Use sk_mark for routing lookup in more places
Subsystems (6)
PATCH 1/1 Use sk_mark for routing lookup in more places
This patch against v2.6.31 adds support for route lookup using sk_mark in some more places. The benefits from this patch are the following. First, SO_MARK option now has effect on UDP sockets too.
Second, ip_queue_xmit() and inet_sk_rebuild_header() could fail to do routing lookup correctly if TCP sockets with SO_MARK were used.
--- net/ipv4/af_inet.c | 1 + net/ipv4/ip_output.c | 1 + net/ipv4/udp.c | 1 + 3 files changed, 3 insertions(+)
net/ipv4/af_inet.c +1 -0
@@ -1103,6 +1103,7 @@ int inet_sk_rebuild_header(struct sock *sk) {struct flowi fl = { .oif = sk->sk_bound_dev_if, + .mark = sk->sk_mark, .nl_u = { .ip4_u = { .daddr = daddr,
net/ipv4/ip_output.c +1 -0
@@ -335,6 +335,7 @@ int ip_queue_xmit(struct sk_buff *skb, int ipfragok) {struct flowi fl = { .oif = sk->sk_bound_dev_if, + .mark = sk->sk_mark, .nl_u = { .ip4_u = { .daddr = daddr, .saddr = inet->saddr,
net/ipv4/udp.c +0 -0
@@ -688,6 +688,7 @@ int udp_sendmsg(struct kiocb *iocb, struct sock *sk,
struct msghdr *msg,
if (rt == NULL) {
struct flowi fl = { .oif = ipc.oif,
+ .mark = sk->sk_mark,
.nl_u = { .ip4_u =
{ .daddr = faddr,
.saddr = saddr,
Quoted 7 lines · Atis Elsts
This patch against v2.6.31 adds support for route lookup using
sk_markin some more places. The benefits from this patch are the following. First,SO_MARKoption now has effect on UDP sockets too.
Second,ip_queue_xmit()andinet_sk_rebuild_header()could fail to do routing lookup correctly if TCP sockets withSO_MARKwere used.Signed-off-by Atis Elsts <[email protected]>
Good catch, thanks !
I used SO_MARK on connected UDP sockets so did not notice the lack of functionality. (ip_route_connect() does use sk->sk_mark)
Quoted 44 lines
--- net/ipv4/af_inet.c | 1 + net/ipv4/ip_output.c | 1 + net/ipv4/udp.c | 1 + 3 files changed, 3 insertions(+)net/ipv4/af_inet.c +1
-0@@ -1103,6 +1103,7 @@ int inet_sk_rebuild_header(struct sock *sk) {struct flowi fl = { .oif = sk->sk_bound_dev_if, + .mark = sk->sk_mark, .nl_u = { .ip4_u = { .daddr = daddr,net/ipv4/ip_output.c +1
-0@@ -335,6 +335,7 @@ int ip_queue_xmit(struct sk_buff *skb, int ipfragok) {struct flowi fl = { .oif = sk->sk_bound_dev_if, + .mark = sk->sk_mark, .nl_u = { .ip4_u = { .daddr = daddr, .saddr = inet->saddr,net/ipv4/udp.c +1
-0@@ -688,6 +688,7 @@ int udp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg, if (rt == NULL) { struct flowi fl = { .oif = ipc.oif, + .mark = sk->sk_mark, .nl_u = { .ip4_u = { .daddr = faddr,.saddr = saddr,
--
From: Eric Dumazet <[email protected]>
Date: Thu, 01 Oct 2009 18:30:06 +0200
Quoted 7 lines
Quoted 6 lines · Atis Elsts
This patch against v2.6.31 adds support for route lookup using
sk_markin some more places. The benefits from this patch are the following. First,SO_MARKoption now has effect on UDP sockets too.
Second,ip_queue_xmit()andinet_sk_rebuild_header()could fail to do routing lookup correctly if TCP sockets withSO_MARKwere used.
Quoted 8 lines
Quoted 1 line
Signed-off-by Atis Elsts <[email protected]>Good catch, thanks !
I used
SO_MARKon connected UDP sockets so did not notice the lack of functionality. (ip_route_connect()does usesk->sk_mark)Acked-by Eric Dumazet <[email protected]>
Applied, thanks.
