This is a note to let you know that I've just added the patch titled
tcp: remove sk_{tr}x_skb_cache
to the 5.15-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary
The filename of the patch is:
tcp-remove-sk_-tr-x_skb_cache.patch
and it can be found in the queue-5.15 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable@xxxxxxxxxxxxxxx> know about it.
commit a838f15e78adc8752ce13a6661ccef7216e20478
Author: Eric Dumazet <edumazet@xxxxxxxxxx>
Date: Wed Sep 22 19:26:43 2021 +0200
tcp: remove sk_{tr}x_skb_cache
[ Upstream commit d8b81175e412c7abebdb5b37d8a84d5fd19b1aad ]
This reverts the following patches :
- commit 2e05fcae83c4 ("tcp: fix compile error if !CONFIG_SYSCTL")
- commit 4f661542a402 ("tcp: fix zerocopy and notsent_lowat issues")
- commit 472c2e07eef0 ("tcp: add one skb cache for tx")
- commit 8b27dae5a2e8 ("tcp: add one skb cache for rx")
Having a cache of one skb (in each direction) per TCP socket is fragile,
since it can cause a significant increase of memory needs,
and not good enough for high speed flows anyway where more than one skb
is needed.
We want instead to add a generic infrastructure, with more flexible
per-cpu caches, for alien NUMA nodes.
Acked-by: Paolo Abeni <pabeni@xxxxxxxxxx>
Acked-by: Mat Martineau <mathew.j.martineau@xxxxxxxxxxxxxxx>
Signed-off-by: Eric Dumazet <edumazet@xxxxxxxxxx>
Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx>
Stable-dep-of: 448a5ce1120c ("udp6: Fix race condition in udp6_sendmsg & connect")
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
diff --git a/Documentation/networking/ip-sysctl.rst b/Documentation/networking/ip-sysctl.rst
index 7890b395e629b..a4d0082bd3b90 100644
--- a/Documentation/networking/ip-sysctl.rst
+++ b/Documentation/networking/ip-sysctl.rst
@@ -991,14 +991,6 @@ tcp_challenge_ack_limit - INTEGER
in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
Default: 1000
-tcp_rx_skb_cache - BOOLEAN
- Controls a per TCP socket cache of one skb, that might help
- performance of some workloads. This might be dangerous
- on systems with a lot of TCP sockets, since it increases
- memory usage.
-
- Default: 0 (disabled)
-
UDP variables
=============
diff --git a/include/net/sock.h b/include/net/sock.h
index 104d80d850e41..d0683b3c57456 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -261,7 +261,6 @@ struct bpf_local_storage;
* @sk_dst_cache: destination cache
* @sk_dst_pending_confirm: need to confirm neighbour
* @sk_policy: flow policy
- * @sk_rx_skb_cache: cache copy of recently accessed RX skb
* @sk_receive_queue: incoming packets
* @sk_wmem_alloc: transmit queue bytes committed
* @sk_tsq_flags: TCP Small Queues flags
@@ -328,7 +327,6 @@ struct bpf_local_storage;
* @sk_peek_off: current peek_offset value
* @sk_send_head: front of stuff to transmit
* @tcp_rtx_queue: TCP re-transmit queue [union with @sk_send_head]
- * @sk_tx_skb_cache: cache copy of recently accessed TX skb
* @sk_security: used by security modules
* @sk_mark: generic packet mark
