linux/net/ipv6/ip6_output.c

1747 lines
44 KiB
C
Raw Normal View History

/*
* IPv6 output functions
* Linux INET6 implementation
*
* Authors:
* Pedro Roque <roque@di.fc.ul.pt>
*
* Based on linux/net/ipv4/ip_output.c
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*
* Changes:
* A.N.Kuznetsov : airthmetics in fragmentation.
* extension headers are implemented.
* route changes now work.
* ip6_forward does not confuse sniffers.
* etc.
*
* H. von Brand : Added missing #include <linux/string.h>
* Imran Patel : frag id should be in NBO
* Kazunori MIYAZAWA @USAGI
* : add ip6_append_data and related functions
* for datagram xmit
*/
#include <linux/errno.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/socket.h>
#include <linux/net.h>
#include <linux/netdevice.h>
#include <linux/if_arp.h>
#include <linux/in6.h>
#include <linux/tcp.h>
#include <linux/route.h>
#include <linux/module.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include <linux/bpf-cgroup.h>
#include <linux/netfilter.h>
#include <linux/netfilter_ipv6.h>
#include <net/sock.h>
#include <net/snmp.h>
#include <net/ipv6.h>
#include <net/ndisc.h>
#include <net/protocol.h>
#include <net/ip6_route.h>
#include <net/addrconf.h>
#include <net/rawv6.h>
#include <net/icmp.h>
#include <net/xfrm.h>
#include <net/checksum.h>
#include <linux/mroute6.h>
#include <net/l3mdev.h>
#include <net/lwtunnel.h>
static int ip6_finish_output2(struct net *net, struct sock *sk, struct sk_buff *skb)
{
struct dst_entry *dst = skb_dst(skb);
struct net_device *dev = dst->dev;
struct neighbour *neigh;
struct in6_addr *nexthop;
int ret;
if (ipv6_addr_is_multicast(&ipv6_hdr(skb)->daddr)) {
struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
if (!(dev->flags & IFF_LOOPBACK) && sk_mc_loop(sk) &&
((mroute6_socket(net, skb) &&
!(IP6CB(skb)->flags & IP6SKB_FORWARDED)) ||
ipv6_chk_mcast_addr(dev, &ipv6_hdr(skb)->daddr,
&ipv6_hdr(skb)->saddr))) {
struct sk_buff *newskb = skb_clone(skb, GFP_ATOMIC);
/* Do not check for IFF_ALLMULTI; multicast routing
is not supported in any case.
*/
if (newskb)
NF_HOOK(NFPROTO_IPV6, NF_INET_POST_ROUTING,
2015-09-16 01:04:16 +00:00
net, sk, newskb, NULL, newskb->dev,
dev_loopback_xmit);
if (ipv6_hdr(skb)->hop_limit == 0) {
IP6_INC_STATS(net, idev,
IPSTATS_MIB_OUTDISCARDS);
kfree_skb(skb);
return 0;
}
}
IP6_UPD_PO_STATS(net, idev, IPSTATS_MIB_OUTMCAST, skb->len);
if (IPV6_ADDR_MC_SCOPE(&ipv6_hdr(skb)->daddr) <=
IPV6_ADDR_SCOPE_NODELOCAL &&
!(dev->flags & IFF_LOOPBACK)) {
kfree_skb(skb);
return 0;
}
}
if (lwtunnel_xmit_redirect(dst->lwtstate)) {
int res = lwtunnel_xmit(skb);
if (res < 0 || res == LWTUNNEL_XMIT_DONE)
return res;
}
rcu_read_lock_bh();
nexthop = rt6_nexthop((struct rt6_info *)dst, &ipv6_hdr(skb)->daddr);
neigh = __ipv6_neigh_lookup_noref(dst->dev, nexthop);
if (unlikely(!neigh))
neigh = __neigh_create(&nd_tbl, nexthop, dst->dev, false);
if (!IS_ERR(neigh)) {
sock_confirm_neigh(skb, neigh);
ret = neigh_output(neigh, skb);
rcu_read_unlock_bh();
return ret;
}
rcu_read_unlock_bh();
IP6_INC_STATS(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTNOROUTES);
kfree_skb(skb);
return -EINVAL;
}
static int ip6_finish_output(struct net *net, struct sock *sk, struct sk_buff *skb)
{
int ret;
ret = BPF_CGROUP_RUN_PROG_INET_EGRESS(sk, skb);
if (ret) {
kfree_skb(skb);
return ret;
}
if ((skb->len > ip6_skb_dst_mtu(skb) && !skb_is_gso(skb)) ||
dst_allfrag(skb_dst(skb)) ||
(IP6CB(skb)->frag_max_size && skb->len > IP6CB(skb)->frag_max_size))
return ip6_fragment(net, sk, skb, ip6_finish_output2);
else
return ip6_finish_output2(net, sk, skb);
}
int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb)
{
struct net_device *dev = skb_dst(skb)->dev;
struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
skb->protocol = htons(ETH_P_IPV6);
skb->dev = dev;
if (unlikely(idev->cnf.disable_ipv6)) {
IP6_INC_STATS(net, idev, IPSTATS_MIB_OUTDISCARDS);
kfree_skb(skb);
return 0;
}
2015-09-16 01:04:16 +00:00
return NF_HOOK_COND(NFPROTO_IPV6, NF_INET_POST_ROUTING,
net, sk, skb, NULL, dev,
ip6_finish_output,
!(IP6CB(skb)->flags & IP6SKB_REROUTED));
}
/*
* xmit an sk_buff (used by TCP, SCTP and DCCP)
* Note : socket lock is not held for SYNACK packets, but might be modified
* by calls to skb_set_owner_w() and ipv6_local_error(),
* which are using proper atomic operations or spinlocks.
*/
int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
__u32 mark, struct ipv6_txoptions *opt, int tclass)
{
struct net *net = sock_net(sk);
const struct ipv6_pinfo *np = inet6_sk(sk);
struct in6_addr *first_hop = &fl6->daddr;
struct dst_entry *dst = skb_dst(skb);
struct ipv6hdr *hdr;
u8 proto = fl6->flowi6_proto;
int seg_len = skb->len;
int hlimit = -1;
u32 mtu;
if (opt) {
unsigned int head_room;
/* First: exthdrs may take lots of space (~8K for now)
MAX_HEADER is not enough.
*/
head_room = opt->opt_nflen + opt->opt_flen;
seg_len += head_room;
head_room += sizeof(struct ipv6hdr) + LL_RESERVED_SPACE(dst->dev);
if (skb_headroom(skb) < head_room) {
struct sk_buff *skb2 = skb_realloc_headroom(skb, head_room);
if (!skb2) {
IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
IPSTATS_MIB_OUTDISCARDS);
kfree_skb(skb);
return -ENOBUFS;
}
consume_skb(skb);
skb = skb2;
/* skb_set_owner_w() changes sk->sk_wmem_alloc atomically,
* it is safe to call in our context (socket lock not held)
*/
skb_set_owner_w(skb, (struct sock *)sk);
}
if (opt->opt_flen)
ipv6_push_frag_opts(skb, opt, &proto);
if (opt->opt_nflen)
ipv6_push_nfrag_opts(skb, opt, &proto, &first_hop,
&fl6->saddr);
}
skb_push(skb, sizeof(struct ipv6hdr));
skb_reset_network_header(skb);
hdr = ipv6_hdr(skb);
/*
* Fill in the IPv6 header
*/
if (np)
hlimit = np->hop_limit;
if (hlimit < 0)
hlimit = ip6_dst_hoplimit(dst);
ipv6: Implement automatic flow label generation on transmit Automatically generate flow labels for IPv6 packets on transmit. The flow label is computed based on skb_get_hash. The flow label will only automatically be set when it is zero otherwise (i.e. flow label manager hasn't set one). This supports the transmit side functionality of RFC 6438. Added an IPv6 sysctl auto_flowlabels to enable/disable this behavior system wide, and added IPV6_AUTOFLOWLABEL socket option to enable this functionality per socket. By default, auto flowlabels are disabled to avoid possible conflicts with flow label manager, however if this feature proves useful we may want to enable it by default. It should also be noted that FreeBSD has already implemented automatic flow labels (including the sysctl and socket option). In FreeBSD, automatic flow labels default to enabled. Performance impact: Running super_netperf with 200 flows for TCP_RR and UDP_RR for IPv6. Note that in UDP case, __skb_get_hash will be called for every packet with explains slight regression. In the TCP case the hash is saved in the socket so there is no regression. Automatic flow labels disabled: TCP_RR: 86.53% CPU utilization 127/195/322 90/95/99% latencies 1.40498e+06 tps UDP_RR: 90.70% CPU utilization 118/168/243 90/95/99% latencies 1.50309e+06 tps Automatic flow labels enabled: TCP_RR: 85.90% CPU utilization 128/199/337 90/95/99% latencies 1.40051e+06 UDP_RR 92.61% CPU utilization 115/164/236 90/95/99% latencies 1.4687e+06 Signed-off-by: Tom Herbert <therbert@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-07-02 04:33:10 +00:00
ip6_flow_hdr(hdr, tclass, ip6_make_flowlabel(net, skb, fl6->flowlabel,
np->autoflowlabel, fl6));
hdr->payload_len = htons(seg_len);
hdr->nexthdr = proto;
hdr->hop_limit = hlimit;
hdr->saddr = fl6->saddr;
hdr->daddr = *first_hop;
skb->protocol = htons(ETH_P_IPV6);
skb->priority = sk->sk_priority;
skb->mark = mark;
mtu = dst_mtu(dst);
if ((skb->len <= mtu) || skb->ignore_df || skb_is_gso(skb)) {
IP6_UPD_PO_STATS(net, ip6_dst_idev(skb_dst(skb)),
IPSTATS_MIB_OUT, skb->len);
/* if egress device is enslaved to an L3 master device pass the
* skb to its handler for processing
*/
skb = l3mdev_ip6_out((struct sock *)sk, skb);
if (unlikely(!skb))
return 0;
/* hooks should never assume socket lock is held.
* we promote our socket to non const
*/
2015-09-16 01:04:16 +00:00
return NF_HOOK(NFPROTO_IPV6, NF_INET_LOCAL_OUT,
net, (struct sock *)sk, skb, NULL, dst->dev,
dst_output);
}
skb->dev = dst->dev;
/* ipv6_local_error() does not require socket lock,
* we promote our socket to non const
*/
ipv6_local_error((struct sock *)sk, EMSGSIZE, fl6, mtu);
IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)), IPSTATS_MIB_FRAGFAILS);
kfree_skb(skb);
return -EMSGSIZE;
}
EXPORT_SYMBOL(ip6_xmit);
static int ip6_call_ra_chain(struct sk_buff *skb, int sel)
{
struct ip6_ra_chain *ra;
struct sock *last = NULL;
read_lock(&ip6_ra_lock);
for (ra = ip6_ra_chain; ra; ra = ra->next) {
struct sock *sk = ra->sk;
if (sk && ra->sel == sel &&
(!sk->sk_bound_dev_if ||
sk->sk_bound_dev_if == skb->dev->ifindex)) {
if (last) {
struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
if (skb2)
rawv6_rcv(last, skb2);
}
last = sk;
}
}
if (last) {
rawv6_rcv(last, skb);
read_unlock(&ip6_ra_lock);
return 1;
}
read_unlock(&ip6_ra_lock);
return 0;
}
static int ip6_forward_proxy_check(struct sk_buff *skb)
{
struct ipv6hdr *hdr = ipv6_hdr(skb);
u8 nexthdr = hdr->nexthdr;
__be16 frag_off;
int offset;
if (ipv6_ext_hdr(nexthdr)) {
offset = ipv6_skip_exthdr(skb, sizeof(*hdr), &nexthdr, &frag_off);
if (offset < 0)
return 0;
} else
offset = sizeof(struct ipv6hdr);
if (nexthdr == IPPROTO_ICMPV6) {
struct icmp6hdr *icmp6;
if (!pskb_may_pull(skb, (skb_network_header(skb) +
offset + 1 - skb->data)))
return 0;
icmp6 = (struct icmp6hdr *)(skb_network_header(skb) + offset);
switch (icmp6->icmp6_type) {
case NDISC_ROUTER_SOLICITATION:
case NDISC_ROUTER_ADVERTISEMENT:
case NDISC_NEIGHBOUR_SOLICITATION:
case NDISC_NEIGHBOUR_ADVERTISEMENT:
case NDISC_REDIRECT:
/* For reaction involving unicast neighbor discovery
* message destined to the proxied address, pass it to
* input function.
*/
return 1;
default:
break;
}
}
/*
* The proxying router can't forward traffic sent to a link-local
* address, so signal the sender and discard the packet. This
* behavior is clarified by the MIPv6 specification.