But Atis, your patch was whitespace corrupted by your email client so I had to apply your patch by hand. Please correct this problem before submitting future patches, thank you.
Here is a followup on this area, thanks.
[RFC] af_packet: fill skb->mark at xmit
skb->mark may be used by classifiers, so fill it in case user set a SO_MARK option on socket.
--- net/packet/af_packet.c | 3 +++ 1 file changed, 3 insertions(+)
net/packet/af_packet.c +3 -0
@@ -490,6 +490,7 @@ static int packet_sendmsg_spkt(struct kiocb *iocb, struct socket *sock, skb->protocol = proto; skb->dev = dev; skb->priority = sk->sk_priority; + skb->mark = sk->sk_mark;if (err) goto out_free;
@@ -856,6 +857,7 @@ static int tpacket_fill_skb(struct packet_sock *po, struct sk_buff *skb, skb->protocol = proto; skb->dev = dev; skb->priority = po->sk.sk_priority; + skb->mark = po->sk.sk_mark;skb_shinfo (skb)->destructor_arg = ph.raw;switch (po->tp_version) {
@@ -1125,6 +1127,7 @@ static int packet_snd(struct socket *sock, skb->protocol = proto; skb->dev = dev; skb->priority = sk->sk_priority; + skb->mark = sk->sk_mark;/* * Now send it
Quoted 7 lines · Eric Dumazet
Here is a followup on this area, thanks.
[RFC]
af_packet: fillskb->markat xmit
skb->markmay be used by classifiers, so fill it in case user set aSO_MARKoption on socket.
Maybe a more generic way to handle this for various protocols would be to fill skb->mark in sock_alloc_send_pskb()
Cool!
As I've already pointed out in a post 2 or so weeks ago, we need the exact same treatment in a ton of places throughout the code (tcp, ipv6, decnet, etc...).
Maybe it would make more sense to create some constructor-like functions for the flowi struct?
Quoted 18 lines · Eric Dumazet
Quoted 7 lines · Eric Dumazet
Here is a followup on this area, thanks.
[RFC]
af_packet: fillskb->markat xmit
skb->markmay be used by classifiers, so fill it in case user set aSO_MARKoption on socket.Maybe a more generic way to handle this for various protocols would be to fill
skb->markinsock_alloc_send_pskb()--
To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
David Miller
maintainerFrom: Maciej Żenczykowski <[email protected]>
Date: Fri, 2 Oct 2009 10:25:13 -0700
Quoted 2 lines
Maybe it would make more sense to create some constructor-like functions for the flowi struct?
Maybe just an initializer like "FLOWI_SOCK(sk)" or similar. So you can say:
struct flowi fl = FLOWI_SOCK(sk);
But the thing is we usually want to initialize all of the details in one go, so we'd need a very messy macro for this that would take many arguments.
It's important to use an initializer rather than assignments in some inline function so that GCC can better coalesce many small members into since large stores to the stack. It doesn't do this as well with real assignment statements.
PATCH 1/1 Use sk_mark for IPv6 routing lookups
In reply to Maciej ŻenczykowskiQuoted 5 lines · Maciej Żenczykowski
Cool!
As I've already pointed out in a post 2 or so weeks ago, we need the exact same treatment in a ton of places throughout the code (tcp, ipv6, decnet, etc...).
Here's a try at the IPv6 part...
Add support for IPv6 route lookups using sk_mark.
