tcp: properly update lost_cnt_hint during shifting
authorYan, Zheng <zheng.z.yan@intel.com>
Sun, 2 Oct 2011 04:21:50 +0000 (04:21 +0000)
committerGreg Kroah-Hartman <gregkh@suse.de>
Fri, 11 Nov 2011 17:36:28 +0000 (09:36 -0800)
[ Upstream commit 1e5289e121372a3494402b1b131b41bfe1cf9b7f ]

lost_skb_hint is used by tcp_mark_head_lost() to mark the first unhandled skb.
lost_cnt_hint is the number of packets or sacked packets before the lost_skb_hint;
When shifting a skb that is before the lost_skb_hint, if tcp_is_fack() is ture,
the skb has already been counted in the lost_cnt_hint; if tcp_is_fack() is false,
tcp_sacktag_one() will increase the lost_cnt_hint. So tcp_shifted_skb() does not
need to adjust the lost_cnt_hint by itself. When shifting a skb that is equal to
lost_skb_hint, the shifted packets will not be counted by tcp_mark_head_lost().
So tcp_shifted_skb() should adjust the lost_cnt_hint even tcp_is_fack(tp) is true.

Signed-off-by: Zheng Yan <zheng.z.yan@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
net/ipv4/tcp_input.c

index b6771f9eb9d40affb7ffbdca7fb70a6f5bc8979e..c68040fe9cd9e04b705607d51cf994aadedf7754 100644 (file)
@@ -1380,9 +1380,7 @@ static int tcp_shifted_skb(struct sock *sk, struct sk_buff *skb,
 
        BUG_ON(!pcount);
 
-       /* Tweak before seqno plays */
-       if (!tcp_is_fack(tp) && tcp_is_sack(tp) && tp->lost_skb_hint &&
-           !before(TCP_SKB_CB(tp->lost_skb_hint)->seq, TCP_SKB_CB(skb)->seq))
+       if (skb == tp->lost_skb_hint)
                tp->lost_cnt_hint += pcount;
 
        TCP_SKB_CB(prev)->end_seq += shifted;