mptcp: adjust tcp rcvspace after moving skbs from ssk to sk queue
Subsystems (2)
Florian Westphal
PATCH v3 1/2 mptcp: adjust tcp rcvspace after moving skbs from ssk to sk queue
TCP does tcp rcvbuf tuning when copying packets to userspace, e.g. in tcp_read_sock(). In case of mptcp, that function is only rarely used (when discarding retransmitted duplicate data).
Instead, skbs are moved from the tcp rx queue to the mptcp socket rx queue.
Adjust subflow rcvbuf when we do so, its the last spot where we can adjust the ssk rcvbuf -- later we only have the mptcp-level socket.
This greatly improves performance on mptcp bulk transfers.
--- net/mptcp/protocol.c | 4 ++++ 1 file changed, 4 insertions(+)
net/mptcp/protocol.c +4 -0
@@ -248,6 +248,9 @@ static bool __mptcp_move_skbs_from_subflow(struct mptcp_sock *msk, *bytes = moved; +if (moved) +tcp_rcv_space_adjust (ssk); +return done; }
@@ -1263,6 +1266,7 @@ static int mptcp_init_sock(struct sock *sk)return ret;sk_sockets_allocated_inc (sk); + sk->sk_rcvbuf =sock_net (sk)->ipv4.sysctl_tcp_rmem[1 ]; sk->sk_sndbuf =sock_net (sk)->ipv4.sysctl_tcp_wmem[2 ];return 0 ;
Signature · 1 line
2.26.2
Florian Westphal
PATCH v3 2/2 mptcp: move recbuf adjustment to recvmsg path
From: Paolo Abeni <pabeni(a)redhat.com>
Place receive window tuning in the recvmsg path.
This makes sure the size is only increased when userspace consumes data.
Previously we would grow the sk receive buffer towards tcp_rmem[2], now we so only if userspace reads data.
Simply adjust the msk rcvbuf size to the largest receive buffer of any of the existing subflows.
---
changes in v3:
- xmas tree fmt
- move SOC_RCVBUF_LOCK test into mptcp_rcv_space_adjust()
net/mptcp/protocol.c | 32 ++++++++++++++++++++++++-------- 1 file changed, 24 insertions(+), 8 deletions(-)
net/mptcp/protocol.c +24 -8
@@ -190,13 +190,6 @@ static bool __mptcp_move_skbs_from_subflow(struct mptcp_sock *msk,return false; } -if (!(sk->sk_userlocks &SOCK_RCVBUF_LOCK )) { -int rcvbuf =max (ssk->sk_rcvbuf, sk->sk_rcvbuf); - -if (rcvbuf > sk->sk_rcvbuf) - sk->sk_rcvbuf = rcvbuf; - } - tp =tcp_sk (ssk); do { u32 map_remaining, offset;
@@ -933,6 +926,25 @@ static bool __mptcp_move_skbs(struct mptcp_sock *msk)return moved >0 ; } +static void mptcp_rcv_space_adjust (struct mptcp_sock *msk) +{ +const struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk; +const struct sock *ssk; +int rcvbuf =0 ; + +if (sk->sk_userlocks &SOCK_RCVBUF_LOCK ) +return ; + +mptcp_for_each_subflow (msk, subflow) { + ssk =mptcp_subflow_tcp_sock (subflow); + rcvbuf =max (ssk->sk_rcvbuf, rcvbuf); + } + +if (rcvbuf) + sk->sk_rcvbuf = rcvbuf; +} +static int mptcp_recvmsg (struct sock *sk,struct msghdr *msg,size_t len,int nonblock,int flags,int *addr_len) {
@@ -962,6 +974,8 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, target =sock_rcvlowat (sk, flags &MSG_WAITALL , len);__mptcp_flush_join_list (msk); +mptcp_rcv_space_adjust (msk); +while (len > (size_t )copied) {int bytes_read;
@@ -975,8 +989,10 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, copied += bytes_read;if (skb_queue_empty (&sk->sk_receive_queue) && -__mptcp_move_skbs (msk)) +__mptcp_move_skbs (msk)) { +mptcp_rcv_space_adjust (msk);continue ; + }/* only the master socket status is relevant here. The exit * conditions mirror closely tcp_recvmsg()
Signature · 1 line
2.26.2
Paolo Abeni
[MPTCP] Re: [PATCH v3 mptcp-next 2/2] mptcp: move recbuf adjustment to recvmsg path
Quoted 76 lines · Florian Westphal
From: Paolo Abeni <pabeni(a)redhat.com>
Place receive window tuning in the recvmsg path.
This makes sure the size is only increased when userspace consumes data.Previously we would grow the sk receive buffer towards tcp_rmem[2], now we so only if userspace reads data.
Simply adjust the msk rcvbuf size to the largest receive buffer of any of the existing subflows.
Signed-off-by Paolo Abeni <pabeni(a)redhat.com>Signed-off-by Florian Westphal <fw(a)strlen.de>---
changes in v3:
- xmas tree fmt
- moveSOC_RCVBUF_LOCKtest intomptcp_rcv_space_adjust()net/mptcp/protocol.c | 32 ++++++++++++++++++++++++-------- 1 file changed, 24 insertions(+), 8 deletions(-)net/mptcp/protocol.c +21
-7@@ -190,13 +190,6 @@ static bool __mptcp_move_skbs_from_subflow(struct mptcp_sock *msk,return false; } -if (!(sk->sk_userlocks &SOCK_RCVBUF_LOCK )) { -int rcvbuf =max (ssk->sk_rcvbuf, sk->sk_rcvbuf); - -if (rcvbuf > sk->sk_rcvbuf) - sk->sk_rcvbuf = rcvbuf; - } - tp =tcp_sk (ssk); do { u32 map_remaining, offset;@@ -933,6 +926,25 @@ static bool __mptcp_move_skbs(struct mptcp_sock *msk)return moved >0 ; } +static void mptcp_rcv_space_adjust (struct mptcp_sock *msk) +{ +const struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk; +const struct sock *ssk; +int rcvbuf =0 ; + +if (sk->sk_userlocks &SOCK_RCVBUF_LOCK ) +return ; + +mptcp_for_each_subflow (msk, subflow) { + ssk =mptcp_subflow_tcp_sock (subflow); + rcvbuf =max (ssk->sk_rcvbuf, rcvbuf); + } + +if (rcvbuf) + sk->sk_rcvbuf = rcvbuf; +} +static int mptcp_recvmsg (struct sock *sk,struct msghdr *msg,size_t len,int nonblock,int flags,int *addr_len) {@@ -962,6 +974,8 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, target =sock_rcvlowat (sk, flags &MSG_WAITALL , len);__mptcp_flush_join_list (msk); +mptcp_rcv_space_adjust (msk); +while (len > (size_t )copied) {int bytes_read;@@ -975,8 +989,10 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, copied += bytes_read;
net/mptcp/protocol.c static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, · line 998
Quoted 10 lines
net/mptcp/protocol.c +3
-1if (skb_queue_empty (&sk->sk_receive_queue) && -__mptcp_move_skbs (msk)) +__mptcp_move_skbs (msk)) { +mptcp_rcv_space_adjust (msk);continue ; + }/* only the master socket status is relevant here. The exit * conditions mirror closely tcp_recvmsg()
LGTM!
Thank you Florian to address the concerns of my twisted mind ;)