---
net/ipv6/datagram.c +1 -0
@@ -147,6 +147,7 @@ ipv4_connected:ipv6_addr_copy (&fl.fl6_dst, &np->daddr);ipv6_addr_copy (&fl.fl6_src, &np->saddr); fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_dport = inet->dport; fl.fl_ip_sport = inet->sport;
net/ipv6/inet6_connection_sock.c +1 -0
@@ -194,6 +194,7 @@ int inet6_csk_xmit(struct sk_buff *skb, int ipfragok) fl.fl6_flowlabel = np->flow_label;IP6_ECN_flow_xmit (sk, fl.fl6_flowlabel); fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_sport = inet->sport; fl.fl_ip_dport = inet->dport;security_sk_classify_flow (sk, &fl);
net/ipv6/ip6mr.c +1 -0
@@ -1523,6 +1523,7 @@ static int ip6mr_forward2(struct sk_buff *skb, struct mfc6_cache *c, int vifi) fl = (struct flowi ) { .oif = vif->link, + .mark = skb->mark, .nl_u = { .ip6_u = { .daddr = ipv6h->daddr, } }
net/ipv6/ipv6_sockglue.c +1 -0
@@ -424,6 +424,7 @@ sticky_done: fl.fl6_flowlabel =0 ; fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark;if (optlen ==0 ) goto update;
net/ipv6/sit.c +1 -0
@@ -629,6 +629,7 @@ static netdev_tx_t ipip6_tunnel_xmit(struct sk_buff *skb, .saddr = tiph->saddr, .tos = RT_TOS(tos) } }, .oif = tunnel->parms.link, + .mark = skb->mark, .proto = IPPROTO_IPV6 };if (ip_route_output_key (dev_net (dev), &rt, &fl)) { stats->tx_carrier_errors++;
net/ipv6/tcp_ipv6.c +4 -0
@@ -243,6 +243,7 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,ipv6_addr_copy (&fl.fl6_src, (saddr ? saddr : &np->saddr)); fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_dport = usin->sin6_port; fl.fl_ip_sport = inet->sport;
@@ -383,6 +384,7 @@ static void tcp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,ipv6_addr_copy (&fl.fl6_dst, &np->daddr);ipv6_addr_copy (&fl.fl6_src, &np->saddr); fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_dport = inet->dport; fl.fl_ip_sport = inet->sport;security_skb_classify_flow (skb, &fl);
@@ -477,6 +479,7 @@ static int tcp_v6_send_synack(struct sock *sk, struct request_sock *req)ipv6_addr_copy (&fl.fl6_src, &treq->loc_addr); fl.fl6_flowlabel =0 ; fl.oif = treq->iif; + fl.mark = sk->sk_mark; fl.fl_ip_dport =inet_rsk (req)->rmt_port; fl.fl_ip_sport =inet_rsk (req)->loc_port;security_req_classify_flow (req, &fl);
@@ -1345,6 +1348,7 @@ static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb, }ipv6_addr_copy (&fl.fl6_src, &treq->loc_addr); fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_dport =inet_rsk (req)->rmt_port; fl.fl_ip_sport =inet_rsk (req)->loc_port;security_req_classify_flow (req, &fl);
net/ipv6/udp.c +2 -0
@@ -879,6 +879,8 @@ do_udp_sendmsg:if (!fl.oif) fl.oif = np->sticky_pktinfo.ipi6_ifindex; + fl.mark = sk->sk_mark; +if (msg->msg_controllen) { opt = &opt_space;memset (opt,0 ,sizeof (struct ipv6_txoptions ));
Re: [PATCH] Use sk_mark for IPv6 routing lookups
Quoted 12 lines · On Friday 02 October 2009 21:19:05 Brian Haley wrote:
Quoted 5 lines · Maciej Żenczykowski
Cool!
As I've already pointed out in a post 2 or so weeks ago, we need the exact same treatment in a ton of places throughout the code (tcp, ipv6, decnet, etc...).
Here's a try at the IPv6 part...
Add support for IPv6 route lookups using
sk_mark.
Not sure if there is need to fill the mark from skb in tunnel xmit functions. In any case, it's not done for GRE or IPIP tunnels at the moment.
Also, in this patch you are doing that for SIT (v6-in-v4) tunnels only, and not doing it for v4-in-v6 or v6-in-v6 tunnels. Any reason for that?
Atis
Re: [PATCH] Use sk_mark for IPv6 routing lookups
Quoted 1 line · Atis Elsts
Not sure if there is need to fill the mark from skb in tunnel xmit functions. In any case, it's not done for GRE or IPIP tunnels at the moment.
Ok, I'll just drop that part, I'm not sure what should be done in this case.
Quoted 1 line
Also, in this patch you are doing that for SIT (v6-in-v4) tunnels only, and not doing it for v4-in-v6 or v6-in-v6 tunnels. Any reason for that?
I just sent that patch out too quickly, here's a better one with the updates.
Add support for IPv6 route lookups using sk_mark.