*/
if (ipv6_addr_type(&hdr->daddr) & IPV6_ADDR_LINKLOCAL) {
dst_link_failure(skb);
return -1;
}
return 0;
}
static inline int ip6_forward_finish(struct net *net, struct sock *sk,
struct sk_buff *skb)
{
return dst_output(net, sk, skb);
}
static unsigned int ip6_dst_mtu_forward(const struct dst_entry *dst)
{
unsigned int mtu;
struct inet6_dev *idev;
if (dst_metric_locked(dst, RTAX_MTU)) {
mtu = dst_metric_raw(dst, RTAX_MTU);
if (mtu)
return mtu;
}
mtu = IPV6_MIN_MTU;
rcu_read_lock();
idev = __in6_dev_get(dst->dev);
if (idev)
mtu = idev->cnf.mtu6;
rcu_read_unlock();
return mtu;
}
net: ip, ipv6: handle gso skbs in forwarding path Marcelo Ricardo Leitner reported problems when the forwarding link path has a lower mtu than the incoming one if the inbound interface supports GRO. Given: Host <mtu1500> R1 <mtu1200> R2 Host sends tcp stream which is routed via R1 and R2. R1 performs GRO. In this case, the kernel will fail to send ICMP fragmentation needed messages (or pkt too big for ipv6), as GSO packets currently bypass dstmtu checks in forward path. Instead, Linux tries to send out packets exceeding the mtu. When locking route MTU on Host (i.e., no ipv4 DF bit set), R1 does not fragment the packets when forwarding, and again tries to send out packets exceeding R1-R2 link mtu. This alters the forwarding dstmtu checks to take the individual gso segment lengths into account. For ipv6, we send out pkt too big error for gso if the individual segments are too big. For ipv4, we either send icmp fragmentation needed, or, if the DF bit is not set, perform software segmentation and let the output path create fragments when the packet is leaving the machine. It is not 100% correct as the error message will contain the headers of the GRO skb instead of the original/segmented one, but it seems to work fine in my (limited) tests. Eric Dumazet suggested to simply shrink mss via ->gso_size to avoid sofware segmentation. However it turns out that skb_segment() assumes skb nr_frags is related to mss size so we would BUG there. I don't want to mess with it considering Herbert and Eric disagree on what the correct behavior should be. Hannes Frederic Sowa notes that when we would shrink gso_size skb_segment would then also need to deal with the case where SKB_MAX_FRAGS would be exceeded. This uses sofware segmentation in the forward path when we hit ipv4 non-DF packets and the outgoing link mtu is too small. Its not perfect, but given the lack of bug reports wrt. GRO fwd being broken this is a rare case anyway. Also its not like this could not be improved later once the dust settles. Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Reported-by: Marcelo Ricardo Leitner <mleitner@redhat.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-02-13 22:09:12 +00:00
static bool ip6_pkt_too_big(const struct sk_buff *skb, unsigned int mtu)
{
if (skb->len <= mtu)
net: ip, ipv6: handle gso skbs in forwarding path Marcelo Ricardo Leitner reported problems when the forwarding link path has a lower mtu than the incoming one if the inbound interface supports GRO. Given: Host <mtu1500> R1 <mtu1200> R2 Host sends tcp stream which is routed via R1 and R2. R1 performs GRO. In this case, the kernel will fail to send ICMP fragmentation needed messages (or pkt too big for ipv6), as GSO packets currently bypass dstmtu checks in forward path. Instead, Linux tries to send out packets exceeding the mtu. When locking route MTU on Host (i.e., no ipv4 DF bit set), R1 does not fragment the packets when forwarding, and again tries to send out packets exceeding R1-R2 link mtu. This alters the forwarding dstmtu checks to take the individual gso segment lengths into account. For ipv6, we send out pkt too big error for gso if the individual segments are too big. For ipv4, we either send icmp fragmentation needed, or, if the DF bit is not set, perform software segmentation and let the output path create fragments when the packet is leaving the machine. It is not 100% correct as the error message will contain the headers of the GRO skb instead of the original/segmented one, but it seems to work fine in my (limited) tests. Eric Dumazet suggested to simply shrink mss via ->gso_size to avoid sofware segmentation. However it turns out that skb_segment() assumes skb nr_frags is related to mss size so we would BUG there. I don't want to mess with it considering Herbert and Eric disagree on what the correct behavior should be. Hannes Frederic Sowa notes that when we would shrink gso_size skb_segment would then also need to deal with the case where SKB_MAX_FRAGS would be exceeded. This uses sofware segmentation in the forward path when we hit ipv4 non-DF packets and the outgoing link mtu is too small. Its not perfect, but given the lack of bug reports wrt. GRO fwd being broken this is a rare case anyway. Also its not like this could not be improved later once the dust settles. Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Reported-by: Marcelo Ricardo Leitner <mleitner@redhat.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-02-13 22:09:12 +00:00
return false;
/* ipv6 conntrack defrag sets max_frag_size + ignore_df */
net: ip, ipv6: handle gso skbs in forwarding path Marcelo Ricardo Leitner reported problems when the forwarding link path has a lower mtu than the incoming one if the inbound interface supports GRO. Given: Host <mtu1500> R1 <mtu1200> R2 Host sends tcp stream which is routed via R1 and R2. R1 performs GRO. In this case, the kernel will fail to send ICMP fragmentation needed messages (or pkt too big for ipv6), as GSO packets currently bypass dstmtu checks in forward path. Instead, Linux tries to send out packets exceeding the mtu. When locking route MTU on Host (i.e., no ipv4 DF bit set), R1 does not fragment the packets when forwarding, and again tries to send out packets exceeding R1-R2 link mtu. This alters the forwarding dstmtu checks to take the individual gso segment lengths into account. For ipv6, we send out pkt too big error for gso if the individual segments are too big. For ipv4, we either send icmp fragmentation needed, or, if the DF bit is not set, perform software segmentation and let the output path create fragments when the packet is leaving the machine. It is not 100% correct as the error message will contain the headers of the GRO skb instead of the original/segmented one, but it seems to work fine in my (limited) tests. Eric Dumazet suggested to simply shrink mss via ->gso_size to avoid sofware segmentation. However it turns out that skb_segment() assumes skb nr_frags is related to mss size so we would BUG there. I don't want to mess with it considering Herbert and Eric disagree on what the correct behavior should be. Hannes Frederic Sowa notes that when we would shrink gso_size skb_segment would then also need to deal with the case where SKB_MAX_FRAGS would be exceeded. This uses sofware segmentation in the forward path when we hit ipv4 non-DF packets and the outgoing link mtu is too small. Its not perfect, but given the lack of bug reports wrt. GRO fwd being broken this is a rare case anyway. Also its not like this could not be improved later once the dust settles. Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Reported-by: Marcelo Ricardo Leitner <mleitner@redhat.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-02-13 22:09:12 +00:00
if (IP6CB(skb)->frag_max_size && IP6CB(skb)->frag_max_size > mtu)
return true;
if (skb->ignore_df)
return false;
if (skb_is_gso(skb) && skb_gso_validate_mtu(skb, mtu))
net: ip, ipv6: handle gso skbs in forwarding path Marcelo Ricardo Leitner reported problems when the forwarding link path has a lower mtu than the incoming one if the inbound interface supports GRO. Given: Host <mtu1500> R1 <mtu1200> R2 Host sends tcp stream which is routed via R1 and R2. R1 performs GRO. In this case, the kernel will fail to send ICMP fragmentation needed messages (or pkt too big for ipv6), as GSO packets currently bypass dstmtu checks in forward path. Instead, Linux tries to send out packets exceeding the mtu. When locking route MTU on Host (i.e., no ipv4 DF bit set), R1 does not fragment the packets when forwarding, and again tries to send out packets exceeding R1-R2 link mtu. This alters the forwarding dstmtu checks to take the individual gso segment lengths into account. For ipv6, we send out pkt too big error for gso if the individual segments are too big. For ipv4, we either send icmp fragmentation needed, or, if the DF bit is not set, perform software segmentation and let the output path create fragments when the packet is leaving the machine. It is not 100% correct as the error message will contain the headers of the GRO skb instead of the original/segmented one, but it seems to work fine in my (limited) tests. Eric Dumazet suggested to simply shrink mss via ->gso_size to avoid sofware segmentation. However it turns out that skb_segment() assumes skb nr_frags is related to mss size so we would BUG there. I don't want to mess with it considering Herbert and Eric disagree on what the correct behavior should be. Hannes Frederic Sowa notes that when we would shrink gso_size skb_segment would then also need to deal with the case where SKB_MAX_FRAGS would be exceeded. This uses sofware segmentation in the forward path when we hit ipv4 non-DF packets and the outgoing link mtu is too small. Its not perfect, but given the lack of bug reports wrt. GRO fwd being broken this is a rare case anyway. Also its not like this could not be improved later once the dust settles. Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Reported-by: Marcelo Ricardo Leitner <mleitner@redhat.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-02-13 22:09:12 +00:00
return false;
return true;
}
int ip6_forward(struct sk_buff *skb)
{
struct dst_entry *dst = skb_dst(skb);
struct ipv6hdr *hdr = ipv6_hdr(skb);
struct inet6_skb_parm *opt = IP6CB(skb);
struct net *net = dev_net(dst->dev);
u32 mtu;
if (net->ipv6.devconf_all->forwarding == 0)
goto error;
if (skb->pkt_type != PACKET_HOST)
goto drop;
if (unlikely(skb->sk))
goto drop;
if (skb_warn_if_lro(skb))
goto drop;
if (!xfrm6_policy_check(NULL, XFRM_POLICY_FWD, skb)) {
__IP6_INC_STATS(net, ip6_dst_idev(dst),
IPSTATS_MIB_INDISCARDS);
goto drop;
}
skb_forward_csum(skb);
/*
* We DO NOT make any processing on
* RA packets, pushing them to user level AS IS
* without ane WARRANTY that application will be able
* to interpret them. The reason is that we
* cannot make anything clever here.
*
* We are not end-node, so that if packet contains
* AH/ESP, we cannot make anything.
* Defragmentation also would be mistake, RA packets
* cannot be fragmented, because there is no warranty
* that different fragments will go along one path. --ANK
*/
if (unlikely(opt->flags & IP6SKB_ROUTERALERT)) {
if (ip6_call_ra_chain(skb, ntohs(opt->ra)))
return 0;
}
/*
* check and decrement ttl
*/
if (hdr->hop_limit <= 1) {
/* Force OUTPUT device used as source address */
skb->dev = dst->dev;
icmpv6_send(skb, ICMPV6_TIME_EXCEED, ICMPV6_EXC_HOPLIMIT, 0);
__IP6_INC_STATS(net, ip6_dst_idev(dst),
IPSTATS_MIB_INHDRERRORS);
kfree_skb(skb);
return -ETIMEDOUT;
}
/* XXX: idev->cnf.proxy_ndp? */
if (net->ipv6.devconf_all->proxy_ndp &&
pneigh_lookup(&nd_tbl, net, &hdr->daddr, skb->dev, 0)) {
int proxied = ip6_forward_proxy_check(skb);
if (proxied > 0)
return ip6_input(skb);
else if (proxied < 0) {
__IP6_INC_STATS(net, ip6_dst_idev(dst),
IPSTATS_MIB_INDISCARDS);
goto drop;
}
}
if (!xfrm6_route_forward(skb)) {
__IP6_INC_STATS(net, ip6_dst_idev(dst),
IPSTATS_MIB_INDISCARDS);
goto drop;
}
dst = skb_dst(skb);
/* IPv6 specs say nothing about it, but it is clear that we cannot
send redirects to source routed frames.
We don't send redirects to frames decapsulated from IPsec.
*/
if (skb->dev == dst->dev && opt->srcrt == 0 && !skb_sec_path(skb)) {
struct in6_addr *target = NULL;
struct inet_peer *peer;
struct rt6_info *rt;
/*
* incoming and outgoing devices are the same
* send a redirect.
*/
rt = (struct rt6_info *) dst;
if (rt->rt6i_flags & RTF_GATEWAY)
target = &rt->rt6i_gateway;
else
target = &hdr->daddr;
peer = inet_getpeer_v6(net->ipv6.peers, &hdr->daddr, 1);
/* Limit redirects both by destination (here)
and by source (inside ndisc_send_redirect)
*/
if (inet_peer_xrlim_allow(peer, 1*HZ))
ndisc_send_redirect(skb, target);
if (peer)
inet_putpeer(peer);
} else {
int addrtype = ipv6_addr_type(&hdr->saddr);
/* This check is security critical. */
if (addrtype == IPV6_ADDR_ANY ||
addrtype & (IPV6_ADDR_MULTICAST | IPV6_ADDR_LOOPBACK))
goto error;
if (addrtype & IPV6_ADDR_LINKLOCAL) {
icmpv6_send(skb, ICMPV6_DEST_UNREACH,
ICMPV6_NOT_NEIGHBOUR, 0);
goto error;
}
}
mtu = ip6_dst_mtu_forward(dst);
if (mtu < IPV6_MIN_MTU)
mtu = IPV6_MIN_MTU;
net: ip, ipv6: handle gso skbs in forwarding path Marcelo Ricardo Leitner reported problems when the forwarding link path has a lower mtu than the incoming one if the inbound interface supports GRO. Given: Host <mtu1500> R1 <mtu1200> R2 Host sends tcp stream which is routed via R1 and R2. R1 performs GRO. In this case, the kernel will fail to send ICMP fragmentation needed messages (or pkt too big for ipv6), as GSO packets currently bypass dstmtu checks in forward path. Instead, Linux tries to send out packets exceeding the mtu. When locking route MTU on Host (i.e., no ipv4 DF bit set), R1 does not fragment the packets when forwarding, and again tries to send out packets exceeding R1-R2 link mtu. This alters the forwarding dstmtu checks to take the individual gso segment lengths into account. For ipv6, we send out pkt too big error for gso if the individual segments are too big. For ipv4, we either send icmp fragmentation needed, or, if the DF bit is not set, perform software segmentation and let the output path create fragments when the packet is leaving the machine. It is not 100% correct as the error message will contain the headers of the GRO skb instead of the original/segmented one, but it seems to work fine in my (limited) tests. Eric Dumazet suggested to simply shrink mss via ->gso_size to avoid sofware segmentation. However it turns out that skb_segment() assumes skb nr_frags is related to mss size so we would BUG there. I don't want to mess with it considering Herbert and Eric disagree on what the correct behavior should be. Hannes Frederic Sowa notes that when we would shrink gso_size skb_segment would then also need to deal with the case where SKB_MAX_FRAGS would be exceeded. This uses sofware segmentation in the forward path when we hit ipv4 non-DF packets and the outgoing link mtu is too small. Its not perfect, but given the lack of bug reports wrt. GRO fwd being broken this is a rare case anyway. Also its not like this could not be improved later once the dust settles. Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Reported-by: Marcelo Ricardo Leitner <mleitner@redhat.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-02-13 22:09:12 +00:00
if (ip6_pkt_too_big(skb, mtu)) {
/* Again, force OUTPUT device used as source address */
skb->dev = dst->dev;
icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
__IP6_INC_STATS(net, ip6_dst_idev(dst),
IPSTATS_MIB_INTOOBIGERRORS);
__IP6_INC_STATS(net, ip6_dst_idev(dst),
IPSTATS_MIB_FRAGFAILS);
kfree_skb(skb);
return -EMSGSIZE;
}
if (skb_cow(skb, dst->dev->hard_header_len)) {
__IP6_INC_STATS(net, ip6_dst_idev(dst),
IPSTATS_MIB_OUTDISCARDS);
goto drop;
}
hdr = ipv6_hdr(skb);
/* Mangling hops number delayed to point after skb COW */
hdr->hop_limit--;
__IP6_INC_STATS(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTFORWDATAGRAMS);
__IP6_ADD_STATS(net, ip6_dst_idev(dst), IPSTATS_MIB_OUTOCTETS, skb->len);
2015-09-16 01:04:16 +00:00
return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD,
net, NULL, skb, skb->dev, dst->dev,
ip6_forward_finish);
error:
__IP6_INC_STATS(net, ip6_dst_idev(dst), IPSTATS_MIB_INADDRERRORS);
drop:
kfree_skb(skb);
return -EINVAL;
}
static void ip6_copy_metadata(struct sk_buff *to, struct sk_buff *from)
{
to->pkt_type = from->pkt_type;
to->priority = from->priority;
to->protocol = from->protocol;
skb_dst_drop(to);
skb_dst_set(to, dst_clone(skb_dst(from)));
to->dev = from->dev;
to->mark = from->mark;
#ifdef CONFIG_NET_SCHED
to->tc_index = from->tc_index;
#endif
nf_copy(to, from);
skb_copy_secmark(to, from);
}
int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
int (*output)(struct net *, struct sock *, struct sk_buff *))
{
struct sk_buff *frag;
struct rt6_info *rt = (struct rt6_info *)skb_dst(skb);
struct ipv6_pinfo *np = skb->sk && !dev_recursion_level() ?
