Re: 2.6.13-mm2 hard lockup

[Michal Piotrowski <michal.k.k.piotrowski@gmail.com>]

Hi,

On 12/09/05, Bernhard Rosenkraenzer <bero@arklinux.org> wrote:
> 2.6.13-mm2 locks up hard occasionally (happened for me twice, each time about
> 20 minutes after booting).
> It left the following in the syslog:
> 
> Sep 11 23:19:55 dhcppc0 kernel: KERNEL: assertion ((int)tp->lost_out >= 0)
> failed at net/ipv4/tcp_input.c (2148)
> Sep 11 23:19:59 dhcppc0 kernel: KERNEL: assertion ((int)tp->lost_out >= 0)
> failed at net/ipv4/tcp_input.c (2148)
> Sep 11 23:25:23 dhcppc0 kernel: KERNEL: assertion ((int)tp->sacked_out >= 0)
> failed at net/ipv4/tcp_input.c (2147)
> Sep 11 23:26:04 dhcppc0 last message repeated 2 times
> Sep 11 23:27:55 dhcppc0 kernel: KERNEL: assertion ((int)tp->sacked_out >= 0)
> failed at net/ipv4/tcp_input.c (2147)
> Sep 11 23:27:57 dhcppc0 kernel: KERNEL: assertion ((int)tp->lost_out >= 0)
> failed at net/ipv4/tcp_input.c (2148)
> Sep 11 23:28:00 dhcppc0 kernel: KERNEL: assertion ((int)tp->sacked_out >= 0)
> failed at net/ipv4/tcp_input.c (2147)
> 
> (This was compiled without sysrq support, therefore no detailed trace)
> 
> LLaP
> bero
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

Someone from netdev has sent this path. I didn't test it.

diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -485,11 +485,6 @@ int tcp_fragment(struct sock *sk, struct
 	TCP_SKB_CB(buff)->when = TCP_SKB_CB(skb)->when;
 	buff->tstamp = skb->tstamp;
 
-	if (TCP_SKB_CB(skb)->sacked & TCPCB_LOST) {
-		tp->lost_out -= tcp_skb_pcount(skb);
-		tp->left_out -= tcp_skb_pcount(skb);
-	}
-
 	old_factor = tcp_skb_pcount(skb);
 
 	/* Fix up tso_factor for both original and new SKB.  */

Here is patch from Herbert Xu


diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -499,7 +499,7 @@ int tcp_fragment(struct sock *sk, struct
 	/* If this packet has been sent out already, we must
 	 * adjust the various packet counters.
 	 */
-	if (after(tp->snd_nxt, TCP_SKB_CB(buff)->end_seq)) {
+	if (!before(tp->snd_nxt, TCP_SKB_CB(buff)->end_seq)) {
 		int diff = old_factor - tcp_skb_pcount(skb) -
 			tcp_skb_pcount(buff);
 

Regards,
Michal Piotrowski