From: "xin.guo" If the new coming segment covers more than one skbs in the ofo queue, and which seq is equal to rcv_nxt, then the sequence range that is duplicated will be sent as DUP SACK, the detail as below, in step6, the {501,2001} range is clearly including too much DUP SACK range, in violation of RFC 2883 rules. 1. client > server: Flags [.], seq 501:1001, ack 1325288529, win 20000, length 500 2. server > client: Flags [.], ack 1, [nop,nop,sack 1 {501:1001}], length 0 3. client > server: Flags [.], seq 1501:2001, ack 1325288529, win 20000, length 500 4. server > client: Flags [.], ack 1, [nop,nop,sack 2 {1501:2001} {501:1001}], length 0 5. client > server: Flags [.], seq 1:2001, ack 1325288529, win 20000, length 2000 6. server > client: Flags [.], ack 2001, [nop,nop,sack 1 {501:2001}], length 0 After this fix, the final ACK is as below: 6. server > client: Flags [.], ack 2001, options [nop,nop,sack 1 {501:1001}], length 0 [edumazet] added a new packetdrill test in the following patch. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: xin.guo Signed-off-by: Eric Dumazet --- net/ipv4/tcp_input.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 19a1542883dfba7211e08dcb44c82b4564d76f04..79e3bfb0108fd4e21de2295e879fbd521daa62f9 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -4845,8 +4845,9 @@ static void tcp_ofo_queue(struct sock *sk) if (before(TCP_SKB_CB(skb)->seq, dsack_high)) { __u32 dsack = dsack_high; + if (before(TCP_SKB_CB(skb)->end_seq, dsack_high)) - dsack_high = TCP_SKB_CB(skb)->end_seq; + dsack = TCP_SKB_CB(skb)->end_seq; tcp_dsack_extend(sk, TCP_SKB_CB(skb)->seq, dsack); } p = rb_next(p); -- 2.50.0.727.gbf7dc18ff4-goog Test DSACK behavior with non contiguous ranges. Without prior fix (tcp: fix tcp_ofo_queue() to avoid including too much DUP SACK range) this would fail with: tcp_dsack_mult.pkt:37: error handling packet: bad value outbound TCP option 5 script packet: 0.100682 . 1:1(0) ack 6001 actual packet: 0.100679 . 1:1(0) ack 6001 win 1097 Signed-off-by: Eric Dumazet Cc: xin.guo --- .../net/packetdrill/tcp_dsack_mult.pkt | 45 +++++++++++++++++++ 1 file changed, 45 insertions(+) create mode 100644 tools/testing/selftests/net/packetdrill/tcp_dsack_mult.pkt diff --git a/tools/testing/selftests/net/packetdrill/tcp_dsack_mult.pkt b/tools/testing/selftests/net/packetdrill/tcp_dsack_mult.pkt new file mode 100644 index 0000000000000000000000000000000000000000..c790d0af635eb471b12a3fde77f62885fed99342 --- /dev/null +++ b/tools/testing/selftests/net/packetdrill/tcp_dsack_mult.pkt @@ -0,0 +1,45 @@ +// SPDX-License-Identifier: GPL-2.0 +// Test various DSACK (RFC 2883) behaviors. + +--mss=1000 + +`./defaults.sh` + + + 0 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3 + +0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0 + +0 bind(3, ..., ...) = 0 + +0 listen(3, 1) = 0 + + +0 < S 0:0(0) win 32792 + +0 > S. 0:0(0) ack 1 + +.1 < . 1:1(0) ack 1 win 1024 + +0 accept(3, ..., ...) = 4 + +// First SACK range. + +0 < P. 1001:2001(1000) ack 1 win 1024 + +0 > . 1:1(0) ack 1 + +// Check SACK coalescing (contiguous sequence). + +0 < P. 2001:3001(1000) ack 1 win 1024 + +0 > . 1:1(0) ack 1 + +// Check we have two SACK ranges for non contiguous sequences. + +0 < P. 4001:5001(1000) ack 1 win 1024 + +0 > . 1:1(0) ack 1 + +// Three ranges. + +0 < P. 7001:8001(1000) ack 1 win 1024 + +0 > . 1:1(0) ack 1 + +// DSACK (1001:3001) + SACK (6001:7001) + +0 < P. 1:6001(6000) ack 1 win 1024 + +0 > . 1:1(0) ack 6001 + +// DSACK (7001:8001) + +0 < P. 6001:8001(2000) ack 1 win 1024 + +0 > . 1:1(0) ack 8001 + +// DSACK for an older segment. + +0 < P. 1:1001(1000) ack 1 win 1024 + +0 > . 1:1(0) ack 8001 -- 2.50.0.727.gbf7dc18ff4-goog