inet6_sk(skb->sk) : NULL;
struct ipv6hdr *tmp_hdr;
struct frag_hdr *fh;
unsigned int mtu, hlen, left, len;
int hroom, troom;
__be32 frag_id;
int ptr, offset = 0, err = 0;
u8 *prevhdr, nexthdr = 0;
err = ip6_find_1stfragopt(skb, &prevhdr);
if (err < 0)
ipv6: Prevent overrun when parsing v6 header options The KASAN warning repoted below was discovered with a syzkaller program. The reproducer is basically: int s = socket(AF_INET6, SOCK_RAW, NEXTHDR_HOP); send(s, &one_byte_of_data, 1, MSG_MORE); send(s, &more_than_mtu_bytes_data, 2000, 0); The socket() call sets the nexthdr field of the v6 header to NEXTHDR_HOP, the first send call primes the payload with a non zero byte of data, and the second send call triggers the fragmentation path. The fragmentation code tries to parse the header options in order to figure out where to insert the fragment option. Since nexthdr points to an invalid option, the calculation of the size of the network header can made to be much larger than the linear section of the skb and data is read outside of it. This fix makes ip6_find_1stfrag return an error if it detects running out-of-bounds. [ 42.361487] ================================================================== [ 42.364412] BUG: KASAN: slab-out-of-bounds in ip6_fragment+0x11c8/0x3730 [ 42.365471] Read of size 840 at addr ffff88000969e798 by task ip6_fragment-oo/3789 [ 42.366469] [ 42.366696] CPU: 1 PID: 3789 Comm: ip6_fragment-oo Not tainted 4.11.0+ #41 [ 42.367628] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.1-1ubuntu1 04/01/2014 [ 42.368824] Call Trace: [ 42.369183] dump_stack+0xb3/0x10b [ 42.369664] print_address_description+0x73/0x290 [ 42.370325] kasan_report+0x252/0x370 [ 42.370839] ? ip6_fragment+0x11c8/0x3730 [ 42.371396] check_memory_region+0x13c/0x1a0 [ 42.371978] memcpy+0x23/0x50 [ 42.372395] ip6_fragment+0x11c8/0x3730 [ 42.372920] ? nf_ct_expect_unregister_notifier+0x110/0x110 [ 42.373681] ? ip6_copy_metadata+0x7f0/0x7f0 [ 42.374263] ? ip6_forward+0x2e30/0x2e30 [ 42.374803] ip6_finish_output+0x584/0x990 [ 42.375350] ip6_output+0x1b7/0x690 [ 42.375836] ? ip6_finish_output+0x990/0x990 [ 42.376411] ? ip6_fragment+0x3730/0x3730 [ 42.376968] ip6_local_out+0x95/0x160 [ 42.377471] ip6_send_skb+0xa1/0x330 [ 42.377969] ip6_push_pending_frames+0xb3/0xe0 [ 42.378589] rawv6_sendmsg+0x2051/0x2db0 [ 42.379129] ? rawv6_bind+0x8b0/0x8b0 [ 42.379633] ? _copy_from_user+0x84/0xe0 [ 42.380193] ? debug_check_no_locks_freed+0x290/0x290 [ 42.380878] ? ___sys_sendmsg+0x162/0x930 [ 42.381427] ? rcu_read_lock_sched_held+0xa3/0x120 [ 42.382074] ? sock_has_perm+0x1f6/0x290 [ 42.382614] ? ___sys_sendmsg+0x167/0x930 [ 42.383173] ? lock_downgrade+0x660/0x660 [ 42.383727] inet_sendmsg+0x123/0x500 [ 42.384226] ? inet_sendmsg+0x123/0x500 [ 42.384748] ? inet_recvmsg+0x540/0x540 [ 42.385263] sock_sendmsg+0xca/0x110 [ 42.385758] SYSC_sendto+0x217/0x380 [ 42.386249] ? SYSC_connect+0x310/0x310 [ 42.386783] ? __might_fault+0x110/0x1d0 [ 42.387324] ? lock_downgrade+0x660/0x660 [ 42.387880] ? __fget_light+0xa1/0x1f0 [ 42.388403] ? __fdget+0x18/0x20 [ 42.388851] ? sock_common_setsockopt+0x95/0xd0 [ 42.389472] ? SyS_setsockopt+0x17f/0x260 [ 42.390021] ? entry_SYSCALL_64_fastpath+0x5/0xbe [ 42.390650] SyS_sendto+0x40/0x50 [ 42.391103] entry_SYSCALL_64_fastpath+0x1f/0xbe [ 42.391731] RIP: 0033:0x7fbbb711e383 [ 42.392217] RSP: 002b:00007ffff4d34f28 EFLAGS: 00000246 ORIG_RAX: 000000000000002c [ 42.393235] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fbbb711e383 [ 42.394195] RDX: 0000000000001000 RSI: 00007ffff4d34f60 RDI: 0000000000000003 [ 42.395145] RBP: 0000000000000046 R08: 00007ffff4d34f40 R09: 0000000000000018 [ 42.396056] R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000400aad [ 42.396598] R13: 0000000000000066 R14: 00007ffff4d34ee0 R15: 00007fbbb717af00 [ 42.397257] [ 42.397411] Allocated by task 3789: [ 42.397702] save_stack_trace+0x16/0x20 [ 42.398005] save_stack+0x46/0xd0 [ 42.398267] kasan_kmalloc+0xad/0xe0 [ 42.398548] kasan_slab_alloc+0x12/0x20 [ 42.398848] __kmalloc_node_track_caller+0xcb/0x380 [ 42.399224] __kmalloc_reserve.isra.32+0x41/0xe0 [ 42.399654] __alloc_skb+0xf8/0x580 [ 42.400003] sock_wmalloc+0xab/0xf0 [ 42.400346] __ip6_append_data.isra.41+0x2472/0x33d0 [ 42.400813] ip6_append_data+0x1a8/0x2f0 [ 42.401122] rawv6_sendmsg+0x11ee/0x2db0 [ 42.401505] inet_sendmsg+0x123/0x500 [ 42.401860] sock_sendmsg+0xca/0x110 [ 42.402209] ___sys_sendmsg+0x7cb/0x930 [ 42.402582] __sys_sendmsg+0xd9/0x190 [ 42.402941] SyS_sendmsg+0x2d/0x50 [ 42.403273] entry_SYSCALL_64_fastpath+0x1f/0xbe [ 42.403718] [ 42.403871] Freed by task 1794: [ 42.404146] save_stack_trace+0x16/0x20 [ 42.404515] save_stack+0x46/0xd0 [ 42.404827] kasan_slab_free+0x72/0xc0 [ 42.405167] kfree+0xe8/0x2b0 [ 42.405462] skb_free_head+0x74/0xb0 [ 42.405806] skb_release_data+0x30e/0x3a0 [ 42.406198] skb_release_all+0x4a/0x60 [ 42.406563] consume_skb+0x113/0x2e0 [ 42.406910] skb_free_datagram+0x1a/0xe0 [ 42.407288] netlink_recvmsg+0x60d/0xe40 [ 42.407667] sock_recvmsg+0xd7/0x110 [ 42.408022] ___sys_recvmsg+0x25c/0x580 [ 42.408395] __sys_recvmsg+0xd6/0x190 [ 42.408753] SyS_recvmsg+0x2d/0x50 [ 42.409086] entry_SYSCALL_64_fastpath+0x1f/0xbe [ 42.409513] [ 42.409665] The buggy address belongs to the object at ffff88000969e780 [ 42.409665] which belongs to the cache kmalloc-512 of size 512 [ 42.410846] The buggy address is located 24 bytes inside of [ 42.410846] 512-byte region [ffff88000969e780, ffff88000969e980) [ 42.411941] The buggy address belongs to the page: [ 42.412405] page:ffffea000025a780 count:1 mapcount:0 mapping: (null) index:0x0 compound_mapcount: 0 [ 42.413298] flags: 0x100000000008100(slab|head) [ 42.413729] raw: 0100000000008100 0000000000000000 0000000000000000 00000001800c000c [ 42.414387] raw: ffffea00002a9500 0000000900000007 ffff88000c401280 0000000000000000 [ 42.415074] page dumped because: kasan: bad access detected [ 42.415604] [ 42.415757] Memory state around the buggy address: [ 42.416222] ffff88000969e880: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 [ 42.416904] ffff88000969e900: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 [ 42.417591] >ffff88000969e980: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc [ 42.418273] ^ [ 42.418588] ffff88000969ea00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 42.419273] ffff88000969ea80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 42.419882] ================================================================== Reported-by: Andrey Konovalov <andreyknvl@google.com> Signed-off-by: Craig Gallek <kraig@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-05-16 18:36:23 +00:00
goto fail;
hlen = err;
nexthdr = *prevhdr;
mtu = ip6_skb_dst_mtu(skb);
/* We must not fragment if the socket is set to force MTU discovery
* or if the skb it not generated by a local socket.
*/
if (unlikely(!skb->ignore_df && skb->len > mtu))
goto fail_toobig;
if (IP6CB(skb)->frag_max_size) {
if (IP6CB(skb)->frag_max_size > mtu)
goto fail_toobig;
/* don't send fragments larger than what we received */
mtu = IP6CB(skb)->frag_max_size;
if (mtu < IPV6_MIN_MTU)
mtu = IPV6_MIN_MTU;
}
if (np && np->frag_size < mtu) {
if (np->frag_size)
mtu = np->frag_size;
}
if (mtu < hlen + sizeof(struct frag_hdr) + 8)
goto fail_toobig;
mtu -= hlen + sizeof(struct frag_hdr);
frag_id = ipv6_select_ident(net, &ipv6_hdr(skb)->daddr,
&ipv6_hdr(skb)->saddr);
if (skb->ip_summed == CHECKSUM_PARTIAL &&
(err = skb_checksum_help(skb)))
goto fail;
ipv6: ip6_fragment: fix headroom tests and skb leak David Woodhouse reports skb_under_panic when we try to push ethernet header to fragmented ipv6 skbs: skbuff: skb_under_panic: text:c1277f1e len:1294 put:14 head:dec98000 data:dec97ffc tail:0xdec9850a end:0xdec98f40 dev:br-lan [..] ip6_finish_output2+0x196/0x4da David further debugged this: [..] offending fragments were arriving here with skb_headroom(skb)==10. Which is reasonable, being the Solos ADSL card's header of 8 bytes followed by 2 bytes of PPP frame type. The problem is that if netfilter ipv6 defragmentation is used, skb_cow() in ip6_forward will only see reassembled skb. Therefore, headroom is overestimated by 8 bytes (we pulled fragment header) and we don't check the skbs in the frag_list either. We can't do these checks in netfilter defrag since outdev isn't known yet. Furthermore, existing tests in ip6_fragment did not consider the fragment or ipv6 header size when checking headroom of the fraglist skbs. While at it, also fix a skb leak on memory allocation -- ip6_fragment must consume the skb. I tested this e1000 driver hacked to not allocate additional headroom (we end up in slowpath, since LL_RESERVED_SPACE is 16). If 2 bytes of headroom are allocated, fastpath is taken (14 byte ethernet header was pulled, so 16 byte headroom available in all fragments). Reported-by: David Woodhouse <dwmw2@infradead.org> Diagnosed-by: David Woodhouse <dwmw2@infradead.org> Signed-off-by: Florian Westphal <fw@strlen.de> Tested-by: David Woodhouse <David.Woodhouse@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-16 15:26:14 +00:00
hroom = LL_RESERVED_SPACE(rt->dst.dev);
if (skb_has_frag_list(skb)) {
unsigned int first_len = skb_pagelen(skb);
struct sk_buff *frag2;
if (first_len - hlen > mtu ||
((first_len - hlen) & 7) ||
ipv6: ip6_fragment: fix headroom tests and skb leak David Woodhouse reports skb_under_panic when we try to push ethernet header to fragmented ipv6 skbs: skbuff: skb_under_panic: text:c1277f1e len:1294 put:14 head:dec98000 data:dec97ffc tail:0xdec9850a end:0xdec98f40 dev:br-lan [..] ip6_finish_output2+0x196/0x4da David further debugged this: [..] offending fragments were arriving here with skb_headroom(skb)==10. Which is reasonable, being the Solos ADSL card's header of 8 bytes followed by 2 bytes of PPP frame type. The problem is that if netfilter ipv6 defragmentation is used, skb_cow() in ip6_forward will only see reassembled skb. Therefore, headroom is overestimated by 8 bytes (we pulled fragment header) and we don't check the skbs in the frag_list either. We can't do these checks in netfilter defrag since outdev isn't known yet. Furthermore, existing tests in ip6_fragment did not consider the fragment or ipv6 header size when checking headroom of the fraglist skbs. While at it, also fix a skb leak on memory allocation -- ip6_fragment must consume the skb. I tested this e1000 driver hacked to not allocate additional headroom (we end up in slowpath, since LL_RESERVED_SPACE is 16). If 2 bytes of headroom are allocated, fastpath is taken (14 byte ethernet header was pulled, so 16 byte headroom available in all fragments). Reported-by: David Woodhouse <dwmw2@infradead.org> Diagnosed-by: David Woodhouse <dwmw2@infradead.org> Signed-off-by: Florian Westphal <fw@strlen.de> Tested-by: David Woodhouse <David.Woodhouse@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-16 15:26:14 +00:00
skb_cloned(skb) ||
skb_headroom(skb) < (hroom + sizeof(struct frag_hdr)))
goto slow_path;
skb_walk_frags(skb, frag) {
/* Correct geometry. */
if (frag->len > mtu ||
((frag->len & 7) && frag->next) ||
ipv6: ip6_fragment: fix headroom tests and skb leak David Woodhouse reports skb_under_panic when we try to push ethernet header to fragmented ipv6 skbs: skbuff: skb_under_panic: text:c1277f1e len:1294 put:14 head:dec98000 data:dec97ffc tail:0xdec9850a end:0xdec98f40 dev:br-lan [..] ip6_finish_output2+0x196/0x4da David further debugged this: [..] offending fragments were arriving here with skb_headroom(skb)==10. Which is reasonable, being the Solos ADSL card's header of 8 bytes followed by 2 bytes of PPP frame type. The problem is that if netfilter ipv6 defragmentation is used, skb_cow() in ip6_forward will only see reassembled skb. Therefore, headroom is overestimated by 8 bytes (we pulled fragment header) and we don't check the skbs in the frag_list either. We can't do these checks in netfilter defrag since outdev isn't known yet. Furthermore, existing tests in ip6_fragment did not consider the fragment or ipv6 header size when checking headroom of the fraglist skbs. While at it, also fix a skb leak on memory allocation -- ip6_fragment must consume the skb. I tested this e1000 driver hacked to not allocate additional headroom (we end up in slowpath, since LL_RESERVED_SPACE is 16). If 2 bytes of headroom are allocated, fastpath is taken (14 byte ethernet header was pulled, so 16 byte headroom available in all fragments). Reported-by: David Woodhouse <dwmw2@infradead.org> Diagnosed-by: David Woodhouse <dwmw2@infradead.org> Signed-off-by: Florian Westphal <fw@strlen.de> Tested-by: David Woodhouse <David.Woodhouse@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-16 15:26:14 +00:00
skb_headroom(frag) < (hlen + hroom + sizeof(struct frag_hdr)))
goto slow_path_clean;
/* Partially cloned skb? */
if (skb_shared(frag))
goto slow_path_clean;
BUG_ON(frag->sk);
if (skb->sk) {
frag->sk = skb->sk;
frag->destructor = sock_wfree;
}
skb->truesize -= frag->truesize;
}
err = 0;
offset = 0;
/* BUILD HEADER */
*prevhdr = NEXTHDR_FRAGMENT;
tmp_hdr = kmemdup(skb_network_header(skb), hlen, GFP_ATOMIC);
if (!tmp_hdr) {
ipv6: ip6_fragment: fix headroom tests and skb leak David Woodhouse reports skb_under_panic when we try to push ethernet header to fragmented ipv6 skbs: skbuff: skb_under_panic: text:c1277f1e len:1294 put:14 head:dec98000 data:dec97ffc tail:0xdec9850a end:0xdec98f40 dev:br-lan [..] ip6_finish_output2+0x196/0x4da David further debugged this: [..] offending fragments were arriving here with skb_headroom(skb)==10. Which is reasonable, being the Solos ADSL card's header of 8 bytes followed by 2 bytes of PPP frame type. The problem is that if netfilter ipv6 defragmentation is used, skb_cow() in ip6_forward will only see reassembled skb. Therefore, headroom is overestimated by 8 bytes (we pulled fragment header) and we don't check the skbs in the frag_list either. We can't do these checks in netfilter defrag since outdev isn't known yet. Furthermore, existing tests in ip6_fragment did not consider the fragment or ipv6 header size when checking headroom of the fraglist skbs. While at it, also fix a skb leak on memory allocation -- ip6_fragment must consume the skb. I tested this e1000 driver hacked to not allocate additional headroom (we end up in slowpath, since LL_RESERVED_SPACE is 16). If 2 bytes of headroom are allocated, fastpath is taken (14 byte ethernet header was pulled, so 16 byte headroom available in all fragments). Reported-by: David Woodhouse <dwmw2@infradead.org> Diagnosed-by: David Woodhouse <dwmw2@infradead.org> Signed-off-by: Florian Westphal <fw@strlen.de> Tested-by: David Woodhouse <David.Woodhouse@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-16 15:26:14 +00:00
err = -ENOMEM;
goto fail;
}
ipv6: ip6_fragment: fix headroom tests and skb leak David Woodhouse reports skb_under_panic when we try to push ethernet header to fragmented ipv6 skbs: skbuff: skb_under_panic: text:c1277f1e len:1294 put:14 head:dec98000 data:dec97ffc tail:0xdec9850a end:0xdec98f40 dev:br-lan [..] ip6_finish_output2+0x196/0x4da David further debugged this: [..] offending fragments were arriving here with skb_headroom(skb)==10. Which is reasonable, being the Solos ADSL card's header of 8 bytes followed by 2 bytes of PPP frame type. The problem is that if netfilter ipv6 defragmentation is used, skb_cow() in ip6_forward will only see reassembled skb. Therefore, headroom is overestimated by 8 bytes (we pulled fragment header) and we don't check the skbs in the frag_list either. We can't do these checks in netfilter defrag since outdev isn't known yet. Furthermore, existing tests in ip6_fragment did not consider the fragment or ipv6 header size when checking headroom of the fraglist skbs. While at it, also fix a skb leak on memory allocation -- ip6_fragment must consume the skb. I tested this e1000 driver hacked to not allocate additional headroom (we end up in slowpath, since LL_RESERVED_SPACE is 16). If 2 bytes of headroom are allocated, fastpath is taken (14 byte ethernet header was pulled, so 16 byte headroom available in all fragments). Reported-by: David Woodhouse <dwmw2@infradead.org> Diagnosed-by: David Woodhouse <dwmw2@infradead.org> Signed-off-by: Florian Westphal <fw@strlen.de> Tested-by: David Woodhouse <David.Woodhouse@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-16 15:26:14 +00:00
frag = skb_shinfo(skb)->frag_list;
skb_frag_list_init(skb);
__skb_pull(skb, hlen);
fh = __skb_push(skb, sizeof(struct frag_hdr));
__skb_push(skb, hlen);
skb_reset_network_header(skb);
memcpy(skb_network_header(skb), tmp_hdr, hlen);
fh->nexthdr = nexthdr;
fh->reserved = 0;
fh->frag_off = htons(IP6_MF);
fh->identification = frag_id;
first_len = skb_pagelen(skb);
skb->data_len = first_len - skb_headlen(skb);
skb->len = first_len;
ipv6_hdr(skb)->payload_len = htons(first_len -
sizeof(struct ipv6hdr));
for (;;) {
/* Prepare header of the next frame,
* before previous one went down. */
if (frag) {
frag->ip_summed = CHECKSUM_NONE;
skb_reset_transport_header(frag);
fh = __skb_push(frag, sizeof(struct frag_hdr));
__skb_push(frag, hlen);
skb_reset_network_header(frag);
memcpy(skb_network_header(frag), tmp_hdr,
hlen);
offset += skb->len - hlen - sizeof(struct frag_hdr);
fh->nexthdr = nexthdr;
fh->reserved = 0;
fh->frag_off = htons(offset);
if (frag->next)
fh->frag_off |= htons(IP6_MF);
fh->identification = frag_id;
ipv6_hdr(frag)->payload_len =
htons(frag->len -
sizeof(struct ipv6hdr));
ip6_copy_metadata(frag, skb);
}
err = output(net, sk, skb);
if (!err)
IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
IPSTATS_MIB_FRAGCREATES);
[IPV6]: SNMPv2 "ipv6IfStatsOutFragCreates" counter error When I tested linux kernel 2.6.71.7 about statistics "ipv6IfStatsOutFragCreates", and found that it couldn't increase correctly. The criteria is RFC 2465: ipv6IfStatsOutFragCreates OBJECT-TYPE SYNTAX Counter32 MAX-ACCESS read-only STATUS current DESCRIPTION "The number of output datagram fragments that have been generated as a result of fragmentation at this output interface." ::= { ipv6IfStatsEntry 15 } I think there are two issues in Linux kernel. 1st: RFC2465 specifies the counter is "The number of output datagram fragments...". I think increasing this counter after output a fragment successfully is better. And it should not be increased even though a fragment is created but failed to output. 2nd: If we send a big ICMP/ICMPv6 echo request to a host, and receive ICMP/ICMPv6 echo reply consisted of some fragments. As we know that in Linux kernel first fragmentation occurs in ICMP layer(maybe saying transport layer is better), but this is not the "real" fragmentation,just do some "pre-fragment" -- allocate space for date, and form a frag_list, etc. The "real" fragmentation happens in IP layer -- set offset and MF flag and so on. So I think in "fast path" for ip_fragment/ip6_fragment, if we send a fragment which "pre-fragment" by upper layer we should also increase "ipv6IfStatsOutFragCreates". Signed-off-by: Wei Dong <weid@nanjing-fnst.com> Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-02 20:41:21 +00:00
if (err || !frag)
break;
skb = frag;
frag = skb->next;
skb->next = NULL;
}
kfree(tmp_hdr);
if (err == 0) {
IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
IPSTATS_MIB_FRAGOKS);
return 0;
}
kfree_skb_list(frag);
IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
IPSTATS_MIB_FRAGFAILS);
return err;
slow_path_clean:
skb_walk_frags(skb, frag2) {
if (frag2 == frag)
break;
frag2->sk = NULL;
frag2->destructor = NULL;
skb->truesize += frag2->truesize;
}
}
slow_path:
left = skb->len - hlen; /* Space per frame */
ptr = hlen; /* Where to start from */
/*
* Fragment the datagram.
