Commit 71b7dec2 authored by Paolo Abeni's avatar Paolo Abeni Committed by David S. Miller

mptcp: less aggressive retransmission strategy

The current mptcp re-inject strategy is very aggressive,
we have mptcp-level retransmissions even on single subflow
connection, if the link in-use is lossy.

Let's be a little more conservative: we do retransmit
only if at least a subflow has write and rtx queue empty.

Additionally use the backup subflows only if the active
subflows are stale - no progresses in at least an rtx period
and ignore stale subflows for rtx timeout update

Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/207Signed-off-by: default avatarPaolo Abeni <pabeni@redhat.com>
Signed-off-by: default avatarMat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parent 33d41c9c
...@@ -308,6 +308,23 @@ int mptcp_pm_get_local_id(struct mptcp_sock *msk, struct sock_common *skc) ...@@ -308,6 +308,23 @@ int mptcp_pm_get_local_id(struct mptcp_sock *msk, struct sock_common *skc)
return mptcp_pm_nl_get_local_id(msk, skc); return mptcp_pm_nl_get_local_id(msk, skc);
} }
void mptcp_pm_subflow_chk_stale(const struct mptcp_sock *msk, struct sock *ssk)
{
struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(ssk);
u32 rcv_tstamp = READ_ONCE(tcp_sk(ssk)->rcv_tstamp);
/* keep track of rtx periods with no progress */
if (!subflow->stale_count) {
subflow->stale_rcv_tstamp = rcv_tstamp;
subflow->stale_count++;
} else if (subflow->stale_rcv_tstamp == rcv_tstamp) {
if (subflow->stale_count < U8_MAX)
subflow->stale_count++;
} else {
subflow->stale_count = 0;
}
}
void mptcp_pm_data_init(struct mptcp_sock *msk) void mptcp_pm_data_init(struct mptcp_sock *msk)
{ {
msk->pm.add_addr_signaled = 0; msk->pm.add_addr_signaled = 0;
......
...@@ -420,7 +420,8 @@ static long mptcp_timeout_from_subflow(const struct mptcp_subflow_context *subfl ...@@ -420,7 +420,8 @@ static long mptcp_timeout_from_subflow(const struct mptcp_subflow_context *subfl
{ {
const struct sock *ssk = mptcp_subflow_tcp_sock(subflow); const struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
return inet_csk(ssk)->icsk_pending ? inet_csk(ssk)->icsk_timeout - jiffies : 0; return inet_csk(ssk)->icsk_pending && !subflow->stale_count ?
inet_csk(ssk)->icsk_timeout - jiffies : 0;
} }
static void mptcp_set_timeout(struct sock *sk) static void mptcp_set_timeout(struct sock *sk)
...@@ -2100,8 +2101,9 @@ static void mptcp_timeout_timer(struct timer_list *t) ...@@ -2100,8 +2101,9 @@ static void mptcp_timeout_timer(struct timer_list *t)
*/ */
static struct sock *mptcp_subflow_get_retrans(const struct mptcp_sock *msk) static struct sock *mptcp_subflow_get_retrans(const struct mptcp_sock *msk)
{ {
struct sock *backup = NULL, *pick = NULL;
struct mptcp_subflow_context *subflow; struct mptcp_subflow_context *subflow;
struct sock *backup = NULL; int min_stale_count = INT_MAX;
sock_owned_by_me((const struct sock *)msk); sock_owned_by_me((const struct sock *)msk);
...@@ -2114,11 +2116,11 @@ static struct sock *mptcp_subflow_get_retrans(const struct mptcp_sock *msk) ...@@ -2114,11 +2116,11 @@ static struct sock *mptcp_subflow_get_retrans(const struct mptcp_sock *msk)
if (!mptcp_subflow_active(subflow)) if (!mptcp_subflow_active(subflow))
continue; continue;
/* still data outstanding at TCP level? Don't retransmit. */ /* still data outstanding at TCP level? skip this */
if (!tcp_write_queue_empty(ssk)) { if (!tcp_rtx_and_write_queues_empty(ssk)) {
if (inet_csk(ssk)->icsk_ca_state >= TCP_CA_Loss) mptcp_pm_subflow_chk_stale(msk, ssk);
min_stale_count = min_t(int, min_stale_count, subflow->stale_count);
continue; continue;
return NULL;
} }
if (subflow->backup) { if (subflow->backup) {
...@@ -2127,10 +2129,15 @@ static struct sock *mptcp_subflow_get_retrans(const struct mptcp_sock *msk) ...@@ -2127,10 +2129,15 @@ static struct sock *mptcp_subflow_get_retrans(const struct mptcp_sock *msk)
continue; continue;
} }
return ssk; if (!pick)
pick = ssk;
} }
return backup; if (pick)
return pick;
/* use backup only if there are no progresses anywhere */
return min_stale_count > 1 ? backup : NULL;
} }
static void mptcp_dispose_initial_subflow(struct mptcp_sock *msk) static void mptcp_dispose_initial_subflow(struct mptcp_sock *msk)
......
...@@ -439,11 +439,13 @@ struct mptcp_subflow_context { ...@@ -439,11 +439,13 @@ struct mptcp_subflow_context {
u8 reset_seen:1; u8 reset_seen:1;
u8 reset_transient:1; u8 reset_transient:1;
u8 reset_reason:4; u8 reset_reason:4;
u8 stale_count;
long delegated_status; long delegated_status;
struct list_head delegated_node; /* link into delegated_action, protected by local BH */ struct list_head delegated_node; /* link into delegated_action, protected by local BH */
u32 setsockopt_seq; u32 setsockopt_seq;
u32 stale_rcv_tstamp;
struct sock *tcp_sock; /* tcp sk backpointer */ struct sock *tcp_sock; /* tcp sk backpointer */
struct sock *conn; /* parent mptcp_sock */ struct sock *conn; /* parent mptcp_sock */
...@@ -690,6 +692,7 @@ void mptcp_crypto_hmac_sha(u64 key1, u64 key2, u8 *msg, int len, void *hmac); ...@@ -690,6 +692,7 @@ void mptcp_crypto_hmac_sha(u64 key1, u64 key2, u8 *msg, int len, void *hmac);
void __init mptcp_pm_init(void); void __init mptcp_pm_init(void);
void mptcp_pm_data_init(struct mptcp_sock *msk); void mptcp_pm_data_init(struct mptcp_sock *msk);
void mptcp_pm_subflow_chk_stale(const struct mptcp_sock *msk, struct sock *ssk);
void mptcp_pm_new_connection(struct mptcp_sock *msk, const struct sock *ssk, int server_side); void mptcp_pm_new_connection(struct mptcp_sock *msk, const struct sock *ssk, int server_side);
void mptcp_pm_fully_established(struct mptcp_sock *msk, const struct sock *ssk, gfp_t gfp); void mptcp_pm_fully_established(struct mptcp_sock *msk, const struct sock *ssk, gfp_t gfp);
bool mptcp_pm_allow_new_subflow(struct mptcp_sock *msk); bool mptcp_pm_allow_new_subflow(struct mptcp_sock *msk);
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment