Message ID | 1469493272-14894-1-git-send-email-subashab@codeaurora.org |
---|---|
State | RFC, archived |
Delegated to: | David Miller |
Headers | show |
On 7/25/16 6:34 PM, Subash Abhinov Kasiviswanathan wrote: > We are seeing incorrect routing when tunneling packets over an > interface and sending it over another interface. This scenario > worked on 3.18 (and earlier) and failed on 4.4 kernel. The rules > / routes / policies were the same across kernels. Can you give an example of your use case -- e.g., commands for others (me) to reproduce?
On Mon, Jul 25, 2016 at 06:34:32PM -0600, Subash Abhinov Kasiviswanathan wrote: > We are seeing incorrect routing when tunneling packets over an > interface and sending it over another interface. This scenario > worked on 3.18 (and earlier) and failed on 4.4 kernel. The rules > / routes / policies were the same across kernels. > > Commit 42a7b32b73d6 ("xfrm: Add oif to dst lookups") allowed > preservation of the oif from a raw packet to a transformed packet. > This causes issues with forwarding scenarios where the > existing oif causes an incorrect route lookup. > > Create a new sysctl which resets oif in xfrm policy. Default value > is 0 which means that oif is preserved on transform. Please don't try to workaround a bug with a sysctl. If we have a bug here, we should fix it. Choosing between bug A and bug B with a sysctl is not what we are doing ;)
> Please don't try to workaround a bug with a sysctl. > If we have a bug here, we should fix it. Choosing > between bug A and bug B with a sysctl is not what > we are doing ;) Sure, this was just a quick hack. > Can you give an example of your use case -- e.g., commands for others > (me) to reproduce? Here is an equivalent set of rules. We see a difference in the oif when reset oif vs preserve it. eth1 is the interface from which traffic is generated while eth0 is the tunnel. -------------- #Commands echo 1 > /proc/sys/net/ipv4/ip_forward echo 1 > /proc/sys/net/ipv4/conf/all/accept_local echo 1 > /proc/sys/net/ipv4/conf/eth0/accept_local echo 1 > /proc/sys/net/ipv4/conf/eth1/accept_local ip addr add 192.168.77.2/24 dev eth0 ip link set eth0 mtu 1400 ip link set eth0 up ip addr add 192.168.33.2/24 dev eth1 ip link set eth1 mtu 1400 ip link set eth1 up ip ru add to 192.168.33.1 lookup 8 prio 4000 ip ru add oif eth1 lookup 8 prio 4010 ip ru add to 192.168.77.1 lookup 9 prio 4030 ip route add default dev eth1 table 8 ip route add default dev eth0 table 9 iptables -t raw -A OUTPUT -j LOG --log-prefix "RAW-OUT >> " iptables -t mangle -A POSTROUTING -j LOG --log-prefix "MAN-PST >> " echo 0 > /proc/sys/net/ipv4/tcp_timestamps # out direction ip xfrm state add src 192.168.77.2 dst 192.168.77.1 proto esp spi 0x1234 mode tunnel enc 'cbc(aes)' 0xbb31df5b207dc1c7a8512eeda0b2d0691e27bc8059dbb82df616bb9955058cd5 auth 'hmac(sha1)' 0x93b43b527d564efb9eac8cd04510b86e409f8ea7 flag af-unspec encap espinudp 4500 4500 0.0.0.0 ip xfrm policy add dir out src 192.168.33.2 tmpl src 192.168.77.2 dst 