*/
troom = rt->dst.dev->needed_tailroom;
/*
* Keep copying data until we run out.
*/
while (left > 0) {
u8 *fragnexthdr_offset;
len = left;
/* IF: it doesn't fit, use 'mtu' - the data space left */
if (len > mtu)
len = mtu;
/* IF: we are not sending up to and including the packet end
then align the next start on an eight byte boundary */
if (len < left) {
len &= ~7;
}
/* Allocate buffer */
frag = alloc_skb(len + hlen + sizeof(struct frag_hdr) +
hroom + troom, GFP_ATOMIC);
if (!frag) {
err = -ENOMEM;
goto fail;
}
/*
* Set up data on packet
*/
ip6_copy_metadata(frag, skb);
skb_reserve(frag, hroom);
skb_put(frag, len + hlen + sizeof(struct frag_hdr));
skb_reset_network_header(frag);
fh = (struct frag_hdr *)(skb_network_header(frag) + hlen);
frag->transport_header = (frag->network_header + hlen +
sizeof(struct frag_hdr));
/*
* Charge the memory for the fragment to any owner
* it might possess
*/
if (skb->sk)
skb_set_owner_w(frag, skb->sk);
/*
* Copy the packet header into the new buffer.
*/
skb_copy_from_linear_data(skb, skb_network_header(frag), hlen);
fragnexthdr_offset = skb_network_header(frag);
fragnexthdr_offset += prevhdr - skb_network_header(skb);
*fragnexthdr_offset = NEXTHDR_FRAGMENT;
/*
* Build fragment header.
*/
fh->nexthdr = nexthdr;
fh->reserved = 0;
fh->identification = frag_id;
/*
* Copy a block of the IP datagram.
*/
BUG_ON(skb_copy_bits(skb, ptr, skb_transport_header(frag),
len));
left -= len;
fh->frag_off = htons(offset);
if (left > 0)
fh->frag_off |= htons(IP6_MF);
ipv6_hdr(frag)->payload_len = htons(frag->len -
sizeof(struct ipv6hdr));
ptr += len;
offset += len;
/*
* Put this fragment into the sending queue.
*/
err = output(net, sk, frag);
if (err)
goto fail;
[IPV6]: SNMPv2 "ipv6IfStatsOutFragCreates" counter error When I tested linux kernel 2.6.71.7 about statistics "ipv6IfStatsOutFragCreates", and found that it couldn't increase correctly. The criteria is RFC 2465: ipv6IfStatsOutFragCreates OBJECT-TYPE SYNTAX Counter32 MAX-ACCESS read-only STATUS current DESCRIPTION "The number of output datagram fragments that have been generated as a result of fragmentation at this output interface." ::= { ipv6IfStatsEntry 15 } I think there are two issues in Linux kernel. 1st: RFC2465 specifies the counter is "The number of output datagram fragments...". I think increasing this counter after output a fragment successfully is better. And it should not be increased even though a fragment is created but failed to output. 2nd: If we send a big ICMP/ICMPv6 echo request to a host, and receive ICMP/ICMPv6 echo reply consisted of some fragments. As we know that in Linux kernel first fragmentation occurs in ICMP layer(maybe saying transport layer is better), but this is not the "real" fragmentation,just do some "pre-fragment" -- allocate space for date, and form a frag_list, etc. The "real" fragmentation happens in IP layer -- set offset and MF flag and so on. So I think in "fast path" for ip_fragment/ip6_fragment, if we send a fragment which "pre-fragment" by upper layer we should also increase "ipv6IfStatsOutFragCreates". Signed-off-by: Wei Dong <weid@nanjing-fnst.com> Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-02 20:41:21 +00:00
IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
IPSTATS_MIB_FRAGCREATES);
}
IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
IPSTATS_MIB_FRAGOKS);
consume_skb(skb);
return err;
fail_toobig:
if (skb->sk && dst_allfrag(skb_dst(skb)))
sk_nocaps_add(skb->sk, NETIF_F_GSO_MASK);
icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
err = -EMSGSIZE;
fail:
IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
IPSTATS_MIB_FRAGFAILS);
kfree_skb(skb);
return err;
}
static inline int ip6_rt_check(const struct rt6key *rt_key,
const struct in6_addr *fl_addr,
const struct in6_addr *addr_cache)
{
return (rt_key->plen != 128 || !ipv6_addr_equal(fl_addr, &rt_key->addr)) &&
(!addr_cache || !ipv6_addr_equal(fl_addr, addr_cache));
}
static struct dst_entry *ip6_sk_dst_check(struct sock *sk,
struct dst_entry *dst,
const struct flowi6 *fl6)
{
struct ipv6_pinfo *np = inet6_sk(sk);
struct rt6_info *rt;
if (!dst)
goto out;
if (dst->ops->family != AF_INET6) {
dst_release(dst);
return NULL;
}
rt = (struct rt6_info *)dst;
/* Yes, checking route validity in not connected
* case is not very simple. Take into account,
* that we do not support routing by source, TOS,
* and MSG_DONTROUTE --ANK (980726)
*
* 1. ip6_rt_check(): If route was host route,
* check that cached destination is current.
* If it is network route, we still may
* check its validity using saved pointer
* to the last used address: daddr_cache.
* We do not want to save whole address now,
* (because main consumer of this service
* is tcp, which has not this problem),
* so that the last trick works only on connected
* sockets.
* 2. oif also should be the same.
*/
if (ip6_rt_check(&rt->rt6i_dst, &fl6->daddr, np->daddr_cache) ||
#ifdef CONFIG_IPV6_SUBTREES
ip6_rt_check(&rt->rt6i_src, &fl6->saddr, np->saddr_cache) ||
#endif
(!(fl6->flowi6_flags & FLOWI_FLAG_SKIP_NH_OIF) &&
(fl6->flowi6_oif && fl6->flowi6_oif != dst->dev->ifindex))) {
dst_release(dst);
dst = NULL;
}
out:
return dst;
}
static int ip6_dst_lookup_tail(struct net *net, const struct sock *sk,
struct dst_entry **dst, struct flowi6 *fl6)
{
#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
struct neighbour *n;
struct rt6_info *rt;
#endif
int err;
int flags = 0;
/* The correct way to handle this would be to do
* ip6_route_get_saddr, and then ip6_route_output; however,
* the route-specific preferred source forces the
* ip6_route_output call _before_ ip6_route_get_saddr.
*
* In source specific routing (no src=any default route),
* ip6_route_output will fail given src=any saddr, though, so
* that's why we try it again later.
*/
if (ipv6_addr_any(&fl6->saddr) && (!*dst || !(*dst)->error)) {
struct rt6_info *rt;
bool had_dst = *dst != NULL;
if (!had_dst)
*dst = ip6_route_output(net, sk, fl6);
rt = (*dst)->error ? NULL : (struct rt6_info *)*dst;
err = ip6_route_get_saddr(net, rt, &fl6->daddr,
sk ? inet6_sk(sk)->srcprefs : 0,
&fl6->saddr);
if (err)
goto out_err_release;
/* If we had an erroneous initial result, pretend it
* never existed and let the SA-enabled version take
* over.
*/
if (!had_dst && (*dst)->error) {
dst_release(*dst);
*dst = NULL;
}
if (fl6->flowi6_oif)
flags |= RT6_LOOKUP_F_IFACE;
}
if (!*dst)
*dst = ip6_route_output_flags(net, sk, fl6, flags);
err = (*dst)->error;
if (err)
goto out_err_release;
#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
/*
* Here if the dst entry we've looked up
* has a neighbour entry that is in the INCOMPLETE
* state and the src address from the flow is
* marked as OPTIMISTIC, we release the found
* dst entry and replace it instead with the
* dst entry of the nexthop router
*/
rt = (struct rt6_info *) *dst;
rcu_read_lock_bh();
n = __ipv6_neigh_lookup_noref(rt->dst.dev,
rt6_nexthop(rt, &fl6->daddr));
err = n && !(n->nud_state & NUD_VALID) ? -EINVAL : 0;
rcu_read_unlock_bh();
if (err) {
struct inet6_ifaddr *ifp;
struct flowi6 fl_gw6;
int redirect;
ifp = ipv6_get_ifaddr(net, &fl6->saddr,
(*dst)->dev, 1);
redirect = (ifp && ifp->flags & IFA_F_OPTIMISTIC);
if (ifp)
in6_ifa_put(ifp);
if (redirect) {
/*
* We need to get the dst entry for the
* default router instead
*/
dst_release(*dst);
memcpy(&fl_gw6, fl6, sizeof(struct flowi6));
memset(&fl_gw6.daddr, 0, sizeof(struct in6_addr));
*dst = ip6_route_output(net, sk, &fl_gw6);
err = (*dst)->error;
if (err)
goto out_err_release;
}
}
#endif
if (ipv6_addr_v4mapped(&fl6->saddr) &&
!(ipv6_addr_v4mapped(&fl6->daddr) || ipv6_addr_any(&fl6->daddr))) {
err = -EAFNOSUPPORT;
goto out_err_release;
}
return 0;
out_err_release:
dst_release(*dst);
*dst = NULL;
if (err == -ENETUNREACH)
IP6_INC_STATS(net, NULL, IPSTATS_MIB_OUTNOROUTES);
return err;
}
/**
* ip6_dst_lookup - perform route lookup on flow
* @sk: socket which provides route info
* @dst: pointer to dst_entry * for result
* @fl6: flow to lookup
*
* This function performs a route lookup on the given flow.
*
* It returns zero on success, or a standard errno code on error.
*/
int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst,
struct flowi6 *fl6)
{
*dst = NULL;
return ip6_dst_lookup_tail(net, sk, dst, fl6);
}
EXPORT_SYMBOL_GPL(ip6_dst_lookup);
/**
* ip6_dst_lookup_flow - perform route lookup on flow with ipsec
* @sk: socket which provides route info
* @fl6: flow to lookup
* @final_dst: final destination address for ipsec lookup
*
* This function performs a route lookup on the given flow.
*
* It returns a valid dst pointer on success, or a pointer encoded
* error code.
*/
struct dst_entry *ip6_dst_lookup_flow(const struct sock *sk, struct flowi6 *fl6,
const struct in6_addr *final_dst)
{
struct dst_entry *dst = NULL;
int err;
err = ip6_dst_lookup_tail(sock_net(sk), sk, &dst, fl6);
if (err)
return ERR_PTR(err);
if (final_dst)
fl6->daddr = *final_dst;
return xfrm_lookup_route(sock_net(sk), dst, flowi6_to_flowi(fl6), sk, 0);
}
EXPORT_SYMBOL_GPL(ip6_dst_lookup_flow);
/**
* ip6_sk_dst_lookup_flow - perform socket cached route lookup on flow
* @sk: socket which provides the dst cache and route info
* @fl6: flow to lookup
* @final_dst: final destination address for ipsec lookup
*
* This function performs a route lookup on the given flow with the
* possibility of using the cached route in the socket if it is valid.
* It will take the socket dst lock when operating on the dst cache.
* As a result, this function can only be used in process context.
*
* It returns a valid dst pointer on success, or a pointer encoded
* error code.