---
net/ipv6/af_inet6.c +1 -0
@@ -654,6 +654,7 @@ int inet6_sk_rebuild_header(struct sock *sk)ipv6_addr_copy (&fl.fl6_src, &np->saddr); fl.fl6_flowlabel = np->flow_label; fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_dport = inet->dport; fl.fl_ip_sport = inet->sport;security_sk_classify_flow (sk, &fl);
net/ipv6/datagram.c +1 -0
@@ -147,6 +147,7 @@ ipv4_connected:ipv6_addr_copy (&fl.fl6_dst, &np->daddr);ipv6_addr_copy (&fl.fl6_src, &np->saddr); fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_dport = inet->dport; fl.fl_ip_sport = inet->sport;
net/ipv6/inet6_connection_sock.c +1 -0
@@ -194,6 +194,7 @@ int inet6_csk_xmit(struct sk_buff *skb, int ipfragok) fl.fl6_flowlabel = np->flow_label;IP6_ECN_flow_xmit (sk, fl.fl6_flowlabel); fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_sport = inet->sport; fl.fl_ip_dport = inet->dport;security_sk_classify_flow (sk, &fl);
net/ipv6/ipv6_sockglue.c +1 -0
@@ -424,6 +424,7 @@ sticky_done: fl.fl6_flowlabel =0 ; fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark;if (optlen ==0 ) goto update;
net/ipv6/syncookies.c +1 -0
@@ -252,6 +252,7 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb) }ipv6_addr_copy (&fl.fl6_src, &ireq6->loc_addr); fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_dport =inet_rsk (req)->rmt_port; fl.fl_ip_sport =inet_sk (sk)->sport;security_req_classify_flow (req, &fl);
net/ipv6/tcp_ipv6.c +4 -0
@@ -243,6 +243,7 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,ipv6_addr_copy (&fl.fl6_src, (saddr ? saddr : &np->saddr)); fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_dport = usin->sin6_port; fl.fl_ip_sport = inet->sport;
@@ -383,6 +384,7 @@ static void tcp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,ipv6_addr_copy (&fl.fl6_dst, &np->daddr);ipv6_addr_copy (&fl.fl6_src, &np->saddr); fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_dport = inet->dport; fl.fl_ip_sport = inet->sport;security_skb_classify_flow (skb, &fl);
@@ -477,6 +479,7 @@ static int tcp_v6_send_synack(struct sock *sk, struct request_sock *req)ipv6_addr_copy (&fl.fl6_src, &treq->loc_addr); fl.fl6_flowlabel =0 ; fl.oif = treq->iif; + fl.mark = sk->sk_mark; fl.fl_ip_dport =inet_rsk (req)->rmt_port; fl.fl_ip_sport =inet_rsk (req)->loc_port;security_req_classify_flow (req, &fl);
@@ -1345,6 +1348,7 @@ static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb, }ipv6_addr_copy (&fl.fl6_src, &treq->loc_addr); fl.oif = sk->sk_bound_dev_if; + fl.mark = sk->sk_mark; fl.fl_ip_dport =inet_rsk (req)->rmt_port; fl.fl_ip_sport =inet_rsk (req)->loc_port;security_req_classify_flow (req, &fl);
net/ipv6/udp.c +2 -0
@@ -879,6 +879,8 @@ do_udp_sendmsg:if (!fl.oif) fl.oif = np->sticky_pktinfo.ipi6_ifindex; + fl.mark = sk->sk_mark; +if (msg->msg_controllen) { opt = &opt_space;memset (opt,0 ,sizeof (struct ipv6_txoptions ));
David Miller
maintainerFrom: Eric Dumazet <[email protected]>
Date: Fri, 02 Oct 2009 07:14:46 +0200
Quoted 8 lines
Here is a followup on this area, thanks.
[RFC]
af_packet: fillskb->markat xmit
skb->markmay be used by classifiers, so fill it in case user set aSO_MARKoption on socket.Signed-off-by Eric Dumazet <[email protected]>
Looks fine, applied, thanks!
Re: [PATCH] Use sk_mark for IPv6 routing lookups
In reply to Brian HaleyFrom: Brian Haley <[email protected]>
Date: Fri, 02 Oct 2009 14:19:05 -0400
Quoted 3 lines
Add support for IPv6 route lookups using
sk_mark.Signed-off-by Brian Haley <[email protected]>
Applied, thanks!