Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
Merge branch 'mptcp-add-tcp_maxseg-sockopt-support'
Matthieu Baerts says:
====================
mptcp: add TCP_MAXSEG sockopt support
The TCP_MAXSEG socket option was not supported by MPTCP, mainly because
it has never been requested before. But there are still valid use-cases,
e.g. with HAProxy.
- Patch 1 is a small cleanup patch in the MPTCP sockopt file.
- Patch 2 expose some code from TCP, to avoid duplicating it in MPTCP.
- Patch 3 adds TCP_MAXSEG sockopt support in MPTCP.
- Patch 4 is not related to the others, it fixes a typo in a comment.
Note that the new TCP_MAXSEG sockopt support has been validated by a new
packetdrill script on the MPTCP CI:
https://github.com/multipath-tcp/packetdrill/pull/161
v1: https://lore.kernel.org/20250716-net-next-mptcp-tcp_maxseg-v1-0-548d3a5666f6@kernel.org
====================
Link: https://patch.msgid.link/20250719-net-next-mptcp-tcp_maxseg-v2-0-8c910fbc5307@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
+46
-14
+1
include/linux/tcp.h
+1
include/linux/tcp.h
···
621
621
void tcp_sock_set_quickack(struct sock *sk, int val);
622
622
int tcp_sock_set_syncnt(struct sock *sk, int val);
623
623
int tcp_sock_set_user_timeout(struct sock *sk, int val);
624
+
int tcp_sock_set_maxseg(struct sock *sk, int val);
624
625
625
626
static inline bool dst_tcp_usec_ts(const struct dst_entry *dst)
626
627
{
+14
-9
net/ipv4/tcp.c
+14
-9
net/ipv4/tcp.c
···
3751
3751
return 0;
3752
3752
}
3753
3753
3754
+
int tcp_sock_set_maxseg(struct sock *sk, int val)
3755
+
{
3756
+
/* Values greater than interface MTU won't take effect. However
3757
+
* at the point when this call is done we typically don't yet
3758
+
* know which interface is going to be used
3759
+
*/
3760
+
if (val && (val < TCP_MIN_MSS || val > MAX_TCP_WINDOW))
3761
+
return -EINVAL;
3762
+
3763
+
tcp_sk(sk)->rx_opt.user_mss = val;
3764
+
return 0;
3765
+
}
3766
+
3754
3767
/*
3755
3768
* Socket option code for TCP.
3756
3769
*/
···
3896
3883
3897
3884
switch (optname) {
3898
3885
case TCP_MAXSEG:
3899
-
/* Values greater than interface MTU won't take effect. However
3900
-
* at the point when this call is done we typically don't yet
3901
-
* know which interface is going to be used
3902
-
*/
3903
-
if (val && (val < TCP_MIN_MSS || val > MAX_TCP_WINDOW)) {
3904
-
err = -EINVAL;
3905
-
break;
3906
-
}
3907
-
tp->rx_opt.user_mss = val;
3886
+
err = tcp_sock_set_maxseg(sk, val);
3908
3887
break;
3909
3888
3910
3889
case TCP_NODELAY:
+1
-1
net/mptcp/protocol.c
+1
-1
net/mptcp/protocol.c
···
1387
1387
* - estimate the faster flow linger time
1388
1388
* - use the above to estimate the amount of byte transferred
1389
1389
* by the faster flow
1390
-
* - check that the amount of queued data is greter than the above,
1390
+
* - check that the amount of queued data is greater than the above,
1391
1391
* otherwise do not use the picked, slower, subflow
1392
1392
* We select the subflow with the shorter estimated time to flush
1393
1393
* the queued mem, which basically ensure the above. We just need
+1
net/mptcp/protocol.h
+1
net/mptcp/protocol.h
+29
-4
net/mptcp/sockopt.c
+29
-4
net/mptcp/sockopt.c
···
798
798
return ret;
799
799
}
800
800
801
+
static int mptcp_setsockopt_all_sf(struct mptcp_sock *msk, int level,
802
+
int optname, sockptr_t optval,
803
+
unsigned int optlen)
804
+
{
805
+
struct mptcp_subflow_context *subflow;
806
+
int ret = 0;
807
+
808
+
mptcp_for_each_subflow(msk, subflow) {
809
+
struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
810
+
811
+
ret = tcp_setsockopt(ssk, level, optname, optval, optlen);
812
+
if (ret)
813
+
break;
814
+
}
815
+
return ret;
816
+
}
817
+
801
818
static int mptcp_setsockopt_sol_tcp(struct mptcp_sock *msk, int optname,
802
819
sockptr_t optval, unsigned int optlen)
803
820
{
···
876
859
&msk->keepalive_cnt,
877
860
val);
878
861
break;
862
+
case TCP_MAXSEG:
863
+
msk->maxseg = val;
864
+
ret = mptcp_setsockopt_all_sf(msk, SOL_TCP, optname, optval,
865
+
optlen);
866
+
break;
879
867
default:
880
868
ret = -ENOPROTOOPT;
881
869
}
···
936
914
937
915
lock_sock(sk);
938
916
ssk = msk->first;
939
-
if (ssk) {
940
-
ret = tcp_getsockopt(ssk, level, optname, optval, optlen);
941
-
goto out;
942
-
}
917
+
if (ssk)
918
+
goto get;
943
919
944
920
ssk = __mptcp_nmpc_sk(msk);
945
921
if (IS_ERR(ssk)) {
···
945
925
goto out;
946
926
}
947
927
928
+
get:
948
929
ret = tcp_getsockopt(ssk, level, optname, optval, optlen);
949
930
950
931
out:
···
1428
1407
return mptcp_put_int_option(msk, optval, optlen, msk->notsent_lowat);
1429
1408
case TCP_IS_MPTCP:
1430
1409
return mptcp_put_int_option(msk, optval, optlen, 1);
1410
+
case TCP_MAXSEG:
1411
+
return mptcp_getsockopt_first_sf_only(msk, SOL_TCP, optname,
1412
+
optval, optlen);
1431
1413
}
1432
1414
return -EOPNOTSUPP;
1433
1415
}
···
1577
1553
tcp_sock_set_keepidle_locked(ssk, msk->keepalive_idle);
1578
1554
tcp_sock_set_keepintvl(ssk, msk->keepalive_intvl);
1579
1555
tcp_sock_set_keepcnt(ssk, msk->keepalive_cnt);
1556
+
tcp_sock_set_maxseg(ssk, msk->maxseg);
1580
1557
1581
1558
inet_assign_bit(TRANSPARENT, ssk, inet_test_bit(TRANSPARENT, sk));
1582
1559
inet_assign_bit(FREEBIND, ssk, inet_test_bit(FREEBIND, sk));