Message ID | 1414690354.9028.9.camel@edumazet-glaptop2.roam.corp.google.com |
---|---|
State | Accepted, archived |
Delegated to: | David Miller |
Headers | show |
From: Eric Dumazet <eric.dumazet@gmail.com> Date: Thu, 30 Oct 2014 10:32:34 -0700 > From: Eric Dumazet <edumazet@google.com> > > Some drivers are unable to perform TX completions in a bound time. > They instead call skb_orphan() > > Problem is skb_fclone_busy() has to detect this case, otherwise > we block TCP retransmits and can freeze unlucky tcp sessions on > mostly idle hosts. > > Signed-off-by: Eric Dumazet <edumazet@google.com> > Fixes: 1f3279ae0c13 ("tcp: avoid retransmits of TCP packets hanging in host queues") Applied, and queued up for -stable, thanks Eric. > This problem is known to hurt users of linux-3.16 kernels used by guests kernels. > David, I can provide backports if you want. Since 3.16 is no longer active, I'll only need to put this into 3.17-stable which I should be able to handle on my own. But thanks for offering, sometimes difficult backports take up a lot of time. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h index 5884f95ff0e9..6c8b6f604e76 100644 --- a/include/linux/skbuff.h +++ b/include/linux/skbuff.h @@ -799,15 +799,19 @@ struct sk_buff_fclones { * @skb: buffer * * Returns true is skb is a fast clone, and its clone is not freed. + * Some drivers call skb_orphan() in their ndo_start_xmit(), + * so we also check that this didnt happen. */ -static inline bool skb_fclone_busy(const struct sk_buff *skb) +static inline bool skb_fclone_busy(const struct sock *sk, + const struct sk_buff *skb) { const struct sk_buff_fclones *fclones; fclones = container_of(skb, struct sk_buff_fclones, skb1); return skb->fclone == SKB_FCLONE_ORIG && - fclones->skb2.fclone == SKB_FCLONE_CLONE; + fclones->skb2.fclone == SKB_FCLONE_CLONE && + fclones->skb2.sk == sk; } static inline struct sk_buff *alloc_skb_fclone(unsigned int size, diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index 3af21296d967..a3d453b94747 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -2126,7 +2126,7 @@ bool tcp_schedule_loss_probe(struct sock *sk) static bool skb_still_in_host_queue(const struct sock *sk, const struct sk_buff *skb) { - if (unlikely(skb_fclone_busy(skb))) { + if (unlikely(skb_fclone_busy(sk, skb))) { NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPSPURIOUS_RTX_HOSTQUEUES); return true; diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c index 4c4e457e7888..88bf289abdc9 100644 --- a/net/xfrm/xfrm_policy.c +++ b/net/xfrm/xfrm_policy.c @@ -1962,7 +1962,7 @@ static int xdst_queue_output(struct sock *sk, struct sk_buff *skb) struct xfrm_policy *pol = xdst->pols[0]; struct xfrm_policy_queue *pq = &pol->polq; - if (unlikely(skb_fclone_busy(skb))) { + if (unlikely(skb_fclone_busy(sk, skb))) { kfree_skb(skb); return 0; }