From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932128AbaELCMq (ORCPT ); Sun, 11 May 2014 22:12:46 -0400 Received: from 1wt.eu ([62.212.114.60]:34467 "EHLO 1wt.eu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752520AbaELBmI (ORCPT ); Sun, 11 May 2014 21:42:08 -0400 Message-Id: <20140512003203.544647249@1wt.eu> User-Agent: quilt/0.48-1 Date: Mon, 12 May 2014 02:33:10 +0200 From: Willy Tarreau To: linux-kernel@vger.kernel.org, stable@vger.kernel.org Cc: Thomas Graf , Eric Dumazet , Hannes Frederic Sowa , Stephen Warren , Fabio Estevam , "David S. Miller" , Willy Tarreau Subject: [ 070/143] ipv6: Dont depend on per socket memory for neighbour discovery In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org 2.6.32-longterm review patch. If anyone has any objections, please let me know. ------------------ messages From: Thomas Graf [ Upstream commit 25a6e6b84fba601eff7c28d30da8ad7cfbef0d43 ] Allocating skbs when sending out neighbour discovery messages currently uses sock_alloc_send_skb() based on a per net namespace socket and thus share a socket wmem buffer space. If a netdevice is temporarily unable to transmit due to carrier loss or for other reasons, the queued up ndisc messages will cosnume all of the wmem space and will thus prevent from any more skbs to be allocated even for netdevices that are able to transmit packets. The number of neighbour discovery messages sent is very limited, use of alloc_skb() bypasses the socket wmem buffer size enforcement while the manual call to skb_set_owner_w() maintains the socket reference needed for the IPv6 output path. This patch has orginally been posted by Eric Dumazet in a modified form. Signed-off-by: Thomas Graf Cc: Eric Dumazet Cc: Hannes Frederic Sowa Cc: Stephen Warren Cc: Fabio Estevam Tested-by: Fabio Estevam Tested-by: Stephen Warren Acked-by: Hannes Frederic Sowa Signed-off-by: David S. Miller Signed-off-by: Willy Tarreau --- net/ipv6/ndisc.c | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) diff --git a/net/ipv6/ndisc.c b/net/ipv6/ndisc.c index f74e4e2..752da21 100644 --- a/net/ipv6/ndisc.c +++ b/net/ipv6/ndisc.c @@ -449,7 +449,6 @@ struct sk_buff *ndisc_build_skb(struct net_device *dev, struct sk_buff *skb; struct icmp6hdr *hdr; int len; - int err; u8 *opt; if (!dev->addr_len) @@ -459,14 +458,12 @@ struct sk_buff *ndisc_build_skb(struct net_device *dev, if (llinfo) len += ndisc_opt_addr_space(dev); - skb = sock_alloc_send_skb(sk, - (MAX_HEADER + sizeof(struct ipv6hdr) + - len + LL_ALLOCATED_SPACE(dev)), - 1, &err); + skb = alloc_skb((MAX_HEADER + sizeof(struct ipv6hdr) + + len + LL_ALLOCATED_SPACE(dev)), GFP_ATOMIC); if (!skb) { ND_PRINTK0(KERN_ERR - "ICMPv6 ND: %s() failed to allocate an skb, err=%d.\n", - __func__, err); + "ICMPv6 ND: %s() failed to allocate an skb.\n", + __func__); return NULL; } @@ -494,6 +491,11 @@ struct sk_buff *ndisc_build_skb(struct net_device *dev, csum_partial(hdr, len, 0)); + /* Manually assign socket ownership as we avoid calling + * sock_alloc_send_pskb() to bypass wmem buffer limits + */ + skb_set_owner_w(skb, sk); + return skb; } -- 1.7.12.2.21.g234cd45.dirty