBE.
>> 
>> Fixes: daba287b299e ("ipv4: fix DO and PROBE pmtu mode regarding local fragmentation with UFO/CORK")
>> Reported-by: [email protected]
>> Closes: https://lore.kernel.org/netdev/[email protected]/
>> Signed-off-by: Alice Mikityanska <[email protected]>
>> Assisted-by: Claude:claude-sonnet-4.6
>> Cc: Willem de Bruijn <[email protected]>
>
> Not sure whether IPv4 and IPv6 should be separate patches.
> IPv6 will have a different Fixes tag.
>
> Per maintainer-netdev.rst, mixing code and test changes in the same
> patch is also discouraged.
>
>> ---
>>  net/ipv4/ip_output.c                         |  2 +
>>  net/ipv6/ip6_output.c                        |  2 +-
>>  tools/testing/selftests/net/Makefile         |  1 +
>>  tools/testing/selftests/net/cork_fragsize.py | 77 ++++++++++++++++++++
>>  4 files changed, 81 insertions(+), 1 deletion(-)
>>  create mode 100755 tools/testing/selftests/net/cork_fragsize.py
>> 
>> diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
>> index e6dd1e5b8c32..2bc997c4fa7b 100644
>> --- a/net/ipv4/ip_output.c
>> +++ b/net/ipv4/ip_output.c
>> @@ -1299,6 +1299,8 @@ static int ip_setup_cork(struct sock *sk, struct inet_cork *cork,
>>  
>>  	cork->fragsize = ip_sk_use_pmtu(sk) ?
>>  			 dst4_mtu(&rt->dst) : READ_ONCE(rt->dst.dev->mtu);
>> +	if (cork->fragsize > IP_MAX_MTU)
>> +		cork->fragsize = IP_MAX_MTU;
>
> if respinning: use min(), like IPv6? It's even more concise and easier to read.
>
>>  
>>  	if (!inetdev_valid_mtu(cork->fragsize))
>>  		return -ENETUNREACH;
>> diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
>> index 2c44e5ed6171..81fb61b94759 100644
>> --- a/net/ipv6/ip6_output.c
>> +++ b/net/ipv6/ip6_output.c
>> @@ -1430,7 +1430,7 @@ static int ip6_setup_cork(struct sock *sk, struct inet_cork_full *cork,
>>  	if (frag_size && frag_size < mtu)
>>  		mtu = frag_size;
>>  
>> -	cork->base.fragsize = mtu;
>> +	cork->base.fragsize = min(mtu, IP6_MAX_MTU);
>
> Not sure whether a limitation that we seriously have to care about, but:
>
> This restricts all IPv6 protocols, while only UDP has the length field
> limitation.
>
> With IPv6 jumbo frames and RAW sockets, could there be a valid case
> for larger fragsize?

True; if a valid jumbo UDP (length=0, with IPv6 HBH exthdr) packet is
crafted manually with SOCK_RAW, it's dropped after my change. I'll
respin, limiting the clamp to UDP sockets only.

I'll also split IPv4/IPv6/selftest into separate patches, also adding a
selftest for SOCK_RAW.