192.168.77.1 proto esp spi 0x1234 mode tunnel # in direction ip xfrm state add src 192.168.77.1 dst 192.168.77.2 proto esp spi 0x4321 mode tunnel enc 'cbc(aes)' 0x5d3ca96d1af2eaa9cf8f1c1cace88f550e2a5b7b82027023287e1fe2a42f7f54 auth 'hmac(sha1)' 0xcd09f850d7c0dd6dc0ed342619c1165571452f9d flag af-unspec encap espinudp 4500 4500 0.0.0.0 ip xfrm policy add dir in dst 192.168.33.2 tmpl src 192.168.77.1 dst 192.168.77.2 proto esp spi 0x4321 mode tunnel ip xfrm policy add dir fwd dst 192.168.33.2 tmpl src 192.168.77.1 dst 192.168.77.2 proto esp spi 0x4321 mode tunnel -------------- Output when resetting oif (3.18) root@vm:~# ping -c 1 -I eth1 192.168.33.1 PING 192.168.33.1 (192.168.33.1) 56(84) bytes of data. RAW-OUT >> IN= OUT=eth0 SRC=192.168.33.2 DST=192.168.33.1 LEN=84 TOS=0x00 PREC=0x00 TTL=64 ID=801 DF PROTO=ICMP TYPE=8 CODE=0 ID=2040 SEQ=1 MAN-PST >> IN= OUT=eth0 SRC=192.168.33.2 DST=192.168.33.1 LEN=84 TOS=0x00 PREC=0x00 TTL=64 ID=801 DF PROTO=ICMP TYPE=8 CODE=0 ID=2040 SEQ=1 RAW-OUT >> IN= OUT=eth0 SRC=192.168.77.2 DST=192.168.77.1 LEN=160 TOS=0x00 PREC=0x00 TTL=64 ID=41757 DF PROTO=UDP SPT=4500 DPT=4500 LEN=140 MAN-PST >> IN= OUT=eth0 SRC=192.168.77.2 DST=192.168.77.1 LEN=160 TOS=0x00 PREC=0x00 TTL=64 ID=41757 DF PROTO=UDP SPT=4500 DPT=4500 LEN=140 -------------- Output when preserving oif (4.4) root@vm:~# ping -c 1 -I eth1 192.168.33.1 PING 192.168.33.1 (192.168.33.1) 56(84) bytes of data. RAW-OUT >> IN= OUT=eth1 SRC=192.168.33.2 DST=192.168.33.1 LEN=84 TOS=0x00 PREC=0x00 TTL=64 ID=20191 DF PROTO=ICMP TYPE=8 CODE=0 ID=2043 SEQ=1 MAN-PST >> IN= OUT=eth1 SRC=192.168.33.2 DST=192.168.33.1 LEN=84 TOS=0x00 PREC=0x00 TTL=64 ID=20191 DF PROTO=ICMP TYPE=8 CODE=0 ID=2043 SEQ=1 RAW-OUT >> IN= OUT=eth1 SRC=192.168.77.2 DST=192.168.77.1 LEN=160 TOS=0x00 PREC=0x00 TTL=64 ID=49515 DF PROTO=UDP SPT=4500 DPT=4500 LEN=140 MAN-PST >> IN= OUT=eth1 SRC=192.168.77.2 DST=192.168.77.1 LEN=160 TOS=0x00 PREC=0x00 TTL=64 ID=49515 DF PROTO=UDP SPT=4500 DPT=4500 LEN=140 -- Qualcomm Innovation Center, Inc. The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
On 7/29/16 12:21 PM, subashab@codeaurora.org wrote: >> Please don't try to workaround a bug with a sysctl. >> If we have a bug here, we should fix it. Choosing >> between bug A and bug B with a sysctl is not what >> we are doing ;) > > Sure, this was just a quick hack. > >> Can you give an example of your use case -- e.g., commands for others >> (me) to reproduce? > > Here is an equivalent set of rules. We see a difference in the oif when > reset oif vs preserve it. > eth1 is the interface from which traffic is generated while eth0 is the > tunnel. > > -------------- > #Commands > echo 1 > /proc/sys/net/ipv4/ip_forward > echo 1 > /proc/sys/net/ipv4/conf/all/accept_local > echo 1 > /proc/sys/net/ipv4/conf/eth0/accept_local > echo 1 > /proc/sys/net/ipv4/conf/eth1/accept_local > > ip addr add 192.168.77.2/24 dev eth0 > ip link set eth0 mtu 1400 > ip link set eth0 up > > ip addr add 192.168.33.2/24 dev eth1 > ip link set eth1 mtu 