14641568562a0c509172e8072bc72ad1c0c571f6
[linux-2.6.git] / linux-2.6-520-vnet+.patch
1 diff -Nurb linux-2.6.22-510/include/linux/netfilter/xt_MARK.h linux-2.6.22-520/include/linux/netfilter/xt_MARK.h
2 --- linux-2.6.22-510/include/linux/netfilter/xt_MARK.h  2007-07-08 19:32:17.000000000 -0400
3 +++ linux-2.6.22-520/include/linux/netfilter/xt_MARK.h  2008-06-04 11:16:24.000000000 -0400
4 @@ -11,6 +11,7 @@
5         XT_MARK_SET=0,
6         XT_MARK_AND,
7         XT_MARK_OR,
8 +       XT_MARK_COPYXID,
9  };
10  
11  struct xt_mark_target_info_v1 {
12 diff -Nurb linux-2.6.22-510/include/linux/netfilter/xt_SETXID.h linux-2.6.22-520/include/linux/netfilter/xt_SETXID.h
13 --- linux-2.6.22-510/include/linux/netfilter/xt_SETXID.h        1969-12-31 19:00:00.000000000 -0500
14 +++ linux-2.6.22-520/include/linux/netfilter/xt_SETXID.h        2008-06-04 11:16:24.000000000 -0400
15 @@ -0,0 +1,14 @@
16 +#ifndef _XT_SETXID_H_target
17 +#define _XT_SETXID_H_target
18 +
19 +/* Version 1 */
20 +enum {
21 +       XT_SET_PACKET_XID=0
22 +};
23 +
24 +struct xt_setxid_target_info_v1 {
25 +       unsigned long mark;
26 +       u_int8_t mode;
27 +};
28 +
29 +#endif /*_XT_SETXID_H_target*/
30 diff -Nurb linux-2.6.22-510/include/linux/netfilter_ipv4/ipt_MARK.h linux-2.6.22-520/include/linux/netfilter_ipv4/ipt_MARK.h
31 --- linux-2.6.22-510/include/linux/netfilter_ipv4/ipt_MARK.h    2007-07-08 19:32:17.000000000 -0400
32 +++ linux-2.6.22-520/include/linux/netfilter_ipv4/ipt_MARK.h    2008-06-04 11:16:24.000000000 -0400
33 @@ -12,6 +12,7 @@
34  #define IPT_MARK_SET   XT_MARK_SET
35  #define IPT_MARK_AND   XT_MARK_AND
36  #define        IPT_MARK_OR     XT_MARK_OR
37 +#define IPT_MARK_COPYXID       XT_MARK_COPYXID
38  
39  #define ipt_mark_target_info_v1 xt_mark_target_info_v1
40  
41 diff -Nurb linux-2.6.22-510/include/linux/netfilter_ipv4/ipt_SETXID.h linux-2.6.22-520/include/linux/netfilter_ipv4/ipt_SETXID.h
42 --- linux-2.6.22-510/include/linux/netfilter_ipv4/ipt_SETXID.h  1969-12-31 19:00:00.000000000 -0500
43 +++ linux-2.6.22-520/include/linux/netfilter_ipv4/ipt_SETXID.h  2008-06-04 11:16:24.000000000 -0400
44 @@ -0,0 +1,13 @@
45 +#ifndef _IPT_SETXID_H_target
46 +#define _IPT_SETXID_H_target
47 +
48 +/* Backwards compatibility for old userspace */
49 +
50 +#include <linux/netfilter/xt_SETXID.h>
51 +
52 +/* Version 1 */
53 +#define IPT_SET_PACKET_XID     XT_SET_PACKET_XID
54 +
55 +#define ipt_setxid_target_info_v1 xt_setxid_target_info_v1
56 +
57 +#endif /*_IPT_SETXID_H_target*/
58 diff -Nurb linux-2.6.22-510/include/linux/skbuff.h linux-2.6.22-520/include/linux/skbuff.h
59 --- linux-2.6.22-510/include/linux/skbuff.h     2007-07-08 19:32:17.000000000 -0400
60 +++ linux-2.6.22-520/include/linux/skbuff.h     2008-06-04 11:16:24.000000000 -0400
61 @@ -302,6 +302,7 @@
62  #endif
63  
64         __u32                   mark;
65 +#define skb_tag                        mark
66  
67         sk_buff_data_t          transport_header;
68         sk_buff_data_t          network_header;
69 diff -Nurb linux-2.6.22-510/include/linux/socket.h linux-2.6.22-520/include/linux/socket.h
70 --- linux-2.6.22-510/include/linux/socket.h     2007-07-08 19:32:17.000000000 -0400
71 +++ linux-2.6.22-520/include/linux/socket.h     2008-06-04 11:16:24.000000000 -0400
72 @@ -288,6 +288,8 @@
73  #define SOL_TIPC       271
74  #define SOL_RXRPC      272
75  
76 +#define SO_SETXID      SO_PEERCRED
77 +
78  /* IPX options */
79  #define IPX_TYPE       1
80  
81 diff -Nurb linux-2.6.22-510/include/linux/vserver/network.h linux-2.6.22-520/include/linux/vserver/network.h
82 --- linux-2.6.22-510/include/linux/vserver/network.h    2008-06-04 11:16:20.000000000 -0400
