diff options
author | Eric Dumazet <edumazet@google.com> | 2021-11-15 22:02:46 +0300 |
---|---|---|
committer | David S. Miller <davem@davemloft.net> | 2021-11-16 16:10:35 +0300 |
commit | f35f821935d8df76f9c92e2431a225bdff938169 (patch) | |
tree | 21b212cdb5b58b515c49bb301a5afb4985fcedb2 /include | |
parent | 3df684c1a3d08a4f649689053a3d527b3b5fda9e (diff) | |
download | linux-f35f821935d8df76f9c92e2431a225bdff938169.tar.xz |
tcp: defer skb freeing after socket lock is released
tcp recvmsg() (or rx zerocopy) spends a fair amount of time
freeing skbs after their payload has been consumed.
A typical ~64KB GRO packet has to release ~45 page
references, eventually going to page allocator
for each of them.
Currently, this freeing is performed while socket lock
is held, meaning that there is a high chance that
BH handler has to queue incoming packets to tcp socket backlog.
This can cause additional latencies, because the user
thread has to process the backlog at release_sock() time,
and while doing so, additional frames can be added
by BH handler.
This patch adds logic to defer these frees after socket
lock is released, or directly from BH handler if possible.
Being able to free these skbs from BH handler helps a lot,
because this avoids the usual alloc/free assymetry,
when BH handler and user thread do not run on same cpu or
NUMA node.
One cpu can now be fully utilized for the kernel->user copy,
and another cpu is handling BH processing and skb/page
allocs/frees (assuming RFS is not forcing use of a single CPU)
Tested:
100Gbit NIC
Max throughput for one TCP_STREAM flow, over 10 runs
MTU : 1500
Before: 55 Gbit
After: 66 Gbit
MTU : 4096+(headers)
Before: 82 Gbit
After: 95 Gbit
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'include')
-rw-r--r-- | include/linux/skbuff.h | 2 | ||||
-rw-r--r-- | include/net/sock.h | 3 | ||||
-rw-r--r-- | include/net/tcp.h | 10 |
3 files changed, 15 insertions, 0 deletions
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h index 686a666d073d..b8b806512e16 100644 --- a/include/linux/skbuff.h +++ b/include/linux/skbuff.h @@ -36,6 +36,7 @@ #include <linux/splice.h> #include <linux/in6.h> #include <linux/if_packet.h> +#include <linux/llist.h> #include <net/flow.h> #include <net/page_pool.h> #if IS_ENABLED(CONFIG_NF_CONNTRACK) @@ -743,6 +744,7 @@ struct sk_buff { }; struct rb_node rbnode; /* used in netem, ip4 defrag, and tcp stack */ struct list_head list; + struct llist_node ll_node; }; union { diff --git a/include/net/sock.h b/include/net/sock.h index 2d40fe4c7718..2578d1f455a7 100644 --- a/include/net/sock.h +++ b/include/net/sock.h @@ -63,6 +63,7 @@ #include <linux/indirect_call_wrapper.h> #include <linux/atomic.h> #include <linux/refcount.h> +#include <linux/llist.h> #include <net/dst.h> #include <net/checksum.h> #include <net/tcp_states.h> @@ -408,6 +409,8 @@ struct sock { struct sk_buff *head; struct sk_buff *tail; } sk_backlog; + struct llist_head defer_list; + #define sk_rmem_alloc sk_backlog.rmem_alloc int sk_forward_alloc; diff --git a/include/net/tcp.h b/include/net/tcp.h index 05c81677aaf7..44e442bf23f9 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -1368,6 +1368,16 @@ static inline bool tcp_checksum_complete(struct sk_buff *skb) } bool tcp_add_backlog(struct sock *sk, struct sk_buff *skb); + +void __sk_defer_free_flush(struct sock *sk); + +static inline void sk_defer_free_flush(struct sock *sk) +{ + if (llist_empty(&sk->defer_list)) + return; + __sk_defer_free_flush(sk); +} + int tcp_filter(struct sock *sk, struct sk_buff *skb); void tcp_set_state(struct sock *sk, int state); void tcp_done(struct sock *sk); |