This patch is needed by the AF-independent approach to provide DCCPv4/6
ECN support. Since this approach handles ECN at in the module, the patch
disables user-side manipulation of ECN bits via TOS/TCLASS.
However, it also fixes a bug occurring in SCTPv6: when passing values that
contain 0x1 in the ECN bits, the internal value of ECT(0) gets turned into
0x3 = CE, i.e. it suggests to the other side that congestion was experienced.
I am working on these issues and would greatly appreciate feedback/suggestions.
(NB - not touched datagram_send_ctl(), since it is used only by udpv6/rawv6).
An alternative for modifying ECN bits in user space is in a subsequent patch.
Signed-off-by: Gerrit Renker <gerrit@xxxxxxxxxxxxxx>
---
net/ipv4/ip_sockglue.c | 7 ++++---
net/ipv6/ipv6_sockglue.c | 2 ++
2 files changed, 6 insertions(+), 3 deletions(-)
--- a/net/ipv4/ip_sockglue.c
+++ b/net/ipv4/ip_sockglue.c
@@ -27,6 +27,7 @@
#include <net/ip.h>
#include <net/icmp.h>
#include <net/tcp_states.h>
+#include <net/inet_ecn.h>
#include <linux/udp.h>
#include <linux/igmp.h>
#include <linux/netfilter.h>
@@ -550,9 +551,9 @@ static int do_ip_setsockopt(struct sock *sk, int level,
inet->cmsg_flags &= ~IP_CMSG_ORIGDSTADDR;
break;
case IP_TOS: /* This sets both TOS and Precedence */
- if (sk->sk_type == SOCK_STREAM) {
- val &= ~3;
- val |= inet->tos & 3;
+ if (sk->sk_type != SOCK_RAW) {
+ val &= ~INET_ECN_MASK;
+ val |= inet->tos & INET_ECN_MASK;
}
if (inet->tos != val) {
inet->tos = val;
--- a/net/ipv6/ipv6_sockglue.c
+++ b/net/ipv6/ipv6_sockglue.c
@@ -318,6 +318,8 @@ static int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
/* RFC 3542, 6.5: default traffic class of 0x0 */
if (val == -1)
val = 0;
+ if (sk->sk_type != SOCK_RAW)
+ val &= ~INET_ECN_MASK;
np->tclass = val;
retv = 0;
break;
--
1.6.0.rc2
--
To unsubscribe from this list: send the line "unsubscribe dccp" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