83 +++ linux-2.6.22-520/include/linux/vserver/network.h    2008-06-04 11:16:24.000000000 -0400
84 @@ -47,6 +47,8 @@
85  #define NXC_TUN_CREATE         0x00000001
86  
87  #define NXC_RAW_ICMP           0x00000100
88 +#define NXC_RAW_SOCKET         0x00000200
89 +#define NXC_RAW_SEND           0x00000400
90  
91  
92  /* address types */
93 diff -Nurb linux-2.6.22-510/include/net/netfilter/nf_conntrack.h linux-2.6.22-520/include/net/netfilter/nf_conntrack.h
94 --- linux-2.6.22-510/include/net/netfilter/nf_conntrack.h       2007-07-08 19:32:17.000000000 -0400
95 +++ linux-2.6.22-520/include/net/netfilter/nf_conntrack.h       2008-06-04 11:16:24.000000000 -0400
96 @@ -131,6 +131,9 @@
97         /* Storage reserved for other modules: */
98         union nf_conntrack_proto proto;
99  
100 +       /* PLANETLAB. VNET-specific */
101 +       int xid[IP_CT_DIR_MAX];
102 +
103         /* features dynamically at the end: helper, nat (both optional) */
104         char data[0];
105  };
106 diff -Nurb linux-2.6.22-510/include/net/raw.h linux-2.6.22-520/include/net/raw.h
107 --- linux-2.6.22-510/include/net/raw.h  2007-07-08 19:32:17.000000000 -0400
108 +++ linux-2.6.22-520/include/net/raw.h  2008-06-04 11:16:24.000000000 -0400
109 @@ -36,7 +36,7 @@
110  
111  extern struct sock *__raw_v4_lookup(struct sock *sk, unsigned short num,
112                                     __be32 raddr, __be32 laddr,
113 -                                   int dif);
114 +                                   int dif, int tag);
115  
116  extern int raw_v4_input(struct sk_buff *skb, struct iphdr *iph, int hash);
117  
118 diff -Nurb linux-2.6.22-510/net/core/skbuff.c linux-2.6.22-520/net/core/skbuff.c
119 --- linux-2.6.22-510/net/core/skbuff.c  2007-07-08 19:32:17.000000000 -0400
120 +++ linux-2.6.22-520/net/core/skbuff.c  2008-06-04 11:16:24.000000000 -0400
121 @@ -56,6 +56,7 @@
122  #include <linux/rtnetlink.h>
123  #include <linux/init.h>
124  #include <linux/scatterlist.h>
125 +#include <linux/vs_network.h>
126  
127  #include <net/protocol.h>
128  #include <net/dst.h>
129 @@ -174,6 +175,7 @@
130         skb->data = data;
131         skb_reset_tail_pointer(skb);
132         skb->end = skb->tail + size;
133 +       if (!in_interrupt()) skb->skb_tag = nx_current_nid(); else skb->skb_tag = -1;
134         /* make sure we initialize shinfo sequentially */
135         shinfo = skb_shinfo(skb);
136         atomic_set(&shinfo->dataref, 1);
137 @@ -443,6 +445,8 @@
138         C(tail);
139         C(end);
140  
141 +       /* Sapan: Cloned skbs aren't owned by anyone. Let the cloner decide who it belongs to. */
142 +
143         atomic_inc(&(skb_shinfo(skb)->dataref));
144         skb->cloned = 1;
145  
146 @@ -492,6 +496,7 @@
147         new->tc_index   = old->tc_index;
148  #endif
149         skb_copy_secmark(new, old);
150 +       new->skb_tag = old->skb_tag;
151         atomic_set(&new->users, 1);
152         skb_shinfo(new)->gso_size = skb_shinfo(old)->gso_size;
153         skb_shinfo(new)->gso_segs = skb_shinfo(old)->gso_segs;
154 diff -Nurb linux-2.6.22-510/net/core/sock.c linux-2.6.22-520/net/core/sock.c
155 --- linux-2.6.22-510/net/core/sock.c    2008-06-04 11:16:21.000000000 -0400
156 +++ linux-2.6.22-520/net/core/sock.c    2008-06-04 11:16:24.000000000 -0400
157 @@ -444,6 +444,19 @@
158                 }
159                 goto set_sndbuf;
160  
161 +       case SO_SETXID:
162 +               if (current_vx_info()) {
163 +                       ret = -EPERM;
164 +                       break;
165 +               }
166 +               if (val < 0 || val > MAX_S_CONTEXT) {