1400 > ip link set eth1 up > > ip ru add to 192.168.33.1 lookup 8 prio 4000 > ip ru add oif eth1 lookup 8 prio 4010 > ip ru add to 192.168.77.1 lookup 9 prio 4030 > > ip route add default dev eth1 table 8 > ip route add default dev eth0 table 9 > > iptables -t raw -A OUTPUT -j LOG --log-prefix "RAW-OUT >> " > iptables -t mangle -A POSTROUTING -j LOG --log-prefix "MAN-PST >> " > > echo 0 > /proc/sys/net/ipv4/tcp_timestamps > > # out direction > ip xfrm state add src 192.168.77.2 dst 192.168.77.1 proto esp spi 0x1234 > mode tunnel enc 'cbc(aes)' > 0xbb31df5b207dc1c7a8512eeda0b2d0691e27bc8059dbb82df616bb9955058cd5 auth > 'hmac(sha1)' 0x93b43b527d564efb9eac8cd04510b86e409f8ea7 flag af-unspec > encap espinudp 4500 4500 0.0.0.0 > > ip xfrm policy add dir out src 192.168.33.2 tmpl src 192.168.77.2 dst > 192.168.77.1 proto esp spi 0x1234 mode tunnel > > # in direction > ip xfrm state add src 192.168.77.1 dst 192.168.77.2 proto esp spi 0x4321 > mode tunnel enc 'cbc(aes)' > 0x5d3ca96d1af2eaa9cf8f1c1cace88f550e2a5b7b82027023287e1fe2a42f7f54 auth > 'hmac(sha1)' 0xcd09f850d7c0dd6dc0ed342619c1165571452f9d flag af-unspec > encap espinudp 4500 4500 0.0.0.0 > > ip xfrm policy add dir in dst 192.168.33.2 tmpl src 192.168.77.1 dst > 192.168.77.2 proto esp spi 0x4321 mode tunnel > ip xfrm policy add dir fwd dst 192.168.33.2 tmpl src 192.168.77.1 dst > 192.168.77.2 proto esp spi 0x4321 mode tunnel > -------------- > > Output when resetting oif (3.18) > > root@vm:~# ping -c 1 -I eth1 192.168.33.1 > PING 192.168.33.1 (192.168.33.1) 56(84) bytes of data. > RAW-OUT >> IN= OUT=eth0 SRC=192.168.33.2 DST=192.168.33.1 LEN=84 > TOS=0x00 PREC=0x00 TTL=64 ID=801 DF PROTO=ICMP TYPE=8 CODE=0 ID=2040 SEQ=1 > MAN-PST >> IN= OUT=eth0 SRC=192.168.33.2 DST=192.168.33.1 LEN=84 > TOS=0x00 PREC=0x00 TTL=64 ID=801 DF PROTO=ICMP TYPE=8 CODE=0 ID=2040 SEQ=1 > RAW-OUT >> IN= OUT=eth0 SRC=192.168.77.2 DST=192.168.77.1 LEN=160 > TOS=0x00 PREC=0x00 TTL=64 ID=41757 DF PROTO=UDP SPT=4500 DPT=4500 LEN=140 > MAN-PST >> IN= OUT=eth0 SRC=192.168.77.2 DST=192.168.77.1 LEN=160 > TOS=0x00 PREC=0x00 TTL=64 ID=41757 DF PROTO=UDP SPT=4500 DPT=4500 LEN=140 > > -------------- > > Output when preserving oif (4.4) > > root@vm:~# ping -c 1 -I eth1 192.168.33.1 > PING 192.168.33.1 (192.168.33.1) 56(84) bytes of data. > RAW-OUT >> IN= OUT=eth1 SRC=192.168.33.2 DST=192.168.33.1 LEN=84 > TOS=0x00 PREC=0x00 TTL=64 ID=20191 DF PROTO=ICMP TYPE=8 CODE=0 ID=2043 > SEQ=1 > MAN-PST >> IN= OUT=eth1 SRC=192.168.33.2 DST=192.168.33.1 LEN=84 > TOS=0x00 PREC=0x00 TTL=64 ID=20191 DF PROTO=ICMP TYPE=8 CODE=0 ID=2043 > SEQ=1 > RAW-OUT >> IN= OUT=eth1 SRC=192.168.77.2 DST=192.168.77.1 LEN=160 > TOS=0x00 PREC=0x00 TTL=64 ID=49515 DF PROTO=UDP SPT=4500 DPT=4500 LEN=140 > MAN-PST >> IN= OUT=eth1 SRC=192.168.77.2 DST=192.168.77.1 LEN=160 > TOS=0x00 PREC=0x00 TTL=64 ID=49515 DF PROTO=UDP SPT=4500 DPT=4500 LEN=140 > I can't explain the iptables output but from a FIB lookup perspective it is using table 8 per the FIB rules, the xfrm is hit and packets