From: Simon Horman <horms@verge.net.au>
To: lvs-devel@vger.kernel.org, netdev@vger.kernel.org
Cc: "Siim Põder" <siim@p6drad-teel.net>,
"Julian Anastasov" <ja@ssi.bg>,
"Malcolm Turnbull" <malcolm@loadbalancer.org>,
"Julius Volz" <juliusv@google.com>,
"Vince Busam" <vbusam@google.com>,
"Herbert Xu" <herbert@gondor.apana.org.au>
Subject: [rfc 1/3] ipvs: handle PARTIAL_CHECKSUM
Date: Mon, 08 Sep 2008 12:04:21 +1000 [thread overview]
Message-ID: <20080908021535.063462678@vergenet.net> (raw)
In-Reply-To: 20080908020420.313463898@vergenet.net
[-- Attachment #1: ipvs-partial_csum-update.patch --]
[-- Type: text/plain, Size: 6335 bytes --]
Now that LVS can load balance locally generated traffic, packets may come
from the loopback device and thus may have a partial checksum.
The existing code allows for the case where there is no checksum at all for
TCP, however Herbert Xu has confirmed that this is not legal.
Signed-off-by: Simon Horman <horms@verge.net.au>
---
This patch implements *_partial_csum_update() in the style
of the existing *_fast_csum_update() code. A subsequent patch
will reimplement these functions in terms of the more standard
inet_proto_csum_replace*() functions.
net/ipv4/ipvs/ip_vs_proto_tcp.c | 37 +++++++++++++++++++++++++++++++++++--
net/ipv4/ipvs/ip_vs_proto_udp.c | 37 +++++++++++++++++++++++++++++++++++--
2 files changed, 70 insertions(+), 4 deletions(-)
Index: lvs-2.6/net/ipv4/ipvs/ip_vs_proto_tcp.c
===================================================================
--- lvs-2.6.orig/net/ipv4/ipvs/ip_vs_proto_tcp.c 2008-09-08 11:46:28.000000000 +1000
+++ lvs-2.6/net/ipv4/ipvs/ip_vs_proto_tcp.c 2008-09-08 11:56:10.000000000 +1000
@@ -134,12 +134,34 @@ tcp_fast_csum_update(int af, struct tcph
}
+static inline void
+tcp_partial_csum_update(int af, struct tcphdr *tcph,
+ const union nf_inet_addr *oldip,
+ const union nf_inet_addr *newip,
+ __be16 oldlen, __be16 newlen)
+{
+#ifdef CONFIG_IP_VS_IPV6
+ if (af == AF_INET6)
+ tcph->check =
+ csum_fold(ip_vs_check_diff16(oldip->ip6, newip->ip6,
+ ip_vs_check_diff2(oldlen, newlen,
+ ~csum_unfold(tcph->check))));
+ else
+#endif
+ tcph->check =
+ csum_fold(ip_vs_check_diff4(oldip->ip, newip->ip,
+ ip_vs_check_diff2(oldlen, newlen,
+ ~csum_unfold(tcph->check))));
+}
+
+
static int
tcp_snat_handler(struct sk_buff *skb,
struct ip_vs_protocol *pp, struct ip_vs_conn *cp)
{
struct tcphdr *tcph;
unsigned int tcphoff;
+ int oldlen;
#ifdef CONFIG_IP_VS_IPV6
if (cp->af == AF_INET6)
@@ -147,6 +169,7 @@ tcp_snat_handler(struct sk_buff *skb,
else
#endif
tcphoff = ip_hdrlen(skb);
+ oldlen = skb->len - tcphoff;
/* csum_check requires unshared skb */
if (!skb_make_writable(skb, tcphoff+sizeof(*tcph)))
@@ -166,7 +189,11 @@ tcp_snat_handler(struct sk_buff *skb,