167 +                       ret = -EINVAL;
168 +                       break;
169 +               }
170 +               sk->sk_xid = val;
171 +               sk->sk_nid = val;
172 +               break;
173 +
174         case SO_RCVBUF:
175                 /* Don't error on this BSD doesn't and if you think
176                    about it this is right. Otherwise apps have to
177 @@ -573,7 +586,7 @@
178                 char devname[IFNAMSIZ];
179  
180                 /* Sorry... */
181 -               if (!capable(CAP_NET_RAW)) {
182 +               if (!nx_capable(CAP_NET_RAW, NXC_RAW_SOCKET)) {
183                         ret = -EPERM;
184                         break;
185                 }
186 diff -Nurb linux-2.6.22-510/net/ipv4/af_inet.c linux-2.6.22-520/net/ipv4/af_inet.c
187 --- linux-2.6.22-510/net/ipv4/af_inet.c 2008-06-04 11:16:21.000000000 -0400
188 +++ linux-2.6.22-520/net/ipv4/af_inet.c 2008-06-04 11:16:24.000000000 -0400
189 @@ -178,6 +178,8 @@
190                         return -EAGAIN;
191                 }
192                 inet->sport = htons(inet->num);
193 +               sk->sk_xid = vx_current_xid();
194 +               if (!in_interrupt()) sk->sk_nid = nx_current_nid(); else sk->sk_nid=0;
195         }
196         release_sock(sk);
197         return 0;
198 @@ -312,6 +314,9 @@
199         if ((protocol == IPPROTO_ICMP) &&
200                 nx_capable(answer->capability, NXC_RAW_ICMP))
201                 goto override;
202 +       if (sock->type == SOCK_RAW &&
203 +               nx_capable(answer->capability, NXC_RAW_SOCKET))
204 +               goto override;
205         if (answer->capability > 0 && !capable(answer->capability))
206                 goto out_rcu_unlock;
207  override:
208 diff -Nurb linux-2.6.22-510/net/ipv4/icmp.c linux-2.6.22-520/net/ipv4/icmp.c
209 --- linux-2.6.22-510/net/ipv4/icmp.c    2008-06-04 11:16:23.000000000 -0400
210 +++ linux-2.6.22-520/net/ipv4/icmp.c    2008-06-04 11:16:24.000000000 -0400
211 @@ -709,7 +709,7 @@
212         if ((raw_sk = sk_head(&raw_v4_htable[hash])) != NULL) {
213                 while ((raw_sk = __raw_v4_lookup(raw_sk, protocol, iph->daddr,
214                                                  iph->saddr,
215 -                                                skb->dev->ifindex)) != NULL) {
216 +                                                skb->dev->ifindex, skb->skb_tag)) != NULL) {
217                         raw_err(raw_sk, skb, info);
218                         raw_sk = sk_next(raw_sk);
219                         iph = (struct iphdr *)skb->data;
220 diff -Nurb linux-2.6.22-510/net/ipv4/ip_options.c linux-2.6.22-520/net/ipv4/ip_options.c
221 --- linux-2.6.22-510/net/ipv4/ip_options.c      2007-07-08 19:32:17.000000000 -0400
222 +++ linux-2.6.22-520/net/ipv4/ip_options.c      2008-06-04 11:16:24.000000000 -0400
223 @@ -409,7 +409,7 @@
224                                         optptr[2] += 8;
225                                         break;
226                                       default:
227 -                                       if (!skb && !capable(CAP_NET_RAW)) {
228 +                                       if (!skb && !nx_capable(CAP_NET_RAW, NXC_RAW_SOCKET)) {
229                                                 pp_ptr = optptr + 3;
230                                                 goto error;
231                                         }
232 @@ -445,7 +445,7 @@
233                                 opt->router_alert = optptr - iph;
234                         break;
235                       case IPOPT_CIPSO:
236 -                       if ((!skb && !capable(CAP_NET_RAW)) || opt->cipso) {
