2.4.1 tcp ack bug ?

2.4.1 tcp ack bug ?

[Giuliano Pochini]

IMHO this tcpdump trace is quite strange:


20:56:25.232532 ppp0 < mc105-v-2.royaume.com.6699 > ppp12.shiny.it.33148: .
96265:97725(1460) ack 77 win 8684 (DF)
20:56:25.242532 ppp0 > ppp12.shiny.it.33148 > mc105-v-2.royaume.com.6699: .
77:77(0) ack 88073 win 62780 <nop,nop, sack 1 {90121:97725} > (DF)
20:56:25.352532 ppp0 < mc105-v-2.royaume.com.6699 > ppp12.shiny.it.33148: P
97725:98313(588) ack 77 win 8684 (DF)
20:56:25.362532 ppp0 > ppp12.shiny.it.33148 > mc105-v-2.royaume.com.6699: .
77:77(0) ack 88073 win 62780 <nop,nop, sack 1 {90121:98313} > (DF)
20:56:26.172532 ppp0 < mc105-v-2.royaume.com.6699 > ppp12.shiny.it.33148: .
88073:89533(1460) ack 77 win 8684 (DF)
20:56:26.302532 ppp0 < mc105-v-2.royaume.com.6699 > ppp12.shiny.it.33148: P
89533:90121(588) ack 77 win 8684 (DF)

Ok, it has just received the missing part, so why it does not ack 98313 ?

20:56:26.312532 ppp0 > ppp12.shiny.it.33148 > mc105-v-2.royaume.com.6699: .
77:77(0) ack 88073 win 62780 <nop,nop, sack 1 {89533:98313} > (DF)
20:56:32.272532 ppp0 < mc105-v-2.royaume.com.6699 > ppp12.shiny.it.33148: .
88073:89533(1460) ack 77 win 8684 (DF)
20:56:32.282532 ppp0 > ppp12.shiny.it.33148 > mc105-v-2.royaume.com.6699: .
77:77(0) ack 98313 win 52560 (DF)


I have other traces where the sack interval looks wrong too, let me know if
you're interested.


[kernel 2.4.1, gcc 2.95.3, tcpdump 3.4, ppp 2.4.0, PowerPC 750]

Bye.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: 2.4.1 tcp ack bug ?

[kuznet]

Hello!

> 20:56:26.172532 ppp0 < mc105-v-2.royaume.com.6699 > ppp12.shiny.it.33148: .
> 88073:89533(1460) ack 77 win 8684 (DF)
....
> Ok, it has just received the missing part, so why it does not ack 98313 ?

Apparently, because this segment has not been received.

Look at checksum errros statistics, for example.

Alexey
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/