tcph->source = cp->vport;
/* Adjust TCP checksums */
- if (!cp->app && (tcph->check != 0)) {
+ if (skb->ip_summed == CHECKSUM_PARTIAL) {
+ tcp_partial_csum_update(cp->af, tcph, &cp->daddr, &cp->vaddr,
+ htonl(oldlen),
+ htonl(skb->len - tcphoff));
+ } else if (!cp->app) {
/* Only port and addr are changed, do fast csum update */
tcp_fast_csum_update(cp->af, tcph, &cp->daddr, &cp->vaddr,
cp->dport, cp->vport);
@@ -204,6 +231,7 @@ tcp_dnat_handler(struct sk_buff *skb,
{
struct tcphdr *tcph;
unsigned int tcphoff;
+ int oldlen;
#ifdef CONFIG_IP_VS_IPV6
if (cp->af == AF_INET6)
@@ -211,6 +239,7 @@ tcp_dnat_handler(struct sk_buff *skb,
else
#endif
tcphoff = ip_hdrlen(skb);
+ oldlen = skb->len - tcphoff;
/* csum_check requires unshared skb */
if (!skb_make_writable(skb, tcphoff+sizeof(*tcph)))
@@ -235,7 +264,11 @@ tcp_dnat_handler(struct sk_buff *skb,
/*
* Adjust TCP checksums
*/
- if (!cp->app && (tcph->check != 0)) {
+ if (skb->ip_summed == CHECKSUM_PARTIAL) {
+ tcp_partial_csum_update(cp->af, tcph, &cp->daddr, &cp->vaddr,
+ htonl(oldlen),
+ htonl(skb->len - tcphoff));
+ } else if (!cp->app) {
/* Only port and addr are changed, do fast csum update */
tcp_fast_csum_update(cp->af, tcph, &cp->vaddr, &cp->daddr,
cp->vport, cp->dport);
Index: lvs-2.6/net/ipv4/ipvs/ip_vs_proto_udp.c
===================================================================
--- lvs-2.6.orig/net/ipv4/ipvs/ip_vs_proto_udp.c 2008-09-08 11:46:28.000000000 +1000
+++ lvs-2.6/net/ipv4/ipvs/ip_vs_proto_udp.c 2008-09-08 11:56:10.000000000 +1000
@@ -141,12 +141,34 @@ udp_fast_csum_update(int af, struct udph
uhdr->check = CSUM_MANGLED_0;
}
+static inline void
+udp_partial_csum_update(int af, struct udphdr *uhdr,
+ const union nf_inet_addr *oldip,
+ const union nf_inet_addr *newip,
+ __be16 oldlen, __be16 newlen)
+{
+#ifdef CONFIG_IP_VS_IPV6
+ if (af == AF_INET6)
+ uhdr->check =
+ csum_fold(ip_vs_check_diff16(oldip->ip6, newip->ip6,
+ ip_vs_check_diff2(oldlen, newlen,
+ ~csum_unfold(uhdr->check))));
+ else
+#endif
+ uhdr->check =
+ csum_fold(ip_vs_check_diff4(oldip->ip, newip->ip,
+ ip_vs_check_diff2(oldlen, newlen,
+ ~csum_unfold(uhdr->check))));
+}
+
+
static int
udp_snat_handler(struct sk_buff *skb,
struct ip_vs_protocol *pp, struct ip_vs_conn *cp)
{
struct udphdr *udph;
unsigned int udphoff;
+ int oldlen;
#ifdef CONFIG_IP_VS_IPV6
if (cp->af == AF_INET6)
@@ -154,6 +176,7 @@ udp_snat_handler(struct sk_buff *skb,
else
#endif
udphoff = ip_hdrlen(skb);
+ oldlen = skb->len - udphoff;
/* csum_check requires unshared skb */
if (!skb_make_writable(skb, udphoff+sizeof(*udph)))
@@ -177,7 +200,11 @@ udp_snat_handler(struct sk_buff *skb,
/*
* Adjust UDP checksums
*/
- if (!cp->app && (udph->check != 0)) {
+ if (skb->ip_summed == CHECKSUM_PARTIAL) {