237 +                       if ((!skb && !nx_capable(CAP_NET_RAW, NXC_RAW_SOCKET)) || opt->cipso) {
238                                 pp_ptr = optptr;
239                                 goto error;
240                         }
241 @@ -458,7 +458,7 @@
242                       case IPOPT_SEC:
243                       case IPOPT_SID:
244                       default:
245 -                       if (!skb && !capable(CAP_NET_RAW)) {
246 +                       if (!skb && !nx_capable(CAP_NET_RAW, NXC_RAW_SOCKET)) {
247                                 pp_ptr = optptr;
248                                 goto error;
249                         }
250 diff -Nurb linux-2.6.22-510/net/ipv4/netfilter/ipt_LOG.c linux-2.6.22-520/net/ipv4/netfilter/ipt_LOG.c
251 --- linux-2.6.22-510/net/ipv4/netfilter/ipt_LOG.c       2008-06-04 11:16:14.000000000 -0400
252 +++ linux-2.6.22-520/net/ipv4/netfilter/ipt_LOG.c       2008-06-04 11:16:24.000000000 -0400
253 @@ -49,6 +49,8 @@
254         else
255                 logflags = NF_LOG_MASK;
256  
257 +       printk("TAG=%d ", skb->skb_tag);
258 +
259         ih = skb_header_pointer(skb, iphoff, sizeof(_iph), &_iph);
260         if (ih == NULL) {
261                 printk("TRUNCATED");
262 diff -Nurb linux-2.6.22-510/net/ipv4/raw.c linux-2.6.22-520/net/ipv4/raw.c
263 --- linux-2.6.22-510/net/ipv4/raw.c     2008-06-04 11:16:21.000000000 -0400
264 +++ linux-2.6.22-520/net/ipv4/raw.c     2008-06-04 11:16:24.000000000 -0400
265 @@ -103,7 +103,7 @@
266  
267  struct sock *__raw_v4_lookup(struct sock *sk, unsigned short num,
268                              __be32 raddr, __be32 laddr,
269 -                            int dif)
270 +                            int dif, int tag)
271  {
272         struct hlist_node *node;
273  
274 @@ -112,6 +112,7 @@
275  
276                 if (inet->num == num                                    &&
277                     !(inet->daddr && inet->daddr != raddr)              &&
278 +                   (!sk->sk_nx_info || tag == 1 || sk->sk_nid == tag)  &&
279                     v4_sock_addr_match(sk->sk_nx_info, inet, laddr)     &&
280                     !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
281                         goto found; /* gotcha */
282 @@ -161,7 +162,7 @@
283                 goto out;
284         sk = __raw_v4_lookup(__sk_head(head), iph->protocol,
285                              iph->saddr, iph->daddr,
286 -                            skb->dev->ifindex);
287 +                            skb->dev->ifindex, skb->skb_tag);
288  
289         while (sk) {
290                 delivered = 1;
291 @@ -174,7 +175,7 @@
292                 }
293                 sk = __raw_v4_lookup(sk_next(sk), iph->protocol,
294                                      iph->saddr, iph->daddr,
295 -                                    skb->dev->ifindex);
296 +                                    skb->dev->ifindex, skb->skb_tag);
297         }
298  out:
299         read_unlock(&raw_v4_lock);
300 @@ -315,7 +316,7 @@
301         }
302  
303         err = -EPERM;
304 -       if (!nx_check(0, VS_ADMIN) && !capable(CAP_NET_RAW) &&
305 +       if (!nx_check(0, VS_ADMIN) && !nx_capable(CAP_NET_RAW, NXC_RAW_SOCKET) &&
306                 sk->sk_nx_info &&
307                 !v4_addr_in_nx_info(sk->sk_nx_info, iph->saddr, NXA_MASK_BIND))
308                 goto error_free;
309 diff -Nurb linux-2.6.22-510/net/netfilter/Kconfig linux-2.6.22-520/net/netfilter/Kconfig
310 --- linux-2.6.22-510/net/netfilter/Kconfig      2007-07-08 19:32:17.000000000 -0400
311 +++ linux-2.6.22-520/net/netfilter/Kconfig      2008-06-04 11:16:24.000000000 -0400