*/
struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
const struct in6_addr *final_dst)
{
struct dst_entry *dst = sk_dst_check(sk, inet6_sk(sk)->dst_cookie);
dst = ip6_sk_dst_check(sk, dst, fl6);
ipv6: Skip XFRM lookup if dst_entry in socket cache is valid At present we perform an xfrm_lookup() for each UDPv6 message we send. The lookup involves querying the flow cache (flow_cache_lookup) and, in case of a cache miss, creating an XFRM bundle. If we miss the flow cache, we can end up creating a new bundle and deriving the path MTU (xfrm_init_pmtu) from on an already transformed dst_entry, which we pass from the socket cache (sk->sk_dst_cache) down to xfrm_lookup(). This can happen only if we're caching the dst_entry in the socket, that is when we're using a connected UDP socket. To put it another way, the path MTU shrinks each time we miss the flow cache, which later on leads to incorrectly fragmented payload. It can be observed with ESPv6 in transport mode: 1) Set up a transformation and lower the MTU to trigger fragmentation # ip xfrm policy add dir out src ::1 dst ::1 \ tmpl src ::1 dst ::1 proto esp spi 1 # ip xfrm state add src ::1 dst ::1 \ proto esp spi 1 enc 'aes' 0x0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b # ip link set dev lo mtu 1500 2) Monitor the packet flow and set up an UDP sink # tcpdump -ni lo -ttt & # socat udp6-listen:12345,fork /dev/null & 3) Send a datagram that needs fragmentation with a connected socket # perl -e 'print "@" x 1470 | socat - udp6:[::1]:12345 2016/06/07 18:52:52 socat[724] E read(3, 0x555bb3d5ba00, 8192): Protocol error 00:00:00.000000 IP6 ::1 > ::1: frag (0|1448) ESP(spi=0x00000001,seq=0x2), length 1448 00:00:00.000014 IP6 ::1 > ::1: frag (1448|32) 00:00:00.000050 IP6 ::1 > ::1: ESP(spi=0x00000001,seq=0x3), length 1272 (^ ICMPv6 Parameter Problem) 00:00:00.000022 IP6 ::1 > ::1: ESP(spi=0x00000001,seq=0x5), length 136 4) Compare it to a non-connected socket # perl -e 'print "@" x 1500' | socat - udp6-sendto:[::1]:12345 00:00:40.535488 IP6 ::1 > ::1: frag (0|1448) ESP(spi=0x00000001,seq=0x6), length 1448 00:00:00.000010 IP6 ::1 > ::1: frag (1448|64) What happens in step (3) is: 1) when connecting the socket in __ip6_datagram_connect(), we perform an XFRM lookup, miss the flow cache, create an XFRM bundle, and cache the destination, 2) afterwards, when sending the datagram, we perform an XFRM lookup, again, miss the flow cache (due to mismatch of flowi6_iif and flowi6_oif, which is an issue of its own), and recreate an XFRM bundle based on the cached (and already transformed) destination. To prevent the recreation of an XFRM bundle, avoid an XFRM lookup altogether whenever we already have a destination entry cached in the socket. This prevents the path MTU shrinkage and brings us on par with UDPv4. The fix also benefits connected PINGv6 sockets, another user of ip6_sk_dst_lookup_flow(), who also suffer messages being transformed twice. Joint work with Hannes Frederic Sowa. Reported-by: Jan Tluka <jtluka@redhat.com> Signed-off-by: Jakub Sitnicki <jkbs@redhat.com> Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-06-08 13:13:34 +00:00
if (!dst)
dst = ip6_dst_lookup_flow(sk, fl6, final_dst);
ipv6: Skip XFRM lookup if dst_entry in socket cache is valid At present we perform an xfrm_lookup() for each UDPv6 message we send. The lookup involves querying the flow cache (flow_cache_lookup) and, in case of a cache miss, creating an XFRM bundle. If we miss the flow cache, we can end up creating a new bundle and deriving the path MTU (xfrm_init_pmtu) from on an already transformed dst_entry, which we pass from the socket cache (sk->sk_dst_cache) down to xfrm_lookup(). This can happen only if we're caching the dst_entry in the socket, that is when we're using a connected UDP socket. To put it another way, the path MTU shrinks each time we miss the flow cache, which later on leads to incorrectly fragmented payload. It can be observed with ESPv6 in transport mode: 1) Set up a transformation and lower the MTU to trigger fragmentation # ip xfrm policy add dir out src ::1 dst ::1 \ tmpl src ::1 dst ::1 proto esp spi 1 # ip xfrm state add src ::1 dst ::1 \ proto esp spi 1 enc 'aes' 0x0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b # ip link set dev lo mtu 1500 2) Monitor the packet flow and set up an UDP sink # tcpdump -ni lo -ttt & # socat udp6-listen:12345,fork /dev/null & 3) Send a datagram that needs fragmentation with a connected socket # perl -e 'print "@" x 1470 | socat - udp6:[::1]:12345 2016/06/07 18:52:52 socat[724] E read(3, 0x555bb3d5ba00, 8192): Protocol error 00:00:00.000000 IP6 ::1 > ::1: frag (0|1448) ESP(spi=0x00000001,seq=0x2), length 1448 00:00:00.000014 IP6 ::1 > ::1: frag (1448|32) 00:00:00.000050 IP6 ::1 > ::1: ESP(spi=0x00000001,seq=0x3), length 1272 (^ ICMPv6 Parameter Problem) 00:00:00.000022 IP6 ::1 > ::1: ESP(spi=0x00000001,seq=0x5), length 136 4) Compare it to a non-connected socket # perl -e 'print "@" x 1500' | socat - udp6-sendto:[::1]:12345 00:00:40.535488 IP6 ::1 > ::1: frag (0|1448) ESP(spi=0x00000001,seq=0x6), length 1448 00:00:00.000010 IP6 ::1 > ::1: frag (1448|64) What happens in step (3) is: 1) when connecting the socket in __ip6_datagram_connect(), we perform an XFRM lookup, miss the flow cache, create an XFRM bundle, and cache the destination, 2) afterwards, when sending the datagram, we perform an XFRM lookup, again, miss the flow cache (due to mismatch of flowi6_iif and flowi6_oif, which is an issue of its own), and recreate an XFRM bundle based on the cached (and already transformed) destination. To prevent the recreation of an XFRM bundle, avoid an XFRM lookup altogether whenever we already have a destination entry cached in the socket. This prevents the path MTU shrinkage and brings us on par with UDPv4. The fix also benefits connected PINGv6 sockets, another user of ip6_sk_dst_lookup_flow(), who also suffer messages being transformed twice. Joint work with Hannes Frederic Sowa. Reported-by: Jan Tluka <jtluka@redhat.com> Signed-off-by: Jakub Sitnicki <jkbs@redhat.com> Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-06-08 13:13:34 +00:00
return dst;
}
EXPORT_SYMBOL_GPL(ip6_sk_dst_lookup_flow);
static inline struct ipv6_opt_hdr *ip6_opt_dup(struct ipv6_opt_hdr *src,
gfp_t gfp)
{
return src ? kmemdup(src, (src->hdrlen + 1) * 8, gfp) : NULL;
}
static inline struct ipv6_rt_hdr *ip6_rthdr_dup(struct ipv6_rt_hdr *src,
gfp_t gfp)
{
return src ? kmemdup(src, (src->hdrlen + 1) * 8, gfp) : NULL;
}
ipv6: ip6_append_data_mtu did not care about pmtudisc and frag_size If the socket had an IPV6_MTU value set, ip6_append_data_mtu lost track of this when appending the second frame on a corked socket. This results in the following splat: [37598.993962] ------------[ cut here ]------------ [37598.994008] kernel BUG at net/core/skbuff.c:2064! [37598.994008] invalid opcode: 0000 [#1] SMP [37598.994008] Modules linked in: tcp_lp uvcvideo videobuf2_vmalloc videobuf2_memops videobuf2_core videodev media vfat fat usb_storage fuse ebtable_nat xt_CHECKSUM bridge stp llc ipt_MASQUERADE nf_conntrack_netbios_ns nf_conntrack_broadcast ip6table_mangle ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 iptable_nat +nf_nat_ipv4 nf_nat iptable_mangle nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack ebtable_filter ebtables ip6table_filter ip6_tables be2iscsi iscsi_boot_sysfs bnx2i cnic uio cxgb4i cxgb4 cxgb3i cxgb3 mdio libcxgbi ib_iser rdma_cm ib_addr iw_cm ib_cm ib_sa ib_mad ib_core iscsi_tcp libiscsi_tcp libiscsi +scsi_transport_iscsi rfcomm bnep iTCO_wdt iTCO_vendor_support snd_hda_codec_conexant arc4 iwldvm mac80211 snd_hda_intel acpi_cpufreq mperf coretemp snd_hda_codec microcode cdc_wdm cdc_acm [37598.994008] snd_hwdep cdc_ether snd_seq snd_seq_device usbnet mii joydev btusb snd_pcm bluetooth i2c_i801 e1000e lpc_ich mfd_core ptp iwlwifi pps_core snd_page_alloc mei cfg80211 snd_timer thinkpad_acpi snd tpm_tis soundcore rfkill tpm tpm_bios vhost_net tun macvtap macvlan kvm_intel kvm uinput binfmt_misc +dm_crypt i915 i2c_algo_bit drm_kms_helper drm i2c_core wmi video [37598.994008] CPU 0 [37598.994008] Pid: 27320, comm: t2 Not tainted 3.9.6-200.fc18.x86_64 #1 LENOVO 27744PG/27744PG [37598.994008] RIP: 0010:[<ffffffff815443a5>] [<ffffffff815443a5>] skb_copy_and_csum_bits+0x325/0x330 [37598.994008] RSP: 0018:ffff88003670da18 EFLAGS: 00010202 [37598.994008] RAX: ffff88018105c018 RBX: 0000000000000004 RCX: 00000000000006c0 [37598.994008] RDX: ffff88018105a6c0 RSI: ffff88018105a000 RDI: ffff8801e1b0aa00 [37598.994008] RBP: ffff88003670da78 R08: 0000000000000000 R09: ffff88018105c040 [37598.994008] R10: ffff8801e1b0aa00 R11: 0000000000000000 R12: 000000000000fff8 [37598.994008] R13: 00000000000004fc R14: 00000000ffff0504 R15: 0000000000000000 [37598.994008] FS: 00007f28eea59740(0000) GS:ffff88023bc00000(0000) knlGS:0000000000000000 [37598.994008] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [37598.994008] CR2: 0000003d935789e0 CR3: 00000000365cb000 CR4: 00000000000407f0 [37598.994008] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [37598.994008] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [37598.994008] Process t2 (pid: 27320, threadinfo ffff88003670c000, task ffff88022c162ee0) [37598.994008] Stack: [37598.994008] ffff88022e098a00 ffff88020f973fc0 0000000000000008 00000000000004c8 [37598.994008] ffff88020f973fc0 00000000000004c4 ffff88003670da78 ffff8801e1b0a200 [37598.994008] 0000000000000018 00000000000004c8 ffff88020f973fc0 00000000000004c4 [37598.994008] Call Trace: [37598.994008] [<ffffffff815fc21f>] ip6_append_data+0xccf/0xfe0 [37598.994008] [<ffffffff8158d9f0>] ? ip_copy_metadata+0x1a0/0x1a0 [37598.994008] [<ffffffff81661f66>] ? _raw_spin_lock_bh+0x16/0x40 [37598.994008] [<ffffffff8161548d>] udpv6_sendmsg+0x1ed/0xc10 [37598.994008] [<ffffffff812a2845>] ? sock_has_perm+0x75/0x90 [37598.994008] [<ffffffff815c3693>] inet_sendmsg+0x63/0xb0 [37598.994008] [<ffffffff812a2973>] ? selinux_socket_sendmsg+0x23/0x30 [37598.994008] [<ffffffff8153a450>] sock_sendmsg+0xb0/0xe0 [37598.994008] [<ffffffff810135d1>] ? __switch_to+0x181/0x4a0 [37598.994008] [<ffffffff8153d97d>] sys_sendto+0x12d/0x180 [37598.994008] [<ffffffff810dfb64>] ? __audit_syscall_entry+0x94/0xf0 [37598.994008] [<ffffffff81020ed1>] ? syscall_trace_enter+0x231/0x240 [37598.994008] [<ffffffff8166a7e7>] tracesys+0xdd/0xe2 [37598.994008] Code: fe 07 00 00 48 c7 c7 04 28 a6 81 89 45 a0 4c 89 4d b8 44 89 5d a8 e8 1b ac b1 ff 44 8b 5d a8 4c 8b 4d b8 8b 45 a0 e9 cf fe ff ff <0f> 0b 66 0f 1f 84 00 00 00 00 00 66 66 66 66 90 55 48 89 e5 48 [37598.994008] RIP [<ffffffff815443a5>] skb_copy_and_csum_bits+0x325/0x330 [37598.994008] RSP <ffff88003670da18> [37599.007323] ---[ end trace d69f6a17f8ac8eee ]--- While there, also check if path mtu discovery is activated for this socket. The logic was adapted from ip6_append_data when first writing on the corked socket. This bug was introduced with commit 0c1833797a5a6ec23ea9261d979aa18078720b74 ("ipv6: fix incorrect ipsec fragment"). v2: a) Replace IPV6_PMTU_DISC_DO with IPV6_PMTUDISC_PROBE. b) Don't pass ipv6_pinfo to ip6_append_data_mtu (suggestion by Gao feng, thanks!). c) Change mtu to unsigned int, else we get a warning about non-matching types because of the min()-macro type-check. Acked-by: Gao feng <gaofeng@cn.fujitsu.com> Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-07-02 06:04:05 +00:00
static void ip6_append_data_mtu(unsigned int *mtu,
int *maxfraglen,
unsigned int fragheaderlen,
struct sk_buff *skb,
ipv6: ip6_append_data_mtu did not care about pmtudisc and frag_size If the socket had an IPV6_MTU value set, ip6_append_data_mtu lost track of this when appending the second frame on a corked socket. This results in the following splat: [37598.993962] ------------[ cut here ]------------ [37598.994008] kernel BUG at net/core/skbuff.c:2064! [37598.994008] invalid opcode: 0000 [#1] SMP [37598.994008] Modules linked in: tcp_lp uvcvideo videobuf2_vmalloc videobuf2_memops videobuf2_core videodev media vfat fat usb_storage fuse ebtable_nat xt_CHECKSUM bridge stp llc ipt_MASQUERADE nf_conntrack_netbios_ns nf_conntrack_broadcast ip6table_mangle ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 iptable_nat +nf_nat_ipv4 nf_nat iptable_mangle nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack ebtable_filter ebtables ip6table_filter ip6_tables be2iscsi iscsi_boot_sysfs bnx2i cnic uio cxgb4i cxgb4 cxgb3i cxgb3 mdio libcxgbi ib_iser rdma_cm ib_addr iw_cm ib_cm ib_sa ib_mad ib_core iscsi_tcp libiscsi_tcp libiscsi +scsi_transport_iscsi rfcomm bnep iTCO_wdt iTCO_vendor_support snd_hda_codec_conexant arc4 iwldvm mac80211 snd_hda_intel acpi_cpufreq mperf coretemp snd_hda_codec microcode cdc_wdm cdc_acm [37598.994008] snd_hwdep cdc_ether snd_seq snd_seq_device usbnet mii joydev btusb snd_pcm bluetooth i2c_i801 e1000e lpc_ich mfd_core ptp iwlwifi pps_core snd_page_alloc mei cfg80211 snd_timer thinkpad_acpi snd tpm_tis soundcore rfkill tpm tpm_bios vhost_net tun macvtap macvlan kvm_intel kvm uinput binfmt_misc +dm_crypt i915 i2c_algo_bit drm_kms_helper drm i2c_core wmi video [37598.994008] CPU 0 [37598.994008] Pid: 27320, comm: t2 Not tainted 3.9.6-200.fc18.x86_64 #1 LENOVO 27744PG/27744PG [37598.994008] RIP: 0010:[<ffffffff815443a5>] [<ffffffff815443a5>] skb_copy_and_csum_bits+0x325/0x330 [37598.994008] RSP: 0018:ffff88003670da18 EFLAGS: 00010202 [37598.994008] RAX: ffff88018105c018 RBX: 0000000000000004 RCX: 00000000000006c0 [37598.994008] RDX: ffff88018105a6c0 RSI: ffff88018105a000 RDI: ffff8801e1b0aa00 [37598.994008] RBP: ffff88003670da78 R08: 0000000000000000 R09: ffff88018105c040 [37598.994008] R10: ffff8801e1b0aa00 R11: 0000000000000000 R12: 000000000000fff8 [37598.994008] R13: 00000000000004fc R14: 00000000ffff0504 R15: 0000000000000000 [37598.994008] FS: 00007f28eea59740(0000) GS:ffff88023bc00000(0000) knlGS:0000000000000000 [37598.994008] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [37598.994008] CR2: 0000003d935789e0 CR3: 00000000365cb000 CR4: 00000000000407f0 [37598.994008] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [37598.994008] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [37598.994008] Process t2 (pid: 27320, threadinfo ffff88003670c000, task ffff88022c162ee0) [37598.994008] Stack: [37598.994008] ffff88022e098a00 ffff88020f973fc0 0000000000000008 00000000000004c8 [37598.994008] ffff88020f973fc0 00000000000004c4 ffff88003670da78 ffff8801e1b0a200 [37598.994008] 0000000000000018 00000000000004c8 ffff88020f973fc0 00000000000004c4 [37598.994008] Call Trace: [37598.994008] [<ffffffff815fc21f>] ip6_append_data+0xccf/0xfe0 [37598.994008] [<ffffffff8158d9f0>] ? ip_copy_metadata+0x1a0/0x1a0 [37598.994008] [<ffffffff81661f66>] ? _raw_spin_lock_bh+0x16/0x40 [37598.994008] [<ffffffff8161548d>] udpv6_sendmsg+0x1ed/0xc10 [37598.994008] [<ffffffff812a2845>] ? sock_has_perm+0x75/0x90 [37598.994008] [<ffffffff815c3693>] inet_sendmsg+0x63/0xb0 [37598.994008] [<ffffffff812a2973>] ? selinux_socket_sendmsg+0x23/0x30 [37598.994008] [<ffffffff8153a450>] sock_sendmsg+0xb0/0xe0 [37598.994008] [<ffffffff810135d1>] ? __switch_to+0x181/0x4a0 [37598.994008] [<ffffffff8153d97d>] sys_sendto+0x12d/0x180 [37598.994008] [<ffffffff810dfb64>] ? __audit_syscall_entry+0x94/0xf0 [37598.994008] [<ffffffff81020ed1>] ? syscall_trace_enter+0x231/0x240 [37598.994008] [<ffffffff8166a7e7>] tracesys+0xdd/0xe2 [37598.994008] Code: fe 07 00 00 48 c7 c7 04 28 a6 81 89 45 a0 4c 89 4d b8 44 89 5d a8 e8 1b ac b1 ff 44 8b 5d a8 4c 8b 4d b8 8b 45 a0 e9 cf fe ff ff <0f> 0b 66 0f 1f 84 00 00 00 00 00 66 66 66 66 90 55 48 89 e5 48 [37598.994008] RIP [<ffffffff815443a5>] skb_copy_and_csum_bits+0x325/0x330 [37598.994008] RSP <ffff88003670da18> [37599.007323] ---[ end trace d69f6a17f8ac8eee ]--- While there, also check if path mtu discovery is activated for this socket. The logic was adapted from ip6_append_data when first writing on the corked socket. This bug was introduced with commit 0c1833797a5a6ec23ea9261d979aa18078720b74 ("ipv6: fix incorrect ipsec fragment"). v2: a) Replace IPV6_PMTU_DISC_DO with IPV6_PMTUDISC_PROBE. b) Don't pass ipv6_pinfo to ip6_append_data_mtu (suggestion by Gao feng, thanks!). c) Change mtu to unsigned int, else we get a warning about non-matching types because of the min()-macro type-check. Acked-by: Gao feng <gaofeng@cn.fujitsu.com> Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-07-02 06:04:05 +00:00
struct rt6_info *rt,
unsigned int orig_mtu)
{
if (!(rt->dst.flags & DST_XFRM_TUNNEL)) {
if (!skb) {
/* first fragment, reserve header_len */
*mtu = orig_mtu - rt->dst.header_len;
} else {
/*
* this fragment is not first, the headers
* space is regarded as data space.