>>  	cork->base.gso_size = ipc6->gso_size;
>>  	cork->base.tx_flags = 0;
>>  	cork->base.mark = ipc6->sockc.mark;
>> diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile
>> index 708d960ae07d..82a1407f6a69 100644
>> --- a/tools/testing/selftests/net/Makefile
>> +++ b/tools/testing/selftests/net/Makefile
>> @@ -24,6 +24,7 @@ TEST_PROGS := \
>>  	cmsg_so_mark.sh \
>>  	cmsg_so_priority.sh \
>>  	cmsg_time.sh \
>> +	cork_fragsize.py \
>>  	double_udp_encap.sh \
>>  	drop_monitor_tests.sh \
>>  	ecmp_rehash.sh \
>> diff --git a/tools/testing/selftests/net/cork_fragsize.py b/tools/testing/selftests/net/cork_fragsize.py
>> new file mode 100755
>> index 000000000000..7a1bb069227a
>> --- /dev/null
>> +++ b/tools/testing/selftests/net/cork_fragsize.py
>> @@ -0,0 +1,77 @@
>> +#!/usr/bin/env python3
>> +# SPDX-License-Identifier: GPL-2.0
>> +
>> +# Test possible UDP length overflow in udp_send_skb/udp_v6_send_skb.
>> +
>> +from lib.py import ksft_run, ksft_exit, ksft_true
>> +from lib.py import ip, NetNS, NetNSEnter
>> +import errno
>> +import socket
>> +import subprocess
>> +
>> +
>> +IP_MTU_DISCOVER = 10
>> +IP_PMTUDISC_PROBE = 3
>> +IPV6_MTU_DISCOVER = 23
>> +IPV6_PMTUDISC_DO = 2
>> +
>> +
>> +def check_dmesg_clean(func) -> bool:
>> +    dmesg = subprocess.Popen(['dmesg'], stdout=subprocess.PIPE)
>> +    result = subprocess.run(['grep', '-q', f'WARNING:.*{func}'], stdin=dmesg.stdout)
>> +    dmesg.wait()
>> +    return result.returncode != 0 and dmesg.returncode == 0
>> +
>> +
>> +def test_ipv6() -> None:
>> +    with NetNS() as ns:
>> +        ip('link add dummy type dummy', ns=ns)
>> +        ip('link set dummy mtu 65576', ns=ns)
>> +        ip('link set dummy up', ns=ns)
>> +        ip('-6 addr add fd00::1/64 dev dummy nodad', ns=ns)
>> +        ip('-6 neigh add fd00::2 lladdr 02:00:00:00:00:02 dev dummy nud permanent', ns=ns)
>> +
>> +        with NetNSEnter(ns):
>> +            with socket.socket(socket.AF_INET6, socket.SOCK_DGRAM) as fd:
>> +                fd.setsockopt(socket.IPPROTO_IPV6, IPV6_MTU_DISCOVER, IPV6_PMTUDISC_DO)
>> +                try:
>> +                    fd.sendto(b' ' * 65528, ('fd00::2', 1234))
>> +                except OSError as e:
>> +                    # Ignore EMSGSIZE: it happens on kernels with the fix.
>> +                    if e.errno != errno.EMSGSIZE:
>> +                        raise
>> +
>> +        ip('link del dummy', ns=ns)
>> +
>> +    ksft_true(check_dmesg_clean('udp_v6_send_skb'), 'WARNING detected in dmesg')
>> +
>> +
>> +def test_ipv4() -> None:
>> +    with NetNS() as ns:
>> +        ip('link add dummy type dummy', ns=ns)
>> +        ip('link set dummy mtu 65556', ns=ns)
>> +        ip('link set dummy up', ns=ns)
>> +        ip('addr add 10.0.0.1/24 dev dummy', ns=ns)
>> +        ip('neigh add 10.0.0.2 lladdr 02:00:00:00:00:02 dev dummy nud permanent', ns=ns)
>> +
>> +        with NetNSEnter(ns):
>> +            with socket.socket(socket.AF_INET, socket.SOCK_DGRAM) as fd:
>> +                fd.setsockopt(socket.IPPROTO_IP, IP_MTU_DISCOVER, IP_PMTUDISC_PROBE)
>> +                try:
>> +                    fd.sendto(b' ' * 65528, ('10.0.0.2', 1234))
>> +                except OSError as e:
>> +                    # Ignore EMSGSIZE: the check happens after the WARN is printed.
>> +                    if e.errno != errno.EMSGSIZE:
>> +                        raise
>> +
>> +        ip('link del dummy', ns=ns)
>> +
>> +    ksft_true(check_dmesg_clean('udp_send_skb'), 'WARNING detected in dmesg')
>> +
>> +
>> +if __name__ == "__main__":
>> +    ksft_run([
>> +        test_ipv6,
>> +        test_ipv4,
>> +    ])
>> +    ksft_exit()
>> -- 
>> 2.55.0
>>