312 @@ -389,6 +389,13 @@
313  
314           To compile it as a module, choose M here.  If unsure, say N.
315  
316 +config NETFILTER_XT_TARGET_SETXID
317 +       tristate '"SETXID" target support'
318 +       depends on NETFILTER_XTABLES
319 +       help
320 +         This option adds a `SETXID' target, which allows you to alter the
321 +         xid of a socket.
322 +
323  config NETFILTER_XT_MATCH_COMMENT
324         tristate  '"comment" match support'
325         depends on NETFILTER_XTABLES
326 diff -Nurb linux-2.6.22-510/net/netfilter/Makefile linux-2.6.22-520/net/netfilter/Makefile
327 --- linux-2.6.22-510/net/netfilter/Makefile     2007-07-08 19:32:17.000000000 -0400
328 +++ linux-2.6.22-520/net/netfilter/Makefile     2008-06-04 11:16:24.000000000 -0400
329 @@ -37,6 +37,7 @@
330  obj-$(CONFIG_NETFILTER_XTABLES) += x_tables.o xt_tcpudp.o
331  
332  # targets
333 +obj-$(CONFIG_NETFILTER_XT_TARGET_SETXID) += xt_SETXID.o
334  obj-$(CONFIG_NETFILTER_XT_TARGET_CLASSIFY) += xt_CLASSIFY.o
335  obj-$(CONFIG_NETFILTER_XT_TARGET_CONNMARK) += xt_CONNMARK.o
336  obj-$(CONFIG_NETFILTER_XT_TARGET_DSCP) += xt_DSCP.o
337 diff -Nurb linux-2.6.22-510/net/netfilter/nf_conntrack_core.c linux-2.6.22-520/net/netfilter/nf_conntrack_core.c
338 --- linux-2.6.22-510/net/netfilter/nf_conntrack_core.c  2007-07-08 19:32:17.000000000 -0400
339 +++ linux-2.6.22-520/net/netfilter/nf_conntrack_core.c  2008-06-04 11:16:24.000000000 -0400
340 @@ -726,6 +726,8 @@
341  
342         /* Overload tuple linked list to put us in unconfirmed list. */
343         list_add(&conntrack->tuplehash[IP_CT_DIR_ORIGINAL].list, &unconfirmed);
344 +       conntrack->xid[IP_CT_DIR_ORIGINAL] = -1;
345 +       conntrack->xid[IP_CT_DIR_REPLY] = -1;
346  
347         write_unlock_bh(&nf_conntrack_lock);
348  
349 diff -Nurb linux-2.6.22-510/net/netfilter/xt_MARK.c linux-2.6.22-520/net/netfilter/xt_MARK.c
350 --- linux-2.6.22-510/net/netfilter/xt_MARK.c    2007-07-08 19:32:17.000000000 -0400
351 +++ linux-2.6.22-520/net/netfilter/xt_MARK.c    2008-06-04 11:19:54.000000000 -0400
352 @@ -5,13 +5,18 @@
353   * This program is free software; you can redistribute it and/or modify
354   * it under the terms of the GNU General Public License version 2 as
355   * published by the Free Software Foundation.