*/
*mtu = orig_mtu;
}
*maxfraglen = ((*mtu - fragheaderlen) & ~7)
+ fragheaderlen - sizeof(struct frag_hdr);
}
}
static int ip6_setup_cork(struct sock *sk, struct inet_cork_full *cork,
struct inet6_cork *v6_cork, struct ipcm6_cookie *ipc6,
struct rt6_info *rt, struct flowi6 *fl6)
{
struct ipv6_pinfo *np = inet6_sk(sk);
unsigned int mtu;
struct ipv6_txoptions *opt = ipc6->opt;
/*
* setup for corking
*/
if (opt) {
if (WARN_ON(v6_cork->opt))
return -EINVAL;
ipv6: flowlabel: do not leave opt->tot_len with garbage When syzkaller team brought us a C repro for the crash [1] that had been reported many times in the past, I finally could find the root cause. If FlowLabel info is merged by fl6_merge_options(), we leave part of the opt_space storage provided by udp/raw/l2tp with random value in opt_space.tot_len, unless a control message was provided at sendmsg() time. Then ip6_setup_cork() would use this random value to perform a kzalloc() call. Undefined behavior and crashes. Fix is to properly set tot_len in fl6_merge_options() At the same time, we can also avoid consuming memory and cpu cycles to clear it, if every option is copied via a kmemdup(). This is the change in ip6_setup_cork(). [1] kasan: CONFIG_KASAN_INLINE enabled kasan: GPF could be caused by NULL-ptr deref or user memory access general protection fault: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 0 PID: 6613 Comm: syz-executor0 Not tainted 4.14.0-rc4+ #127 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 task: ffff8801cb64a100 task.stack: ffff8801cc350000 RIP: 0010:ip6_setup_cork+0x274/0x15c0 net/ipv6/ip6_output.c:1168 RSP: 0018:ffff8801cc357550 EFLAGS: 00010203 RAX: dffffc0000000000 RBX: ffff8801cc357748 RCX: 0000000000000010 RDX: 0000000000000002 RSI: ffffffff842bd1d9 RDI: 0000000000000014 RBP: ffff8801cc357620 R08: ffff8801cb17f380 R09: ffff8801cc357b10 R10: ffff8801cb64a100 R11: 0000000000000000 R12: ffff8801cc357ab0 R13: ffff8801cc357b10 R14: 0000000000000000 R15: ffff8801c3bbf0c0 FS: 00007f9c5c459700(0000) GS:ffff8801db200000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000020324000 CR3: 00000001d1cf2000 CR4: 00000000001406f0 DR0: 0000000020001010 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000600 Call Trace: ip6_make_skb+0x282/0x530 net/ipv6/ip6_output.c:1729 udpv6_sendmsg+0x2769/0x3380 net/ipv6/udp.c:1340 inet_sendmsg+0x11f/0x5e0 net/ipv4/af_inet.c:762 sock_sendmsg_nosec net/socket.c:633 [inline] sock_sendmsg+0xca/0x110 net/socket.c:643 SYSC_sendto+0x358/0x5a0 net/socket.c:1750 SyS_sendto+0x40/0x50 net/socket.c:1718 entry_SYSCALL_64_fastpath+0x1f/0xbe RIP: 0033:0x4520a9 RSP: 002b:00007f9c5c458c08 EFLAGS: 00000216 ORIG_RAX: 000000000000002c RAX: ffffffffffffffda RBX: 0000000000718000 RCX: 00000000004520a9 RDX: 0000000000000001 RSI: 0000000020fd1000 RDI: 0000000000000016 RBP: 0000000000000086 R08: 0000000020e0afe4 R09: 000000000000001c R10: 0000000000000000 R11: 0000000000000216 R12: 00000000004bb1ee R13: 00000000ffffffff R14: 0000000000000016 R15: 0000000000000029 Code: e0 07 83 c0 03 38 d0 7c 08 84 d2 0f 85 ea 0f 00 00 48 8d 79 04 48 b8 00 00 00 00 00 fc ff df 45 8b 74 24 04 48 89 fa 48 c1 ea 03 <0f> b6 14 02 48 89 f8 83 e0 07 83 c0 03 38 d0 7c 08 84 d2 0f 85 RIP: ip6_setup_cork+0x274/0x15c0 net/ipv6/ip6_output.c:1168 RSP: ffff8801cc357550 Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Dmitry Vyukov <dvyukov@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-21 19:26:23 +00:00
v6_cork->opt = kzalloc(sizeof(*opt), sk->sk_allocation);
if (unlikely(!v6_cork->opt))
return -ENOBUFS;
ipv6: flowlabel: do not leave opt->tot_len with garbage When syzkaller team brought us a C repro for the crash [1] that had been reported many times in the past, I finally could find the root cause. If FlowLabel info is merged by fl6_merge_options(), we leave part of the opt_space storage provided by udp/raw/l2tp with random value in opt_space.tot_len, unless a control message was provided at sendmsg() time. Then ip6_setup_cork() would use this random value to perform a kzalloc() call. Undefined behavior and crashes. Fix is to properly set tot_len in fl6_merge_options() At the same time, we can also avoid consuming memory and cpu cycles to clear it, if every option is copied via a kmemdup(). This is the change in ip6_setup_cork(). [1] kasan: CONFIG_KASAN_INLINE enabled kasan: GPF could be caused by NULL-ptr deref or user memory access general protection fault: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 0 PID: 6613 Comm: syz-executor0 Not tainted 4.14.0-rc4+ #127 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 task: ffff8801cb64a100 task.stack: ffff8801cc350000 RIP: 0010:ip6_setup_cork+0x274/0x15c0 net/ipv6/ip6_output.c:1168 RSP: 0018:ffff8801cc357550 EFLAGS: 00010203 RAX: dffffc0000000000 RBX: ffff8801cc357748 RCX: 0000000000000010 RDX: 0000000000000002 RSI: ffffffff842bd1d9 RDI: 0000000000000014 RBP: ffff8801cc357620 R08: ffff8801cb17f380 R09: ffff8801cc357b10 R10: ffff8801cb64a100 R11: 0000000000000000 R12: ffff8801cc357ab0 R13: ffff8801cc357b10 R14: 0000000000000000 R15: ffff8801c3bbf0c0 FS: 00007f9c5c459700(0000) GS:ffff8801db200000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000020324000 CR3: 00000001d1cf2000 CR4: 00000000001406f0 DR0: 0000000020001010 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000600 Call Trace: ip6_make_skb+0x282/0x530 net/ipv6/ip6_output.c:1729 udpv6_sendmsg+0x2769/0x3380 net/ipv6/udp.c:1340 inet_sendmsg+0x11f/0x5e0 net/ipv4/af_inet.c:762 sock_sendmsg_nosec net/socket.c:633 [inline] sock_sendmsg+0xca/0x110 net/socket.c:643 SYSC_sendto+0x358/0x5a0 net/socket.c:1750 SyS_sendto+0x40/0x50 net/socket.c:1718 entry_SYSCALL_64_fastpath+0x1f/0xbe RIP: 0033:0x4520a9 RSP: 002b:00007f9c5c458c08 EFLAGS: 00000216 ORIG_RAX: 000000000000002c RAX: ffffffffffffffda RBX: 0000000000718000 RCX: 00000000004520a9 RDX: 0000000000000001 RSI: 0000000020fd1000 RDI: 0000000000000016 RBP: 0000000000000086 R08: 0000000020e0afe4 R09: 000000000000001c R10: 0000000000000000 R11: 0000000000000216 R12: 00000000004bb1ee R13: 00000000ffffffff R14: 0000000000000016 R15: 0000000000000029 Code: e0 07 83 c0 03 38 d0 7c 08 84 d2 0f 85 ea 0f 00 00 48 8d 79 04 48 b8 00 00 00 00 00 fc ff df 45 8b 74 24 04 48 89 fa 48 c1 ea 03 <0f> b6 14 02 48 89 f8 83 e0 07 83 c0 03 38 d0 7c 08 84 d2 0f 85 RIP: ip6_setup_cork+0x274/0x15c0 net/ipv6/ip6_output.c:1168 RSP: ffff8801cc357550 Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Dmitry Vyukov <dvyukov@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-21 19:26:23 +00:00
v6_cork->opt->tot_len = sizeof(*opt);
v6_cork->opt->opt_flen = opt->opt_flen;
v6_cork->opt->opt_nflen = opt->opt_nflen;
v6_cork->opt->dst0opt = ip6_opt_dup(opt->dst0opt,
sk->sk_allocation);
if (opt->dst0opt && !v6_cork->opt->dst0opt)
return -ENOBUFS;
v6_cork->opt->dst1opt = ip6_opt_dup(opt->dst1opt,
sk->sk_allocation);
if (opt->dst1opt && !v6_cork->opt->dst1opt)
return -ENOBUFS;
v6_cork->opt->hopopt = ip6_opt_dup(opt->hopopt,
sk->sk_allocation);
if (opt->hopopt && !v6_cork->opt->hopopt)
return -ENOBUFS;
v6_cork->opt->srcrt = ip6_rthdr_dup(opt->srcrt,
sk->sk_allocation);
if (opt->srcrt && !v6_cork->opt->srcrt)
return -ENOBUFS;
/* need source address above miyazawa*/
}
dst_hold(&rt->dst);
cork->base.dst = &rt->dst;
cork->fl.u.ip6 = *fl6;
v6_cork->hop_limit = ipc6->hlimit;
v6_cork->tclass = ipc6->tclass;
if (rt->dst.flags & DST_XFRM_TUNNEL)
mtu = np->pmtudisc >= IPV6_PMTUDISC_PROBE ?
rt->dst.dev->mtu : dst_mtu(&rt->dst);
else
mtu = np->pmtudisc >= IPV6_PMTUDISC_PROBE ?
rt->dst.dev->mtu : dst_mtu(rt->dst.path);
if (np->frag_size < mtu) {
if (np->frag_size)
mtu = np->frag_size;
}
cork->base.fragsize = mtu;
if (dst_allfrag(rt->dst.path))
cork->base.flags |= IPCORK_ALLFRAG;
cork->base.length = 0;
return 0;
}
static int __ip6_append_data(struct sock *sk,
struct flowi6 *fl6,
struct sk_buff_head *queue,
struct inet_cork *cork,
struct inet6_cork *v6_cork,
struct page_frag *pfrag,
int getfrag(void *from, char *to, int offset,
int len, int odd, struct sk_buff *skb),
void *from, int length, int transhdrlen,
unsigned int flags, struct ipcm6_cookie *ipc6,
const struct sockcm_cookie *sockc)
{
struct sk_buff *skb, *skb_prev = NULL;
unsigned int maxfraglen, fragheaderlen, mtu, orig_mtu;
int exthdrlen = 0;
int dst_exthdrlen = 0;
int hh_len;
int copy;
int err;
int offset = 0;
__u8 tx_flags = 0;
u32 tskey = 0;
struct rt6_info *rt = (struct rt6_info *)cork->dst;
struct ipv6_txoptions *opt = v6_cork->opt;
int csummode = CHECKSUM_NONE;
unsigned int maxnonfragsize, headersize;
skb = skb_peek_tail(queue);
if (!skb) {
exthdrlen = opt ? opt->opt_flen : 0;
dst_exthdrlen = rt->dst.header_len - rt->rt6i_nfheader_len;
}
mtu = cork->fragsize;
orig_mtu = mtu;
hh_len = LL_RESERVED_SPACE(rt->dst.dev);
fragheaderlen = sizeof(struct ipv6hdr) + rt->rt6i_nfheader_len +
(opt ? opt->opt_nflen : 0);
maxfraglen = ((mtu - fragheaderlen) & ~7) + fragheaderlen -
sizeof(struct frag_hdr);
headersize = sizeof(struct ipv6hdr) +
(opt ? opt->opt_flen + opt->opt_nflen : 0) +
(dst_allfrag(&rt->dst) ?
sizeof(struct frag_hdr) : 0) +
rt->rt6i_nfheader_len;
if (cork->length + length > mtu - headersize && ipc6->dontfrag &&
(sk->sk_protocol == IPPROTO_UDP ||
sk->sk_protocol == IPPROTO_RAW)) {
ipv6_local_rxpmtu(sk, fl6, mtu - headersize +
sizeof(struct ipv6hdr));
goto emsgsize;
}
if (ip6_sk_ignore_df(sk))
maxnonfragsize = sizeof(struct ipv6hdr) + IPV6_MAXPLEN;
else
maxnonfragsize = mtu;
if (cork->length + length > maxnonfragsize - headersize) {
emsgsize:
ipv6_local_error(sk, EMSGSIZE, fl6,
mtu - headersize +
sizeof(struct ipv6hdr));
return -EMSGSIZE;
}
/* CHECKSUM_PARTIAL only with no extension headers and when
* we are not going to fragment
*/
if (transhdrlen && sk->sk_protocol == IPPROTO_UDP &&
headersize == sizeof(struct ipv6hdr) &&
length <= mtu - headersize &&
!(flags & MSG_MORE) &&
rt->dst.dev->features & (NETIF_F_IPV6_CSUM | NETIF_F_HW_CSUM))
csummode = CHECKSUM_PARTIAL;
if (sk->sk_type == SOCK_DGRAM || sk->sk_type == SOCK_RAW) {
sock_tx_timestamp(sk, sockc->tsflags, &tx_flags);
if (tx_flags & SKBTX_ANY_SW_TSTAMP &&
sk->sk_tsflags & SOF_TIMESTAMPING_OPT_ID)
tskey = sk->sk_tskey++;
}
/*
* Let's try using as much space as possible.