* @sk_cgrp_data: cgroup data for this cgroup
@@ -394,7 +392,6 @@ struct sock {
atomic_t sk_drops;
int sk_rcvlowat;
struct sk_buff_head sk_error_queue;
- struct sk_buff *sk_rx_skb_cache;
struct sk_buff_head sk_receive_queue;
/*
* The backlog queue is special, it is always used with
@@ -447,7 +444,6 @@ struct sock {
struct sk_buff *sk_send_head;
struct rb_root tcp_rtx_queue;
};
- struct sk_buff *sk_tx_skb_cache;
struct sk_buff_head sk_write_queue;
__s32 sk_peek_off;
int sk_write_pending;
@@ -1596,18 +1592,10 @@ static inline void sk_mem_uncharge(struct sock *sk, int size)
__sk_mem_reclaim(sk, 1 << 20);
}
-DECLARE_STATIC_KEY_FALSE(tcp_tx_skb_cache_key);
static inline void sk_wmem_free_skb(struct sock *sk, struct sk_buff *skb)
{
sk_wmem_queued_add(sk, -skb->truesize);
sk_mem_uncharge(sk, skb->truesize);
- if (static_branch_unlikely(&tcp_tx_skb_cache_key) &&
- !sk->sk_tx_skb_cache && !skb_cloned(skb)) {
- skb_ext_reset(skb);
- skb_zcopy_clear(skb, true);
- sk->sk_tx_skb_cache = skb;
- return;
- }
__kfree_skb(skb);
}
@@ -2667,7 +2655,6 @@ static inline void skb_setup_tx_timestamp(struct sk_buff *skb, __u16 tsflags)
&skb_shinfo(skb)->tskey);
}
-DECLARE_STATIC_KEY_FALSE(tcp_rx_skb_cache_key);
/**
* sk_eat_skb - Release a skb if it is no longer needed
* @sk: socket to eat this skb from
@@ -2679,12 +2666,6 @@ DECLARE_STATIC_KEY_FALSE(tcp_rx_skb_cache_key);
static inline void sk_eat_skb(struct sock *sk, struct sk_buff *skb)
{
__skb_unlink(skb, &sk->sk_receive_queue);
- if (static_branch_unlikely(&tcp_rx_skb_cache_key) &&
- !sk->sk_rx_skb_cache) {
- sk->sk_rx_skb_cache = skb;
- skb_orphan(skb);
- return;
- }
__kfree_skb(skb);
}
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index e46b11507edc2..4d1c55723ad56 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -133,10 +133,6 @@ void inet_sock_destruct(struct sock *sk)
struct inet_sock *inet = inet_sk(sk);
__skb_queue_purge(&sk->sk_receive_queue);
- if (sk->sk_rx_skb_cache) {
- __kfree_skb(sk->sk_rx_skb_cache);
- sk->sk_rx_skb_cache = NULL;
- }
__skb_queue_purge(&sk->sk_error_queue);
sk_mem_reclaim(sk);
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 1f22e72074fdc..1aa3ecaf456c7 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -540,18 +540,6 @@ static struct ctl_table ipv4_table[] = {
.extra1 = &sysctl_fib_sync_mem_min,
.extra2 = &sysctl_fib_sync_mem_max,
},
- {
- .procname = "tcp_rx_skb_cache",
- .data = &tcp_rx_skb_cache_key.key,
- .mode = 0644,
- .proc_handler = proc_do_static_key,
- },
- {
- .procname = "tcp_tx_skb_cache",
- .data = &tcp_tx_skb_cache_key.key,
- .mode = 0644,
- .proc_handler = proc_do_static_key,
- },
{ }
};
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index fc0fa1f2ca9b1..bd7930b2be278 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -325,11 +325,6 @@ struct tcp_splice_state {
unsigned long tcp_memory_pressure __read_mostly;
EXPORT_SYMBOL_GPL(tcp_memory_pressure);
-DEFINE_STATIC_KEY_FALSE(tcp_rx_skb_cache_key);
-EXPORT_SYMBOL(tcp_rx_skb_cache_key);