356 + *
357   */
358  
359  #include <linux/module.h>
360 +#include <linux/version.h>
361  #include <linux/skbuff.h>
362  #include <linux/ip.h>
363  #include <net/checksum.h>
364 +#include <net/route.h>
365 +#include <net/inet_hashtables.h>
366  
367 +#include <net/netfilter/nf_conntrack.h>
368  #include <linux/netfilter/x_tables.h>
369  #include <linux/netfilter/xt_MARK.h>
370  
371 @@ -21,6 +26,48 @@
372  MODULE_ALIAS("ipt_MARK");
373  MODULE_ALIAS("ip6t_MARK");
374  
375 +static inline u_int16_t
376 +get_dst_port(struct nf_conntrack_tuple *tuple)
377 +{
378 +       switch (tuple->dst.protonum) {
379 +       case IPPROTO_GRE:
380 +               /* XXX Truncate 32-bit GRE key to 16 bits */
381 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,11)
382 +               return tuple->dst.u.gre.key;
383 +#else
384 +               return htons(ntohl(tuple->dst.u.gre.key));
385 +#endif  
386 +       case IPPROTO_ICMP:
387 +               /* Bind on ICMP echo ID */
388 +               return tuple->src.u.icmp.id;
389 +       case IPPROTO_TCP:
390 +               return tuple->dst.u.tcp.port;
391 +       case IPPROTO_UDP:
392 +               return tuple->dst.u.udp.port;
393 +       default:
394 +               return tuple->dst.u.all;
395 +       }
396 +}
397 +
398 +static inline u_int16_t
399 +get_src_port(struct nf_conntrack_tuple *tuple)
400 +{
401 +       switch (tuple->dst.protonum) {
402 +       case IPPROTO_GRE:
403 +               /* XXX Truncate 32-bit GRE key to 16 bits */
404 +               return htons(ntohl(tuple->src.u.gre.key));
405 +       case IPPROTO_ICMP:
406 +               /* Bind on ICMP echo ID */
407 +               return tuple->src.u.icmp.id;
408 +       case IPPROTO_TCP:
409 +               return tuple->src.u.tcp.port;
410 +       case IPPROTO_UDP:
411 +               return tuple->src.u.udp.port;
412 +       default:
413 +               return tuple->src.u.all;
414 +       }
415 +}
416 +
417  static unsigned int
418  target_v0(struct sk_buff **pskb,
419           const struct net_device *in,
420 @@ -35,6 +82,8 @@
421         return XT_CONTINUE;
422  }
423  
424 +extern DEFINE_PER_CPU(int, sknid_elevator);
425 +
426  static unsigned int
427  target_v1(struct sk_buff **pskb,
428           const struct net_device *in,
429 @@ -44,7 +93,20 @@
430           const void *targinfo)
431  {
432         const struct xt_mark_target_info_v1 *markinfo = targinfo;
433 -       int mark = 0;
434 +       enum ip_conntrack_info ctinfo;
435 +      struct sock *connection_sk;
436 +      int dif;
437 +      struct nf_conn *ct;
438 +      extern struct inet_hashinfo tcp_hashinfo;
439 +      enum ip_conntrack_dir dir;
440 +      int *curtag;
441 +      u_int32_t src_ip;
442 +      u_int32_t dst_ip;
443 +      u_int16_t proto, src_port;
444 +      u_int32_t ip;
445 +      u_int16_t port;
446 +
447 +       int mark = -1;
448  
449         switch (markinfo->mode) {
450         case XT_MARK_SET:
451 @@ -58,13 +120,76 @@
452         case XT_MARK_OR:
453                 mark = (*pskb)->mark | markinfo->mark;
454                 break;
455 +
456 +               case XT_MARK_COPYXID: 
457 +                                             
458 +                                             ct = nf_ct_get((*pskb), &ctinfo);
459 +                                             if (!ct) 
460 +                                                     break;
461 +
462 +                                             dir = CTINFO2DIR(ctinfo);
463 +                                             src_ip = ct->tuplehash[dir].tuple.src.u3.ip;
464 +                                             dst_ip = ct->tuplehash[dir].tuple.dst.u3.ip;
465 +                                             src_port = get_src_port(&ct->tuplehash[dir].tuple);
466 +                                             proto = ct->tuplehash[dir].tuple.dst.protonum;
467 +
468 +                                             dif = ((struct rtable *)(*pskb)->dst)->rt_iif;
469 +                                             ip = ct->tuplehash[dir].tuple.dst.u3.ip;
470 +                                             port = get_dst_port(&ct->tuplehash[dir].tuple);
471 +
472 +                                             if (proto == 1 || proto == 17) {
473 +                                                     if ((*pskb)->mark>0) /* The packet is marked, it's going out */
