tcp: change tcp_skb_pcount() location

Our goal is to access no more than one cache line access per skb in
a write or receive queue when doing the various walks.

After recent TCP_SKB_CB() reorganizations, it is almost done.

Last part is tcp_skb_pcount() which currently uses
skb_shinfo(skb)->gso_segs, which is a terrible choice, because it needs
3 cache lines in current kernel (skb->head, skb->end, and
shinfo->gso_segs are all in 3 different cache lines, far from skb->cb)

This very simple patch reuses space currently taken by tcp_tw_isn
only in input path, as tcp_skb_pcount is only needed for skb stored in
write queue.

This considerably speeds up tcp_ack(), granted we avoid shinfo->tx_flags
to get SKBTX_ACK_TSTAMP, which seems possible.

This also speeds up all sack processing in general.

This speeds up tcp_sendmsg() because it no longer has to access/dirty
shinfo.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
Eric Dumazet
2014-09-24 04:11:22 -07:00
committed by David S. Miller
parent dc83d4d8f6
commit cd7d8498c9
4 changed files with 33 additions and 11 deletions

View File

@ -698,7 +698,16 @@ static inline u32 tcp_skb_timestamp(const struct sk_buff *skb)
struct tcp_skb_cb {
__u32 seq; /* Starting sequence number */
__u32 end_seq; /* SEQ + FIN + SYN + datalen */
__u32 tcp_tw_isn; /* isn chosen by tcp_timewait_state_process() */
union {
/* Note : tcp_tw_isn is used in input path only
* (isn chosen by tcp_timewait_state_process())
*
* tcp_gso_segs is used in write queue only,
* cf tcp_skb_pcount()
*/
__u32 tcp_tw_isn;
__u32 tcp_gso_segs;
};
__u8 tcp_flags; /* TCP header flags. (tcp[13]) */
__u8 sacked; /* State flags for SACK/FACK. */
@ -746,7 +755,17 @@ TCP_ECN_create_request(struct request_sock *req, const struct sk_buff *skb,
*/
static inline int tcp_skb_pcount(const struct sk_buff *skb)
{
return skb_shinfo(skb)->gso_segs;
return TCP_SKB_CB(skb)->tcp_gso_segs;
}
static inline void tcp_skb_pcount_set(struct sk_buff *skb, int segs)
{
TCP_SKB_CB(skb)->tcp_gso_segs = segs;
}
static inline void tcp_skb_pcount_add(struct sk_buff *skb, int segs)
{
TCP_SKB_CB(skb)->tcp_gso_segs += segs;
}
/* This is valid iff tcp_skb_pcount() > 1. */