+ udp_partial_csum_update(cp->af, udph, &cp->daddr, &cp->vaddr,
+ htonl(oldlen),
+ htonl(skb->len - udphoff));
+ } else if (!cp->app && (udph->check != 0)) {
/* Only port and addr are changed, do fast csum update */
udp_fast_csum_update(cp->af, udph, &cp->daddr, &cp->vaddr,
cp->dport, cp->vport);
@@ -216,6 +243,7 @@ udp_dnat_handler(struct sk_buff *skb,
{
struct udphdr *udph;
unsigned int udphoff;
+ int oldlen;
#ifdef CONFIG_IP_VS_IPV6
if (cp->af == AF_INET6)
@@ -223,6 +251,7 @@ udp_dnat_handler(struct sk_buff *skb,
else
#endif
udphoff = ip_hdrlen(skb);
+ oldlen = skb->len - udphoff;
/* csum_check requires unshared skb */
if (!skb_make_writable(skb, udphoff+sizeof(*udph)))
@@ -247,7 +276,11 @@ udp_dnat_handler(struct sk_buff *skb,
/*
* Adjust UDP checksums
*/
- if (!cp->app && (udph->check != 0)) {
+ if (skb->ip_summed == CHECKSUM_PARTIAL) {
+ udp_partial_csum_update(cp->af, udph, &cp->daddr, &cp->vaddr,
+ htonl(oldlen),
+ htonl(skb->len - udphoff));
+ } else if (!cp->app && (udph->check != 0)) {
/* Only port and addr are changed, do fast csum update */
udp_fast_csum_update(cp->af, udph, &cp->vaddr, &cp->daddr,
cp->vport, cp->dport);
--
next prev parent reply other threads:[~2008-09-08 2:04 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-09-08 2:04 [rfc 0/3] IPVS: checksum updates Simon Horman
2008-09-08 2:04 ` Simon Horman [this message]
2008-09-08 7:24 ` [rfc 1/3] ipvs: handle PARTIAL_CHECKSUM Herbert Xu
2008-09-08 9:05 ` Simon Horman
2008-09-08 9:54 ` Herbert Xu
2008-09-08 2:04 ` [rfc 2/3] ipvs: Use inet_proto_csum_replace*() Simon Horman
2008-09-08 2:04 ` [rfc 3/3] ipvs: Consolidate checksuming code Simon Horman
2008-09-08 10:03 ` [rfc 0/3] IPVS: checksum updates Julius Volz
2008-09-08 10:41 ` Simon Horman
2008-09-08 11:42 ` Julius Volz
2008-09-08 11:57 ` Simon Horman
2008-09-08 12:04 ` Simon Horman
2008-09-08 12:14 ` Julius Volz
2008-09-08 12:34 ` Simon Horman
2008-09-08 13:12 ` Julius Volz
2008-09-08 13:20 ` Simon Horman
2008-09-08 13:42 ` Julius Volz
2008-09-08 15:32 ` Julius Volz
2008-09-08 23:22 ` Simon Horman
2008-09-08 23:40 ` Simon Horman
2008-09-09 9:30 ` Julius Volz
2008-09-09 11:31 ` Simon Horman
2008-09-10 17:30 ` Julius Volz
2008-09-10 23:29 ` Simon Horman
2008-09-11 13:07 ` Wensong Zhang
2008-09-11 13:45 ` Simon Horman
2008-09-11 13:55 ` Julius Volz
2008-09-11 14:43 ` Wensong Zhang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20080908021535.063462678@vergenet.net \
--to=horms@verge.net.au \
--cc=herbert@gondor.apana.org.au \
--cc=ja@ssi.bg \
--cc=juliusv@google.com \
--cc=lvs-devel@vger.kernel.org \
--cc=malcolm@loadbalancer.org \
--cc=netdev@vger.kernel.org \
--cc=siim@p6drad-teel.net \
--cc=vbusam@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).