474 +                                                     {
475 +                                                             if (ct->xid[0]>0 && ct->xid[0]!=(*pskb)->mark && hooknum==NF_IP_LOCAL_IN) {
476 +                                                                     /* Neil Spring's bug. */
477 +                                                                       break;
478 +                                                             }
479 +                                                             else
480 +                                                                     ct->xid[0]=(*pskb)->mark;
481 +                                                     }
482 +
483 +                                                     if (ct->xid[0] > 0) {
484 +                                                             mark = ct->xid[0];
485         }
486  
487 +                                             }
488 +                                             else if (proto == 6) { 
489 +                                                     if ((*pskb)->sk) {
490 +                                                             connection_sk = (*pskb)->sk;
491 +                                                             sock_hold(connection_sk);
492 +                                                     }
493 +                                                     else 
494 +                                                             connection_sk = inet_lookup_established(&tcp_hashinfo, src_ip, src_port, ip, port, dif);
495 +                                                             
496 +
497 +                                                     if (connection_sk) {
498 +                                                             if (connection_sk->sk_state == TCP_TIME_WAIT) {
499 +                                                                     inet_twsk_put(inet_twsk(connection_sk));
500 +                                                                     break;
501 +                                                             }
502 +                                                             connection_sk->sk_peercred.gid = connection_sk->sk_peercred.uid = ct->xid[dir];
503 +                                                             ct->xid[!dir]=connection_sk->sk_nid;
504 +                                                             if (connection_sk->sk_nid != 0) 
505 +                                                                     mark = connection_sk->sk_nid;
506 +                                                             sock_put(connection_sk);
507 +                                                     }
508 +                                                     else 
509 +                                                             mark = -1 ; 
510 +                                             }
511 +                                             break;
512 +       }
513 +       if (mark != -1) {
514         (*pskb)->mark = mark;
515 +       }
516 +
517 +       curtag=&__get_cpu_var(sknid_elevator);
518 +       if (mark > 0 && *curtag==-2) 
519 +       {
520 +               *curtag = mark;
521 +       }
522         return XT_CONTINUE;
523  }
524  
525 -
526  static int
527  checkentry_v0(const char *tablename,
528               const void *entry,
529 @@ -92,7 +217,8 @@
530  
531         if (markinfo->mode != XT_MARK_SET
532             && markinfo->mode != XT_MARK_AND
533 -           && markinfo->mode != XT_MARK_OR) {
534 +           && markinfo->mode != XT_MARK_OR
535 +           && markinfo->mode != XT_MARK_COPYXID) {
536                 printk(KERN_WARNING "MARK: unknown mode %u\n",
537                        markinfo->mode);
538                 return 0;
539 diff -Nurb linux-2.6.22-510/net/netfilter/xt_SETXID.c linux-2.6.22-520/net/netfilter/xt_SETXID.c
540 --- linux-2.6.22-510/net/netfilter/xt_SETXID.c  1969-12-31 19:00:00.000000000 -0500
541 +++ linux-2.6.22-520/net/netfilter/xt_SETXID.c  2008-06-04 11:16:24.000000000 -0400
542 @@ -0,0 +1,79 @@
543 +#include <linux/module.h>
544 +#include <linux/skbuff.h>
545 +#include <linux/ip.h>
546 +#include <net/checksum.h>
547 +#include <linux/vs_network.h>
548 +
549 +#include <linux/netfilter/x_tables.h>
550 +#include <linux/netfilter/xt_SETXID.h>
551 +