shift to 192.168.77.1 and go out what you have as eth0. Take a look at: perf record -e fib:* -a -g perf script And then run tcpdump on both eth0 and eth1. For me on "eth0" (which is really eth11 for my VM setup) I see this on the ping: 20:50:11.389837 ARP, Request who-has 192.168.77.2 tell 192.168.77.1, length 28 20:50:11.390079 ARP, Reply 192.168.77.2 is-at 02:00:12:34:02:0a, length 28 20:50:11.390101 IP 192.168.77.1 > 192.168.77.2: ICMP 192.168.77.1 udp port 4500 unreachable, length 168 So the packets are going out "eth0" as expected. That said, the commands you have given do not totally transfer to another setup. In my case I have 2 VMs with eth11 and eth12 directly connected (VM1 eth11 <--> VM2 eth11 and ditto for eth12). You have given one side of the commands and I have configured the other side with the .1 addresses but not bothered to translate the xfrm commands. That said, this seems like a contrived example -- you pin ping to device eth1 (-I eth1), you are pinging a host on the network for eth1 but want packets to go out eth0 via the xfrm. Can you elaborate on the real use case and problem here?
> I can't explain the iptables output but from a FIB lookup perspective > it is using table 8 per the FIB rules, the xfrm is hit and packets > shift to 192.168.77.1 and go out what you have as eth0. > > Take a look at: > perf record -e fib:* -a -g > perf script > > And then run tcpdump on both eth0 and eth1. For me on "eth0" (which is > really eth11 for my VM setup) I see this on the ping: > You can try running these commands as is on UML. We tried these out on 3.18 as well as on 4.4. > 20:50:11.389837 ARP, Request who-has 192.168.77.2 tell 192.168.77.1, > length 28 > 20:50:11.390079 ARP, Reply 192.168.77.2 is-at 02:00:12:34:02:0a, length > 28 > 20:50:11.390101 IP 192.168.77.1 > 192.168.77.2: ICMP 192.168.77.1 udp > port 4500 unreachable, length 168 > > So the packets are going out "eth0" as expected. > > That said, the commands you have given do not totally transfer to > another setup. In my case I have 2 VMs with eth11 and eth12 directly > connected (VM1 eth11 <--> VM2 eth11 and ditto for eth12). You have > given one side of the commands and I have configured the other side > with the .1 addresses but not bothered to translate the xfrm commands. > > That said, this seems like a contrived example -- you pin ping to > device eth1 (-I eth1), you are pinging a host on the network for eth1 > but want packets to go out eth0 via the xfrm. Can you elaborate on the > real use case and problem here? Applications may be bound to a specific interface but would try to send data over multiple types of networks. Our use case here is wifi calling. In this case, we try to force packets to go over wifi after encryption. The rules which we were using worked on 3.18 but we ran into issues on 4.4. Debugging narrowed us down to this oif preservation through xfrm. -- Qualcomm Innovation Center, Inc. The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
