summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorPaolo Abeni <pabeni@redhat.com>2024-02-20 12:00:01 +0100
committerJakub Kicinski <kuba@kernel.org>2024-02-22 02:05:01 +0100
commitf796feabb9f5b1e5c48780a7a0023ab4b82336dd (patch)
treef50132512a9ed9f9a500aa2b79140f5cde52e8be
parentMerge branch 'net-phy-marvell-88q2xxx-add-driver-for-the-marvell-88q2220-phy' (diff)
downloadlinux-f796feabb9f5b1e5c48780a7a0023ab4b82336dd.tar.xz
linux-f796feabb9f5b1e5c48780a7a0023ab4b82336dd.zip
udp: add local "peek offset enabled" flag
We want to re-organize the struct sock layout. The sk_peek_off field location is problematic, as most protocols want it in the RX read area, while UDP wants it on a cacheline different from sk_receive_queue. Create a local (inside udp_sock) copy of the 'peek offset is enabled' flag and place it inside the same cacheline of reader_queue. Check such flag before reading sk_peek_off. This will save potential false sharing and cache misses in the fast-path. Tested under UDP flood with small packets. The struct sock layout update causes a 4% performance drop, and this patch restores completely the original tput. Signed-off-by: Paolo Abeni <pabeni@redhat.com> Reviewed-by: Eric Dumazet <edumazet@google.com> Reviewed-by: Willem de Bruijn <willemb@google.com> Link: https://lore.kernel.org/r/67ab679c15fbf49fa05b3ffe05d91c47ab84f147.1708426665.git.pabeni@redhat.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
-rw-r--r--include/linux/udp.h10
-rw-r--r--net/ipv4/af_inet.c2
-rw-r--r--net/ipv4/udp.c2
-rw-r--r--net/ipv6/af_inet6.c2
4 files changed, 13 insertions, 3 deletions
diff --git a/include/linux/udp.h b/include/linux/udp.h
index d04188714dca..3748e82b627b 100644
--- a/include/linux/udp.h
+++ b/include/linux/udp.h
@@ -92,6 +92,9 @@ struct udp_sock {
/* This fields follows rcvbuf value, and is touched by udp_recvmsg */
int forward_threshold;
+
+ /* Cache friendly copy of sk->sk_peek_off >= 0 */
+ bool peeking_with_offset;
};
#define udp_test_bit(nr, sk) \
@@ -109,6 +112,13 @@ struct udp_sock {
#define udp_sk(ptr) container_of_const(ptr, struct udp_sock, inet.sk)
+static inline int udp_set_peek_off(struct sock *sk, int val)
+{
+ sk_set_peek_off(sk, val);
+ WRITE_ONCE(udp_sk(sk)->peeking_with_offset, val >= 0);
+ return 0;
+}
+
static inline void udp_set_no_check6_tx(struct sock *sk, bool val)
{
udp_assign_bit(NO_CHECK6_TX, sk, val);
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index ad278009e469..5daebdcbca32 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -1103,7 +1103,7 @@ const struct proto_ops inet_dgram_ops = {
.recvmsg = inet_recvmsg,
.mmap = sock_no_mmap,
.splice_eof = inet_splice_eof,
- .set_peek_off = sk_set_peek_off,
+ .set_peek_off = udp_set_peek_off,
#ifdef CONFIG_COMPAT
.compat_ioctl = inet_compat_ioctl,
#endif
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index f631b0a21af4..38cce7cc51f6 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -1589,7 +1589,7 @@ int udp_init_sock(struct sock *sk)
void skb_consume_udp(struct sock *sk, struct sk_buff *skb, int len)
{
- if (unlikely(READ_ONCE(sk->sk_peek_off) >= 0)) {
+ if (unlikely(READ_ONCE(udp_sk(sk)->peeking_with_offset))) {
bool slow = lock_sock_fast(sk);
sk_peek_offset_bwd(sk, len);
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index 959bfd9f6344..b90d46533cdc 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -736,7 +736,7 @@ const struct proto_ops inet6_dgram_ops = {
.recvmsg = inet6_recvmsg, /* retpoline's sake */
.read_skb = udp_read_skb,
.mmap = sock_no_mmap,
- .set_peek_off = sk_set_peek_off,
+ .set_peek_off = udp_set_peek_off,
#ifdef CONFIG_COMPAT
.compat_ioctl = inet6_compat_ioctl,
#endif