552 +MODULE_LICENSE("GPL");
553 +MODULE_AUTHOR("");
554 +MODULE_DESCRIPTION("");
555 +MODULE_ALIAS("ipt_SETXID");
556 +
557 +static unsigned int
558 +target_v1(struct sk_buff **pskb,
559 +         const struct net_device *in,
560 +         const struct net_device *out,
561 +         unsigned int hooknum,
562 +         const struct xt_target *target,
563 +         const void *targinfo)
564 +{
565 +       const struct xt_setxid_target_info_v1 *setxidinfo = targinfo;
566 +
567 +       switch (setxidinfo->mode) {
568 +       case XT_SET_PACKET_XID:
569 +                (*pskb)->skb_tag = setxidinfo->mark;
570 +               break;
571 +       }
572 +       return XT_CONTINUE;
573 +}
574 +
575 +
576 +static int
577 +checkentry_v1(const char *tablename,
578 +             const void *entry,
579 +             const struct xt_target *target,
580 +             void *targinfo,
581 +             unsigned int hook_mask)
582 +{
583 +       struct xt_setxid_target_info_v1 *setxidinfo = targinfo;
584 +
585 +       if (setxidinfo->mode != XT_SET_PACKET_XID) {
586 +               printk(KERN_WARNING "SETXID: unknown mode %u\n",
587 +                      setxidinfo->mode);
588 +               return 0;
589 +       }
590 +
591 +       return 1;
592 +}
593 +
594 +static struct xt_target xt_setxid_target[] = {
595 +       {
596 +               .name           = "SETXID",
597 +               .family         = AF_INET,
598 +               .revision       = 1,
599 +               .checkentry     = checkentry_v1,
600 +               .target         = target_v1,
601 +               .targetsize     = sizeof(struct xt_setxid_target_info_v1),
602 +               .table          = "mangle",
603 +               .me             = THIS_MODULE,
604 +       }
605 +};
606 +
607 +static int __init init(void)
608 +{
609 +       int err;
610 +
611 +       err = xt_register_targets(xt_setxid_target, ARRAY_SIZE(xt_setxid_target));
612 +       return err;
613 +}
614 +
615 +static void __exit fini(void)
616 +{
617 +       xt_unregister_targets(xt_setxid_target, ARRAY_SIZE(xt_setxid_target));
618 +}
619 +
620 +module_init(init);
621 +module_exit(fini);
622 diff -Nurb linux-2.6.22-510/net/packet/af_packet.c linux-2.6.22-520/net/packet/af_packet.c
623 --- linux-2.6.22-510/net/packet/af_packet.c     2007-07-08 19:32:17.000000000 -0400
624 +++ linux-2.6.22-520/net/packet/af_packet.c     2008-06-04 11:16:24.000000000 -0400
625 @@ -78,6 +78,7 @@
626  #include <linux/poll.h>
627  #include <linux/module.h>
628  #include <linux/init.h>
629 +#include <linux/vs_network.h>
630  
631  #ifdef CONFIG_INET
632  #include <net/inet_common.h>
633 @@ -324,6 +325,9 @@
634         __be16 proto=0;
635         int err;
636  
637 +       if (!nx_capable(CAP_NET_RAW, NXC_RAW_SEND))
638 +               return -EPERM;
639 +
640         /*
641          *      Get and verify the address.
642          */
643 @@ -416,10 +420,22 @@
644         return err;
645  }
646  
647 +extern DEFINE_PER_CPU(int, sknid_elevator);
648  static inline unsigned int run_filter(struct sk_buff *skb, struct sock *sk,
649                                       unsigned int res)
650  {
651         struct sk_filter *filter;
652 +       int tag = skb->skb_tag;
653 +       int *elevator=&__get_cpu_var(sknid_elevator);
654 +
655 +       if (sk->sk_nx_info && !(tag == 1 || sk->sk_nid == tag)) {
656 +               *elevator=-2;
657 +               return 0;
658 +       }
659 +       else if (!sk->sk_nx_info && *elevator) {
660 +               /* Root has already seen this packet */
661 +               return 0;
662 +       }
663  
664         rcu_read_lock_bh();
665         filter = rcu_dereference(sk->sk_filter);
666 @@ -711,6 +727,9 @@
667         unsigned char *addr;
668         int ifindex, err, reserve = 0;
669  
670 +       if (!nx_capable(CAP_NET_RAW, NXC_RAW_SEND)) 
671 +               return -EPERM;
672 +
673         /*
674          *      Get and verify the address.
675          */
676 @@ -984,7 +1003,7 @@
677         __be16 proto = (__force __be16)protocol; /* weird, but documented */
678         int err;
679  
680 -       if (!capable(CAP_NET_RAW))
681 +       if (!nx_capable(CAP_NET_RAW, NXC_RAW_SOCKET))
682                 return -EPERM;
683         if (sock->type != SOCK_DGRAM && sock->type != SOCK_RAW &&
684             sock->type != SOCK_PACKET)