diff --git a/include/net/netns/xfrm.h b/include/net/netns/xfrm.h index 24cd394..2e1beca 100644 --- a/include/net/netns/xfrm.h +++ b/include/net/netns/xfrm.h @@ -64,6 +64,7 @@ struct netns_xfrm { u32 sysctl_aevent_rseqth; int sysctl_larval_drop; u32 sysctl_acq_expires; + int sysctl_reset_oif; #ifdef CONFIG_SYSCTL struct ctl_table_header *sysctl_hdr; #endif diff --git a/net/ipv4/xfrm4_policy.c b/net/ipv4/xfrm4_policy.c index 7b0edb3..4dc3733 100644 --- a/net/ipv4/xfrm4_policy.c +++ b/net/ipv4/xfrm4_policy.c @@ -29,7 +29,8 @@ static struct dst_entry *__xfrm4_dst_lookup(struct net *net, struct flowi4 *fl4, memset(fl4, 0, sizeof(*fl4)); fl4->daddr = daddr->a4; fl4->flowi4_tos = tos; - fl4->flowi4_oif = oif; + if (!net->xfrm.sysctl_reset_oif) + fl4->flowi4_oif = oif; if (saddr) fl4->saddr = saddr->a4; diff --git a/net/ipv6/xfrm6_policy.c b/net/ipv6/xfrm6_policy.c index c074771..13e72d7 100644 --- a/net/ipv6/xfrm6_policy.c +++ b/net/ipv6/xfrm6_policy.c @@ -36,7 +36,8 @@ static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos, int oif, int err; memset(&fl6, 0, sizeof(fl6)); - fl6.flowi6_oif = oif; + if (!net->xfrm.sysctl_reset_oif) + fl6.flowi6_oif = oif; fl6.flowi6_flags = FLOWI_FLAG_SKIP_NH_OIF; memcpy(&fl6.daddr, daddr, sizeof(fl6.daddr)); if (saddr) diff --git a/net/xfrm/xfrm_sysctl.c b/net/xfrm/xfrm_sysctl.c index 05a6e3d..c9d374b 100644 --- a/net/xfrm/xfrm_sysctl.c +++ b/net/xfrm/xfrm_sysctl.c @@ -9,6 +9,7 @@ static void __net_init __xfrm_sysctl_init(struct net *net) net->xfrm.sysctl_aevent_rseqth = XFRM_AE_SEQT_SIZE; net->xfrm.sysctl_larval_drop = 1; net->xfrm.sysctl_acq_expires = 30; + net->xfrm.sysctl_reset_oif = 0; } #ifdef CONFIG_SYSCTL @@ -37,6 +38,12 @@ static struct ctl_table xfrm_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "xfrm_reset_oif", + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, {} }; @@ -53,6 +60,7 @@ int __net_init xfrm_sysctl_init(struct net *net) table[1].data = &net->xfrm.sysctl_aevent_rseqth; table[2].data = &net->xfrm.sysctl_larval_drop; table[3].data = &net->xfrm.sysctl_acq_expires; + table[4].data = &net->xfrm.sysctl_reset_oif; /* Don't export sysctls to unprivileged users */ if (net->user_ns != &init_user_ns)
We are seeing incorrect routing when tunneling packets over an interface and sending it over another interface. This scenario worked on 3.18 (and earlier) and failed on 4.4 kernel. The rules / routes / policies were the same across kernels. Commit 42a7b32b73d6 ("xfrm: Add oif to dst lookups") allowed preservation of the oif from a raw packet to a transformed packet. This causes issues with forwarding scenarios where the existing oif causes an incorrect route lookup. Create a new sysctl which resets oif in xfrm policy. Default value is 0 which means that oif is preserved on transform. Signed-off-by: Subash Abhinov Kasiviswanathan <subashab@codeaurora.org> --- include/net/netns/xfrm.h | 1 + net/ipv4/xfrm4_policy.c | 3 ++- net/ipv6/xfrm6_policy.c | 3 ++- net/xfrm/xfrm_sysctl.c | 8 ++++++++ 4 files changed, 13 insertions(+), 2 deletions(-)