From: Paolo Abeni Date: Thu, 11 Feb 2021 23:30:41 +0000 (-0800) Subject: mptcp: better msk receive window updates X-Git-Tag: v5.15~1635^2~21^2~28^2~1 X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=e3859603ba13e7545372b76ab08436993d540a5a;p=platform%2Fkernel%2Flinux-starfive.git mptcp: better msk receive window updates Move mptcp_cleanup_rbuf() related checks inside the mentioned helper and extend them to mirror TCP checks more closely. Additionally drop the 'rmem_pending' hack, since commit 879526030c8b ("mptcp: protect the rx path with the msk socket spinlock") we can use instead 'rmem_released'. Fixes: ea4ca586b16f ("mptcp: refine MPTCP-level ack scheduling") Signed-off-by: Paolo Abeni Signed-off-by: Mat Martineau Signed-off-by: David S. Miller --- diff --git a/net/mptcp/options.c b/net/mptcp/options.c index e0d21c0..82a37cc 100644 --- a/net/mptcp/options.c +++ b/net/mptcp/options.c @@ -498,8 +498,8 @@ static bool mptcp_established_options_dss(struct sock *sk, struct sk_buff *skb, { struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk); struct mptcp_sock *msk = mptcp_sk(subflow->conn); + u64 snd_data_fin_enable, ack_seq; unsigned int dss_size = 0; - u64 snd_data_fin_enable; struct mptcp_ext *mpext; unsigned int ack_size; bool ret = false; @@ -531,13 +531,14 @@ static bool mptcp_established_options_dss(struct sock *sk, struct sk_buff *skb, return ret; } + ack_seq = READ_ONCE(msk->ack_seq); if (READ_ONCE(msk->use_64bit_ack)) { ack_size = TCPOLEN_MPTCP_DSS_ACK64; - opts->ext_copy.data_ack = READ_ONCE(msk->ack_seq); + opts->ext_copy.data_ack = ack_seq; opts->ext_copy.ack64 = 1; } else { ack_size = TCPOLEN_MPTCP_DSS_ACK32; - opts->ext_copy.data_ack32 = (uint32_t)READ_ONCE(msk->ack_seq); + opts->ext_copy.data_ack32 = (uint32_t)ack_seq; opts->ext_copy.ack64 = 0; } opts->ext_copy.use_ack = 1; diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index c11fcf6..0dbf5cb 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -457,7 +457,18 @@ static bool mptcp_subflow_cleanup_rbuf(struct sock *ssk) static void mptcp_cleanup_rbuf(struct mptcp_sock *msk) { struct sock *ack_hint = READ_ONCE(msk->ack_hint); + int old_space = READ_ONCE(msk->old_wspace); struct mptcp_subflow_context *subflow; + struct sock *sk = (struct sock *)msk; + bool cleanup; + + /* this is a simple superset of what tcp_cleanup_rbuf() implements + * so that we don't have to acquire the ssk socket lock most of the time + * to do actually nothing + */ + cleanup = __mptcp_space(sk) - old_space >= max(0, old_space); + if (!cleanup) + return; /* if the hinted ssk is still active, try to use it */ if (likely(ack_hint)) { @@ -1865,7 +1876,7 @@ static void __mptcp_splice_receive_queue(struct sock *sk) skb_queue_splice_tail_init(&sk->sk_receive_queue, &msk->receive_queue); } -static bool __mptcp_move_skbs(struct mptcp_sock *msk, unsigned int rcv) +static bool __mptcp_move_skbs(struct mptcp_sock *msk) { struct sock *sk = (struct sock *)msk; unsigned int moved = 0; @@ -1885,13 +1896,10 @@ static bool __mptcp_move_skbs(struct mptcp_sock *msk, unsigned int rcv) slowpath = lock_sock_fast(ssk); mptcp_data_lock(sk); + __mptcp_update_rmem(sk); done = __mptcp_move_skbs_from_subflow(msk, ssk, &moved); mptcp_data_unlock(sk); - if (moved && rcv) { - WRITE_ONCE(msk->rmem_pending, min(rcv, moved)); - tcp_cleanup_rbuf(ssk, 1); - WRITE_ONCE(msk->rmem_pending, 0); - } + tcp_cleanup_rbuf(ssk, moved); unlock_sock_fast(ssk, slowpath); } while (!done); @@ -1904,6 +1912,7 @@ static bool __mptcp_move_skbs(struct mptcp_sock *msk, unsigned int rcv) ret |= __mptcp_ofo_queue(msk); __mptcp_splice_receive_queue(sk); mptcp_data_unlock(sk); + mptcp_cleanup_rbuf(msk); } if (ret) mptcp_check_data_fin((struct sock *)msk); @@ -1933,7 +1942,7 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, target = sock_rcvlowat(sk, flags & MSG_WAITALL, len); while (copied < len) { - int bytes_read, old_space; + int bytes_read; bytes_read = __mptcp_recvmsg_mskq(msk, msg, len - copied); if (unlikely(bytes_read < 0)) { @@ -1944,14 +1953,11 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, copied += bytes_read; - if (skb_queue_empty(&msk->receive_queue) && - __mptcp_move_skbs(msk, len - copied)) - continue; - /* be sure to advertise window change */ - old_space = READ_ONCE(msk->old_wspace); - if ((tcp_space(sk) - old_space) >= old_space) - mptcp_cleanup_rbuf(msk); + mptcp_cleanup_rbuf(msk); + + if (skb_queue_empty(&msk->receive_queue) && __mptcp_move_skbs(msk)) + continue; /* only the master socket status is relevant here. The exit * conditions mirror closely tcp_recvmsg() @@ -1979,7 +1985,7 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, /* race breaker: the shutdown could be after the * previous receive queue check */ - if (__mptcp_move_skbs(msk, len - copied)) + if (__mptcp_move_skbs(msk)) continue; break; } @@ -2012,7 +2018,7 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, /* .. race-breaker: ssk might have gotten new data * after last __mptcp_move_skbs() returned false. */ - if (unlikely(__mptcp_move_skbs(msk, 0))) + if (unlikely(__mptcp_move_skbs(msk))) set_bit(MPTCP_DATA_READY, &msk->flags); } else if (unlikely(!test_bit(MPTCP_DATA_READY, &msk->flags))) { /* data to read but mptcp_wait_data() cleared DATA_READY */ diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h index 6a164ad..8d9f0ff 100644 --- a/net/mptcp/protocol.h +++ b/net/mptcp/protocol.h @@ -234,7 +234,6 @@ struct mptcp_sock { u64 wnd_end; unsigned long timer_ival; u32 token; - int rmem_pending; int rmem_released; unsigned long flags; bool can_ack; @@ -293,7 +292,7 @@ static inline struct mptcp_sock *mptcp_sk(const struct sock *sk) static inline int __mptcp_space(const struct sock *sk) { - return tcp_space(sk) + READ_ONCE(mptcp_sk(sk)->rmem_pending); + return tcp_space(sk) + READ_ONCE(mptcp_sk(sk)->rmem_released); } static inline struct mptcp_data_frag *mptcp_send_head(const struct sock *sk)