* Use MTU if total length of the message fits into the MTU.
* Otherwise, we need to reserve fragment header and
* fragment alignment (= 8-15 octects, in total).
*
* Note that we may need to "move" the data from the tail of
* of the buffer to the new fragment when we split
* the message.
*
* FIXME: It may be fragmented into multiple chunks
* at once if non-fragmentable extension headers
* are too large.
* --yoshfuji
*/
cork->length += length;
if (!skb)
goto alloc_new_skb;
while (length > 0) {
/* Check if the remaining data fits into current packet. */
copy = (cork->length <= mtu && !(cork->flags & IPCORK_ALLFRAG) ? mtu : maxfraglen) - skb->len;
if (copy < length)
copy = maxfraglen - skb->len;
if (copy <= 0) {
char *data;
unsigned int datalen;
unsigned int fraglen;
unsigned int fraggap;
unsigned int alloclen;
alloc_new_skb:
/* There's no room in the current skb */
if (skb)
fraggap = skb->len - maxfraglen;
else
fraggap = 0;
/* update mtu and maxfraglen if necessary */
if (!skb || !skb_prev)
ip6_append_data_mtu(&mtu, &maxfraglen,
ipv6: ip6_append_data_mtu did not care about pmtudisc and frag_size If the socket had an IPV6_MTU value set, ip6_append_data_mtu lost track of this when appending the second frame on a corked socket. This results in the following splat: [37598.993962] ------------[ cut here ]------------ [37598.994008] kernel BUG at net/core/skbuff.c:2064! [37598.994008] invalid opcode: 0000 [#1] SMP [37598.994008] Modules linked in: tcp_lp uvcvideo videobuf2_vmalloc videobuf2_memops videobuf2_core videodev media vfat fat usb_storage fuse ebtable_nat xt_CHECKSUM bridge stp llc ipt_MASQUERADE nf_conntrack_netbios_ns nf_conntrack_broadcast ip6table_mangle ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 iptable_nat +nf_nat_ipv4 nf_nat iptable_mangle nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack ebtable_filter ebtables ip6table_filter ip6_tables be2iscsi iscsi_boot_sysfs bnx2i cnic uio cxgb4i cxgb4 cxgb3i cxgb3 mdio libcxgbi ib_iser rdma_cm ib_addr iw_cm ib_cm ib_sa ib_mad ib_core iscsi_tcp libiscsi_tcp libiscsi +scsi_transport_iscsi rfcomm bnep iTCO_wdt iTCO_vendor_support snd_hda_codec_conexant arc4 iwldvm mac80211 snd_hda_intel acpi_cpufreq mperf coretemp snd_hda_codec microcode cdc_wdm cdc_acm [37598.994008] snd_hwdep cdc_ether snd_seq snd_seq_device usbnet mii joydev btusb snd_pcm bluetooth i2c_i801 e1000e lpc_ich mfd_core ptp iwlwifi pps_core snd_page_alloc mei cfg80211 snd_timer thinkpad_acpi snd tpm_tis soundcore rfkill tpm tpm_bios vhost_net tun macvtap macvlan kvm_intel kvm uinput binfmt_misc +dm_crypt i915 i2c_algo_bit drm_kms_helper drm i2c_core wmi video [37598.994008] CPU 0 [37598.994008] Pid: 27320, comm: t2 Not tainted 3.9.6-200.fc18.x86_64 #1 LENOVO 27744PG/27744PG [37598.994008] RIP: 0010:[<ffffffff815443a5>] [<ffffffff815443a5>] skb_copy_and_csum_bits+0x325/0x330 [37598.994008] RSP: 0018:ffff88003670da18 EFLAGS: 00010202 [37598.994008] RAX: ffff88018105c018 RBX: 0000000000000004 RCX: 00000000000006c0 [37598.994008] RDX: ffff88018105a6c0 RSI: ffff88018105a000 RDI: ffff8801e1b0aa00 [37598.994008] RBP: ffff88003670da78 R08: 0000000000000000 R09: ffff88018105c040 [37598.994008] R10: ffff8801e1b0aa00 R11: 0000000000000000 R12: 000000000000fff8 [37598.994008] R13: 00000000000004fc R14: 00000000ffff0504 R15: 0000000000000000 [37598.994008] FS: 00007f28eea59740(0000) GS:ffff88023bc00000(0000) knlGS:0000000000000000 [37598.994008] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [37598.994008] CR2: 0000003d935789e0 CR3: 00000000365cb000 CR4: 00000000000407f0 [37598.994008] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [37598.994008] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [37598.994008] Process t2 (pid: 27320, threadinfo ffff88003670c000, task ffff88022c162ee0) [37598.994008] Stack: [37598.994008] ffff88022e098a00 ffff88020f973fc0 0000000000000008 00000000000004c8 [37598.994008] ffff88020f973fc0 00000000000004c4 ffff88003670da78 ffff8801e1b0a200 [37598.994008] 0000000000000018 00000000000004c8 ffff88020f973fc0 00000000000004c4 [37598.994008] Call Trace: [37598.994008] [<ffffffff815fc21f>] ip6_append_data+0xccf/0xfe0 [37598.994008] [<ffffffff8158d9f0>] ? ip_copy_metadata+0x1a0/0x1a0 [37598.994008] [<ffffffff81661f66>] ? _raw_spin_lock_bh+0x16/0x40 [37598.994008] [<ffffffff8161548d>] udpv6_sendmsg+0x1ed/0xc10 [37598.994008] [<ffffffff812a2845>] ? sock_has_perm+0x75/0x90 [37598.994008] [<ffffffff815c3693>] inet_sendmsg+0x63/0xb0 [37598.994008] [<ffffffff812a2973>] ? selinux_socket_sendmsg+0x23/0x30 [37598.994008] [<ffffffff8153a450>] sock_sendmsg+0xb0/0xe0 [37598.994008] [<ffffffff810135d1>] ? __switch_to+0x181/0x4a0 [37598.994008] [<ffffffff8153d97d>] sys_sendto+0x12d/0x180 [37598.994008] [<ffffffff810dfb64>] ? __audit_syscall_entry+0x94/0xf0 [37598.994008] [<ffffffff81020ed1>] ? syscall_trace_enter+0x231/0x240 [37598.994008] [<ffffffff8166a7e7>] tracesys+0xdd/0xe2 [37598.994008] Code: fe 07 00 00 48 c7 c7 04 28 a6 81 89 45 a0 4c 89 4d b8 44 89 5d a8 e8 1b ac b1 ff 44 8b 5d a8 4c 8b 4d b8 8b 45 a0 e9 cf fe ff ff <0f> 0b 66 0f 1f 84 00 00 00 00 00 66 66 66 66 90 55 48 89 e5 48 [37598.994008] RIP [<ffffffff815443a5>] skb_copy_and_csum_bits+0x325/0x330 [37598.994008] RSP <ffff88003670da18> [37599.007323] ---[ end trace d69f6a17f8ac8eee ]--- While there, also check if path mtu discovery is activated for this socket. The logic was adapted from ip6_append_data when first writing on the corked socket. This bug was introduced with commit 0c1833797a5a6ec23ea9261d979aa18078720b74 ("ipv6: fix incorrect ipsec fragment"). v2: a) Replace IPV6_PMTU_DISC_DO with IPV6_PMTUDISC_PROBE. b) Don't pass ipv6_pinfo to ip6_append_data_mtu (suggestion by Gao feng, thanks!). c) Change mtu to unsigned int, else we get a warning about non-matching types because of the min()-macro type-check. Acked-by: Gao feng <gaofeng@cn.fujitsu.com> Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-07-02 06:04:05 +00:00
fragheaderlen, skb, rt,
orig_mtu);
skb_prev = skb;
/*
* If remaining data exceeds the mtu,
* we know we need more fragment(s).
*/
datalen = length + fraggap;
if (datalen > (cork->length <= mtu && !(cork->flags & IPCORK_ALLFRAG) ? mtu : maxfraglen) - fragheaderlen)
datalen = maxfraglen - fragheaderlen - rt->dst.trailer_len;
if ((flags & MSG_MORE) &&
!(rt->dst.dev->features&NETIF_F_SG))
alloclen = mtu;
else
alloclen = datalen + fragheaderlen;
alloclen += dst_exthdrlen;
if (datalen != length + fraggap) {
/*
* this is not the last fragment, the trailer
* space is regarded as data space.
*/
datalen += rt->dst.trailer_len;
}
alloclen += rt->dst.trailer_len;
fraglen = datalen + fragheaderlen;
/*
* We just reserve space for fragment header.
* Note: this may be overallocation if the message
* (without MSG_MORE) fits into the MTU.
*/
alloclen += sizeof(struct frag_hdr);
copy = datalen - transhdrlen - fraggap;
if (copy < 0) {
err = -EINVAL;
goto error;
}
if (transhdrlen) {
skb = sock_alloc_send_skb(sk,
alloclen + hh_len,
(flags & MSG_DONTWAIT), &err);
} else {
skb = NULL;
if (refcount_read(&sk->sk_wmem_alloc) <=
2 * sk->sk_sndbuf)
skb = sock_wmalloc(sk,
alloclen + hh_len, 1,
sk->sk_allocation);
if (unlikely(!skb))
err = -ENOBUFS;
}
if (!skb)
goto error;
/*
* Fill in the control structures
*/
skb->protocol = htons(ETH_P_IPV6);
skb->ip_summed = csummode;
skb->csum = 0;
/* reserve for fragmentation and ipsec header */
skb_reserve(skb, hh_len + sizeof(struct frag_hdr) +
dst_exthdrlen);
/* Only the initial fragment is time stamped */
skb_shinfo(skb)->tx_flags = tx_flags;
tx_flags = 0;
skb_shinfo(skb)->tskey = tskey;
tskey = 0;
/*
* Find where to start putting bytes
*/
data = skb_put(skb, fraglen);
skb_set_network_header(skb, exthdrlen);
data += fragheaderlen;
skb->transport_header = (skb->network_header +
fragheaderlen);
if (fraggap) {
skb->csum = skb_copy_and_csum_bits(
skb_prev, maxfraglen,
data + transhdrlen, fraggap, 0);
skb_prev->csum = csum_sub(skb_prev->csum,
skb->csum);
data += fraggap;
pskb_trim_unique(skb_prev, maxfraglen);
}
if (copy > 0 &&
getfrag(from, data + transhdrlen, offset,
copy, fraggap, skb) < 0) {
err = -EFAULT;
kfree_skb(skb);
goto error;
}
offset += copy;
length -= datalen - fraggap;
transhdrlen = 0;
exthdrlen = 0;
dst_exthdrlen = 0;
if ((flags & MSG_CONFIRM) && !skb_prev)
skb_set_dst_pending_confirm(skb, 1);
/*
* Put the packet on the pending queue
*/
__skb_queue_tail(queue, skb);
continue;
}
if (copy > length)
copy = length;
if (!(rt->dst.dev->features&NETIF_F_SG)) {
unsigned int off;
off = skb->len;
if (getfrag(from, skb_put(skb, copy),
offset, copy, off, skb) < 0) {
__skb_trim(skb, off);
err = -EFAULT;
goto error;
}
} else {
int i = skb_shinfo(skb)->nr_frags;
net: use a per task frag allocator We currently use a per socket order-0 page cache for tcp_sendmsg() operations. This page is used to build fragments for skbs. Its done to increase probability of coalescing small write() into single segments in skbs still in write queue (not yet sent) But it wastes a lot of memory for applications handling many mostly idle sockets, since each socket holds one page in sk->sk_sndmsg_page Its also quite inefficient to build TSO 64KB packets, because we need about 16 pages per skb on arches where PAGE_SIZE = 4096, so we hit page allocator more than wanted. This patch adds a per task frag allocator and uses bigger pages, if available. An automatic fallback is done in case of memory pressure. (up to 32768 bytes per frag, thats order-3 pages on x86) This increases TCP stream performance by 20% on loopback device, but also benefits on other network devices, since 8x less frags are mapped on transmit and unmapped on tx completion. Alexander Duyck mentioned a probable performance win on systems with IOMMU enabled. Its possible some SG enabled hardware cant cope with bigger fragments, but their ndo_start_xmit() should already handle this, splitting a fragment in sub fragments, since some arches have PAGE_SIZE=65536 Successfully tested on various ethernet devices. (ixgbe, igb, bnx2x, tg3, mellanox mlx4) Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Ben Hutchings <bhutchings@solarflare.com> Cc: Vijay Subramanian <subramanian.vijay@gmail.com> Cc: Alexander Duyck <alexander.h.duyck@intel.com> Tested-by: Vijay Subramanian <subramanian.vijay@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-09-23 23:04:42 +00:00
err = -ENOMEM;
if (!sk_page_frag_refill(sk, pfrag))
goto error;
net: use a per task frag allocator We currently use a per socket order-0 page cache for tcp_sendmsg() operations. This page is used to build fragments for skbs. Its done to increase probability of coalescing small write() into single segments in skbs still in write queue (not yet sent) But it wastes a lot of memory for applications handling many mostly idle sockets, since each socket holds one page in sk->sk_sndmsg_page Its also quite inefficient to build TSO 64KB packets, because we need about 16 pages per skb on arches where PAGE_SIZE = 4096, so we hit page allocator more than wanted. This patch adds a per task frag allocator and uses bigger pages, if available. An automatic fallback is done in case of memory pressure. (up to 32768 bytes per frag, thats order-3 pages on x86) This increases TCP stream performance by 20% on loopback device, but also benefits on other network devices, since 8x less frags are mapped on transmit and unmapped on tx completion. Alexander Duyck mentioned a probable performance win on systems with IOMMU enabled. Its possible some SG enabled hardware cant cope with bigger fragments, but their ndo_start_xmit() should already handle this, splitting a fragment in sub fragments, since some arches have PAGE_SIZE=65536 Successfully tested on various ethernet devices. (ixgbe, igb, bnx2x, tg3, mellanox mlx4) Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Ben Hutchings <bhutchings@solarflare.com> Cc: Vijay Subramanian <subramanian.vijay@gmail.com> Cc: Alexander Duyck <alexander.h.duyck@intel.com> Tested-by: Vijay Subramanian <subramanian.vijay@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-09-23 23:04:42 +00:00
if (!skb_can_coalesce(skb, i, pfrag->page,
pfrag->offset)) {
err = -EMSGSIZE;
if (i == MAX_SKB_FRAGS)
goto error;
__skb_fill_page_desc(skb, i, pfrag->page,
pfrag->offset, 0);
skb_shinfo(skb)->nr_frags = ++i;
get_page(pfrag->page);
}
net: use a per task frag allocator We currently use a per socket order-0 page cache for tcp_sendmsg() operations. This page is used to build fragments for skbs. Its done to increase probability of coalescing small write() into single segments in skbs still in write queue (not yet sent) But it wastes a lot of memory for applications handling many mostly idle sockets, since each socket holds one page in sk->sk_sndmsg_page Its also quite inefficient to build TSO 64KB packets, because we need about 16 pages per skb on arches where PAGE_SIZE = 4096, so we hit page allocator more than wanted. This patch adds a per task frag allocator and uses bigger pages, if available. An automatic fallback is done in case of memory pressure. (up to 32768 bytes per frag, thats order-3 pages on x86) This increases TCP stream performance by 20% on loopback device, but also benefits on other network devices, since 8x less frags are mapped on transmit and unmapped on tx completion. Alexander Duyck mentioned a probable performance win on systems with IOMMU enabled. Its possible some SG enabled hardware cant cope with bigger fragments, but their ndo_start_xmit() should already handle this, splitting a fragment in sub fragments, since some arches have PAGE_SIZE=65536 Successfully tested on various ethernet devices. (ixgbe, igb, bnx2x, tg3, mellanox mlx4) Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Ben Hutchings <bhutchings@solarflare.com> Cc: Vijay Subramanian <subramanian.vijay@gmail.com> Cc: Alexander Duyck <alexander.h.duyck@intel.com> Tested-by: Vijay Subramanian <subramanian.vijay@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-09-23 23:04:42 +00:00