-
-DEFINE_STATIC_KEY_FALSE(tcp_tx_skb_cache_key);
-
void tcp_enter_memory_pressure(struct sock *sk)
{
unsigned long val;
@@ -866,18 +861,6 @@ struct sk_buff *sk_stream_alloc_skb(struct sock *sk, int size, gfp_t gfp,
{
struct sk_buff *skb;
- if (likely(!size)) {
- skb = sk->sk_tx_skb_cache;
- if (skb) {
- skb->truesize = SKB_TRUESIZE(skb_end_offset(skb));
- sk->sk_tx_skb_cache = NULL;
- pskb_trim(skb, 0);
- INIT_LIST_HEAD(&skb->tcp_tsorted_anchor);
- skb_shinfo(skb)->tx_flags = 0;
- memset(TCP_SKB_CB(skb), 0, sizeof(struct tcp_skb_cb));
- return skb;
- }
- }
/* The TCP header must be at least 32-bit aligned. */
size = ALIGN(size, 4);
@@ -2952,11 +2935,6 @@ void tcp_write_queue_purge(struct sock *sk)
sk_wmem_free_skb(sk, skb);
}
tcp_rtx_queue_purge(sk);
- skb = sk->sk_tx_skb_cache;
- if (skb) {
- __kfree_skb(skb);
- sk->sk_tx_skb_cache = NULL;
- }
INIT_LIST_HEAD(&tcp_sk(sk)->tsorted_sent_queue);
sk_mem_reclaim(sk);
tcp_clear_all_retrans_hints(tcp_sk(sk));
@@ -2999,10 +2977,6 @@ int tcp_disconnect(struct sock *sk, int flags)
tcp_clear_xmit_timers(sk);
__skb_queue_purge(&sk->sk_receive_queue);
- if (sk->sk_rx_skb_cache) {
- __kfree_skb(sk->sk_rx_skb_cache);
- sk->sk_rx_skb_cache = NULL;
- }
WRITE_ONCE(tp->copied_seq, tp->rcv_nxt);
tp->urg_data = 0;
tcp_write_queue_purge(sk);
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index db05ab4287e30..45d12484552fc 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -1975,7 +1975,6 @@ static void tcp_v4_fill_cb(struct sk_buff *skb, const struct iphdr *iph,
int tcp_v4_rcv(struct sk_buff *skb)
{
struct net *net = dev_net(skb->dev);
- struct sk_buff *skb_to_free;
int sdif = inet_sdif(skb);
int dif = inet_iif(skb);
const struct iphdr *iph;
@@ -2124,17 +2123,12 @@ int tcp_v4_rcv(struct sk_buff *skb)
tcp_segs_in(tcp_sk(sk), skb);
ret = 0;
if (!sock_owned_by_user(sk)) {
- skb_to_free = sk->sk_rx_skb_cache;
- sk->sk_rx_skb_cache = NULL;
ret = tcp_v4_do_rcv(sk, skb);
} else {
if (tcp_add_backlog(sk, skb))
goto discard_and_relse;
- skb_to_free = NULL;
}
bh_unlock_sock(sk);
- if (skb_to_free)
- __kfree_skb(skb_to_free);
put_and_return:
if (refcounted)
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index b6f5a4474d8bc..7b81ab2b57011 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -1626,7 +1626,6 @@ static void tcp_v6_fill_cb(struct sk_buff *skb, const struct ipv6hdr *hdr,
INDIRECT_CALLABLE_SCOPE int tcp_v6_rcv(struct sk_buff *skb)
{
- struct sk_buff *skb_to_free;
int sdif = inet6_sdif(skb);
int dif = inet6_iif(skb);
const struct tcphdr *th;
@@ -1762,17 +1761,12 @@ INDIRECT_CALLABLE_SCOPE int tcp_v6_rcv(struct sk_buff *skb)
tcp_segs_in(tcp_sk(sk), skb);
ret = 0;
if (!sock_owned_by_user(sk)) {
- skb_to_free = sk->sk_rx_skb_cache;
- sk->sk_rx_skb_cache = NULL;
ret = tcp_v6_do_rcv(sk, skb);
} else {
if (tcp_add_backlog(sk, skb))
goto discard_and_relse;
- skb_to_free = NULL;
}
bh_unlock_sock(sk);
- if (skb_to_free)
- __kfree_skb(skb_to_free);
put_and_return:
if (refcounted)
sock_put(sk);
