ECN is not currently supported on inet6 sockets.
This is a hack to copy over the ECN bits when ECN-capable transport
has been enabled on the inet_sk socket.
This allows to support DCCPv6 ECN support, but needs to be revised:
* the inet_sock already has an u8 `tos' field for the DSCP/ECN bits;
* the same field is present as u8 `tclass' field in ipv6_pinfo;
* the DS field (RFC 2474) and the ECN bits (RFC 3168) are AF-independent,
so TOS/Traffic Class could be made AF-independent;
* sockets that include ipv6_pinfo also include struct inet_sock.
Ideally there should be one field, e.g. `inet->dsfield', but doing
this requires some more careful analysis, since inet_sk(sk)->tos and
inet6_sk(sk)->tclass are referenced all over the source code.
Leaving this for the moment, to be completed later.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
---
include/net/inet_ecn.h | 8 ++++++--
1 files changed, 6 insertions(+), 2 deletions(-)
--- a/include/net/inet_ecn.h
+++ b/include/net/inet_ecn.h
@@ -46,9 +46,13 @@ static inline __u8 INET_ECN_encapsulate(__u8 outer, __u8 inner)
(label) &= ~htonl(INET_ECN_MASK << 20); \
} while (0)
+/*
+ * XXX Hack: copy over the ECN bits from inet_sk to inet6_sk.
+ * This should be replaced by a single `dsfield' in inet
+ */
#define IP6_ECN_flow_xmit(sk, label) do { \
- if (INET_ECN_is_capable(inet6_sk(sk)->tclass)) \
- (label) |= htonl(INET_ECN_ECT_0 << 20); \
+ IP6_ECN_flow_init(label); \
+ (label) |= htonl((inet_sk(sk)->tos & INET_ECN_MASK) << 20); \
} while (0)
static inline int IP_ECN_set_ce(struct iphdr *iph)
--
1.6.0.rc2
--
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