copy = min_t(int, copy, pfrag->size - pfrag->offset);
if (getfrag(from,
net: use a per task frag allocator We currently use a per socket order-0 page cache for tcp_sendmsg() operations. This page is used to build fragments for skbs. Its done to increase probability of coalescing small write() into single segments in skbs still in write queue (not yet sent) But it wastes a lot of memory for applications handling many mostly idle sockets, since each socket holds one page in sk->sk_sndmsg_page Its also quite inefficient to build TSO 64KB packets, because we need about 16 pages per skb on arches where PAGE_SIZE = 4096, so we hit page allocator more than wanted. This patch adds a per task frag allocator and uses bigger pages, if available. An automatic fallback is done in case of memory pressure. (up to 32768 bytes per frag, thats order-3 pages on x86) This increases TCP stream performance by 20% on loopback device, but also benefits on other network devices, since 8x less frags are mapped on transmit and unmapped on tx completion. Alexander Duyck mentioned a probable performance win on systems with IOMMU enabled. Its possible some SG enabled hardware cant cope with bigger fragments, but their ndo_start_xmit() should already handle this, splitting a fragment in sub fragments, since some arches have PAGE_SIZE=65536 Successfully tested on various ethernet devices. (ixgbe, igb, bnx2x, tg3, mellanox mlx4) Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Ben Hutchings <bhutchings@solarflare.com> Cc: Vijay Subramanian <subramanian.vijay@gmail.com> Cc: Alexander Duyck <alexander.h.duyck@intel.com> Tested-by: Vijay Subramanian <subramanian.vijay@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-09-23 23:04:42 +00:00
page_address(pfrag->page) + pfrag->offset,
offset, copy, skb->len, skb) < 0)
goto error_efault;
pfrag->offset += copy;
skb_frag_size_add(&skb_shinfo(skb)->frags[i - 1], copy);
skb->len += copy;
skb->data_len += copy;
skb->truesize += copy;
refcount_add(copy, &sk->sk_wmem_alloc);
}
offset += copy;
length -= copy;
}
net: use a per task frag allocator We currently use a per socket order-0 page cache for tcp_sendmsg() operations. This page is used to build fragments for skbs. Its done to increase probability of coalescing small write() into single segments in skbs still in write queue (not yet sent) But it wastes a lot of memory for applications handling many mostly idle sockets, since each socket holds one page in sk->sk_sndmsg_page Its also quite inefficient to build TSO 64KB packets, because we need about 16 pages per skb on arches where PAGE_SIZE = 4096, so we hit page allocator more than wanted. This patch adds a per task frag allocator and uses bigger pages, if available. An automatic fallback is done in case of memory pressure. (up to 32768 bytes per frag, thats order-3 pages on x86) This increases TCP stream performance by 20% on loopback device, but also benefits on other network devices, since 8x less frags are mapped on transmit and unmapped on tx completion. Alexander Duyck mentioned a probable performance win on systems with IOMMU enabled. Its possible some SG enabled hardware cant cope with bigger fragments, but their ndo_start_xmit() should already handle this, splitting a fragment in sub fragments, since some arches have PAGE_SIZE=65536 Successfully tested on various ethernet devices. (ixgbe, igb, bnx2x, tg3, mellanox mlx4) Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Ben Hutchings <bhutchings@solarflare.com> Cc: Vijay Subramanian <subramanian.vijay@gmail.com> Cc: Alexander Duyck <alexander.h.duyck@intel.com> Tested-by: Vijay Subramanian <subramanian.vijay@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-09-23 23:04:42 +00:00
return 0;
net: use a per task frag allocator We currently use a per socket order-0 page cache for tcp_sendmsg() operations. This page is used to build fragments for skbs. Its done to increase probability of coalescing small write() into single segments in skbs still in write queue (not yet sent) But it wastes a lot of memory for applications handling many mostly idle sockets, since each socket holds one page in sk->sk_sndmsg_page Its also quite inefficient to build TSO 64KB packets, because we need about 16 pages per skb on arches where PAGE_SIZE = 4096, so we hit page allocator more than wanted. This patch adds a per task frag allocator and uses bigger pages, if available. An automatic fallback is done in case of memory pressure. (up to 32768 bytes per frag, thats order-3 pages on x86) This increases TCP stream performance by 20% on loopback device, but also benefits on other network devices, since 8x less frags are mapped on transmit and unmapped on tx completion. Alexander Duyck mentioned a probable performance win on systems with IOMMU enabled. Its possible some SG enabled hardware cant cope with bigger fragments, but their ndo_start_xmit() should already handle this, splitting a fragment in sub fragments, since some arches have PAGE_SIZE=65536 Successfully tested on various ethernet devices. (ixgbe, igb, bnx2x, tg3, mellanox mlx4) Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Ben Hutchings <bhutchings@solarflare.com> Cc: Vijay Subramanian <subramanian.vijay@gmail.com> Cc: Alexander Duyck <alexander.h.duyck@intel.com> Tested-by: Vijay Subramanian <subramanian.vijay@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-09-23 23:04:42 +00:00
error_efault:
err = -EFAULT;
error:
cork->length -= length;
IP6_INC_STATS(sock_net(sk), rt->rt6i_idev, IPSTATS_MIB_OUTDISCARDS);
return err;
}
int ip6_append_data(struct sock *sk,
int getfrag(void *from, char *to, int offset, int len,
int odd, struct sk_buff *skb),
void *from, int length, int transhdrlen,
struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
struct rt6_info *rt, unsigned int flags,
const struct sockcm_cookie *sockc)
{
struct inet_sock *inet = inet_sk(sk);
struct ipv6_pinfo *np = inet6_sk(sk);
int exthdrlen;
int err;
if (flags&MSG_PROBE)
return 0;
if (skb_queue_empty(&sk->sk_write_queue)) {
/*
* setup for corking
*/
err = ip6_setup_cork(sk, &inet->cork, &np->cork,
ipc6, rt, fl6);
if (err)
return err;
exthdrlen = (ipc6->opt ? ipc6->opt->opt_flen : 0);
length += exthdrlen;
transhdrlen += exthdrlen;
} else {
fl6 = &inet->cork.fl.u.ip6;
transhdrlen = 0;
}
return __ip6_append_data(sk, fl6, &sk->sk_write_queue, &inet->cork.base,
&np->cork, sk_page_frag(sk), getfrag,
from, length, transhdrlen, flags, ipc6, sockc);
}
EXPORT_SYMBOL_GPL(ip6_append_data);
static void ip6_cork_release(struct inet_cork_full *cork,
struct inet6_cork *v6_cork)
{
if (v6_cork->opt) {
kfree(v6_cork->opt->dst0opt);
kfree(v6_cork->opt->dst1opt);
kfree(v6_cork->opt->hopopt);
kfree(v6_cork->opt->srcrt);
kfree(v6_cork->opt);
v6_cork->opt = NULL;
}
if (cork->base.dst) {
dst_release(cork->base.dst);
cork->base.dst = NULL;
cork->base.flags &= ~IPCORK_ALLFRAG;
}
memset(&cork->fl, 0, sizeof(cork->fl));
}
struct sk_buff *__ip6_make_skb(struct sock *sk,
struct sk_buff_head *queue,
struct inet_cork_full *cork,
struct inet6_cork *v6_cork)
{
struct sk_buff *skb, *tmp_skb;
struct sk_buff **tail_skb;
struct in6_addr final_dst_buf, *final_dst = &final_dst_buf;
struct ipv6_pinfo *np = inet6_sk(sk);
struct net *net = sock_net(sk);
struct ipv6hdr *hdr;
struct ipv6_txoptions *opt = v6_cork->opt;
struct rt6_info *rt = (struct rt6_info *)cork->base.dst;
struct flowi6 *fl6 = &cork->fl.u.ip6;
unsigned char proto = fl6->flowi6_proto;
skb = __skb_dequeue(queue);
if (!skb)
goto out;
tail_skb = &(skb_shinfo(skb)->frag_list);
/* move skb->data to ip header from ext header */
if (skb->data < skb_network_header(skb))
__skb_pull(skb, skb_network_offset(skb));
while ((tmp_skb = __skb_dequeue(queue)) != NULL) {
__skb_pull(tmp_skb, skb_network_header_len(skb));
*tail_skb = tmp_skb;
tail_skb = &(tmp_skb->next);
skb->len += tmp_skb->len;
skb->data_len += tmp_skb->len;
skb->truesize += tmp_skb->truesize;
tmp_skb->destructor = NULL;
tmp_skb->sk = NULL;
}
/* Allow local fragmentation. */
skb->ignore_df = ip6_sk_ignore_df(sk);
*final_dst = fl6->daddr;
__skb_pull(skb, skb_network_header_len(skb));
if (opt && opt->opt_flen)
ipv6_push_frag_opts(skb, opt, &proto);
if (opt && opt->opt_nflen)
ipv6_push_nfrag_opts(skb, opt, &proto, &final_dst, &fl6->saddr);
skb_push(skb, sizeof(struct ipv6hdr));
skb_reset_network_header(skb);
hdr = ipv6_hdr(skb);
ip6_flow_hdr(hdr, v6_cork->tclass,
ipv6: Implement automatic flow label generation on transmit Automatically generate flow labels for IPv6 packets on transmit. The flow label is computed based on skb_get_hash. The flow label will only automatically be set when it is zero otherwise (i.e. flow label manager hasn't set one). This supports the transmit side functionality of RFC 6438. Added an IPv6 sysctl auto_flowlabels to enable/disable this behavior system wide, and added IPV6_AUTOFLOWLABEL socket option to enable this functionality per socket. By default, auto flowlabels are disabled to avoid possible conflicts with flow label manager, however if this feature proves useful we may want to enable it by default. It should also be noted that FreeBSD has already implemented automatic flow labels (including the sysctl and socket option). In FreeBSD, automatic flow labels default to enabled. Performance impact: Running super_netperf with 200 flows for TCP_RR and UDP_RR for IPv6. Note that in UDP case, __skb_get_hash will be called for every packet with explains slight regression. In the TCP case the hash is saved in the socket so there is no regression. Automatic flow labels disabled: TCP_RR: 86.53% CPU utilization 127/195/322 90/95/99% latencies 1.40498e+06 tps UDP_RR: 90.70% CPU utilization 118/168/243 90/95/99% latencies 1.50309e+06 tps Automatic flow labels enabled: TCP_RR: 85.90% CPU utilization 128/199/337 90/95/99% latencies 1.40051e+06 UDP_RR 92.61% CPU utilization 115/164/236 90/95/99% latencies 1.4687e+06 Signed-off-by: Tom Herbert <therbert@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-07-02 04:33:10 +00:00
ip6_make_flowlabel(net, skb, fl6->flowlabel,
np->autoflowlabel, fl6));
hdr->hop_limit = v6_cork->hop_limit;
hdr->nexthdr = proto;
hdr->saddr = fl6->saddr;
hdr->daddr = *final_dst;
skb->priority = sk->sk_priority;
skb->mark = sk->sk_mark;
skb_dst_set(skb, dst_clone(&rt->dst));
IP6_UPD_PO_STATS(net, rt->rt6i_idev, IPSTATS_MIB_OUT, skb->len);
if (proto == IPPROTO_ICMPV6) {
struct inet6_dev *idev = ip6_dst_idev(skb_dst(skb));
ICMP6MSGOUT_INC_STATS(net, idev, icmp6_hdr(skb)->icmp6_type);
ICMP6_INC_STATS(net, idev, ICMP6_MIB_OUTMSGS);
}
ip6_cork_release(cork, v6_cork);
out:
return skb;
}
int ip6_send_skb(struct sk_buff *skb)
{
struct net *net = sock_net(skb->sk);
struct rt6_info *rt = (struct rt6_info *)skb_dst(skb);
int err;
err = ip6_local_out(net, skb->sk, skb);
if (err) {
if (err > 0)
err = net_xmit_errno(err);
if (err)
IP6_INC_STATS(net, rt->rt6i_idev,
IPSTATS_MIB_OUTDISCARDS);
}
return err;
}
int ip6_push_pending_frames(struct sock *sk)
{
struct sk_buff *skb;
skb = ip6_finish_skb(sk);
if (!skb)
return 0;
return ip6_send_skb(skb);
}
EXPORT_SYMBOL_GPL(ip6_push_pending_frames);
static void __ip6_flush_pending_frames(struct sock *sk,
struct sk_buff_head *queue,
struct inet_cork_full *cork,
struct inet6_cork *v6_cork)
{
struct sk_buff *skb;
while ((skb = __skb_dequeue_tail(queue)) != NULL) {
if (skb_dst(skb))
IP6_INC_STATS(sock_net(sk), ip6_dst_idev(skb_dst(skb)),
IPSTATS_MIB_OUTDISCARDS);
kfree_skb(skb);
}
ip6_cork_release(cork, v6_cork);
}
void ip6_flush_pending_frames(struct sock *sk)
{
__ip6_flush_pending_frames(sk, &sk->sk_write_queue,
&inet_sk(sk)->cork, &inet6_sk(sk)->cork);
}
EXPORT_SYMBOL_GPL(ip6_flush_pending_frames);
struct sk_buff *ip6_make_skb(struct sock *sk,
int getfrag(void *from, char *to, int offset,
int len, int odd, struct sk_buff *skb),
void *from, int length, int transhdrlen,
struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
struct rt6_info *rt, unsigned int flags,
const struct sockcm_cookie *sockc)
{
struct inet_cork_full cork;
struct inet6_cork v6_cork;
struct sk_buff_head queue;
int exthdrlen = (ipc6->opt ? ipc6->opt->opt_flen : 0);
int err;
if (flags & MSG_PROBE)
return NULL;
__skb_queue_head_init(&queue);
cork.base.flags = 0;
cork.base.addr = 0;
cork.base.opt = NULL;
v6_cork.opt = NULL;
err = ip6_setup_cork(sk, &cork, &v6_cork, ipc6, rt, fl6);
if (err)
return ERR_PTR(err);
if (ipc6->dontfrag < 0)
ipc6->dontfrag = inet6_sk(sk)->dontfrag;
err = __ip6_append_data(sk, fl6, &queue, &cork.base, &v6_cork,
&current->task_frag, getfrag, from,
length + exthdrlen, transhdrlen + exthdrlen,
flags, ipc6, sockc);
if (err) {
__ip6_flush_pending_frames(sk, &queue, &cork, &v6_cork);
return ERR_PTR(err);
}
return __ip6_make_skb(sk, &queue, &cork, &v6_cork);
}