Commit fdc0bde90a689b9145f2b6f271c03f4c99d09667

Authored by Denis V. Lunev
Committed by David S. Miller
1 parent f6e0b239a2

icmp: icmp_sk() should not use smp_processor_id() in preemptible code

Pass namespace into icmp_xmit_lock, obtain socket inside and return
it as a result for caller.

Thanks Alexey Dobryan for this report:

Steps to reproduce:

	CONFIG_PREEMPT=y
	CONFIG_DEBUG_PREEMPT=y
	tracepath <something>

BUG: using smp_processor_id() in preemptible [00000000] code: tracepath/3205
caller is icmp_sk+0x15/0x30
Pid: 3205, comm: tracepath Not tainted 2.6.27-rc4 #1

Call Trace:
 [<ffffffff8031af14>] debug_smp_processor_id+0xe4/0xf0
 [<ffffffff80409405>] icmp_sk+0x15/0x30
 [<ffffffff8040a17b>] icmp_send+0x4b/0x3f0
 [<ffffffff8025a415>] ? trace_hardirqs_on_caller+0xd5/0x160
 [<ffffffff8025a4ad>] ? trace_hardirqs_on+0xd/0x10
 [<ffffffff8023a475>] ? local_bh_enable_ip+0x95/0x110
 [<ffffffff804285b9>] ? _spin_unlock_bh+0x39/0x40
 [<ffffffff8025a26c>] ? mark_held_locks+0x4c/0x90
 [<ffffffff8025a4ad>] ? trace_hardirqs_on+0xd/0x10
 [<ffffffff8025a415>] ? trace_hardirqs_on_caller+0xd5/0x160
 [<ffffffff803e91b4>] ip_fragment+0x8d4/0x900
 [<ffffffff803e7030>] ? ip_finish_output2+0x0/0x290
 [<ffffffff803e91e0>] ? ip_finish_output+0x0/0x60
 [<ffffffff803e6650>] ? dst_output+0x0/0x10
 [<ffffffff803e922c>] ip_finish_output+0x4c/0x60
 [<ffffffff803e92e3>] ip_output+0xa3/0xf0
 [<ffffffff803e68d0>] ip_local_out+0x20/0x30
 [<ffffffff803e753f>] ip_push_pending_frames+0x27f/0x400
 [<ffffffff80406313>] udp_push_pending_frames+0x233/0x3d0
 [<ffffffff804067d1>] udp_sendmsg+0x321/0x6f0
 [<ffffffff8040d155>] inet_sendmsg+0x45/0x80
 [<ffffffff803b967f>] sock_sendmsg+0xdf/0x110
 [<ffffffff8024a100>] ? autoremove_wake_function+0x0/0x40
 [<ffffffff80257ce5>] ? validate_chain+0x415/0x1010
 [<ffffffff8027dc10>] ? __do_fault+0x140/0x450
 [<ffffffff802597d0>] ? __lock_acquire+0x260/0x590
 [<ffffffff803b9e55>] ? sockfd_lookup_light+0x45/0x80
 [<ffffffff803ba50a>] sys_sendto+0xea/0x120
 [<ffffffff80428e42>] ? _spin_unlock_irqrestore+0x42/0x80
 [<ffffffff803134bc>] ? __up_read+0x4c/0xb0
 [<ffffffff8024e0c6>] ? up_read+0x26/0x30
 [<ffffffff8020b8bb>] system_call_fastpath+0x16/0x1b

icmp6_sk() is similar.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Showing 2 changed files with 26 additions and 19 deletions Side-by-side Diff

... ... @@ -204,18 +204,22 @@
204 204 return net->ipv4.icmp_sk[smp_processor_id()];
205 205 }
206 206  
207   -static inline int icmp_xmit_lock(struct sock *sk)
  207 +static inline struct sock *icmp_xmit_lock(struct net *net)
208 208 {
  209 + struct sock *sk;
  210 +
209 211 local_bh_disable();
210 212  
  213 + sk = icmp_sk(net);
  214 +
211 215 if (unlikely(!spin_trylock(&sk->sk_lock.slock))) {
212 216 /* This can happen if the output path signals a
213 217 * dst_link_failure() for an outgoing ICMP packet.
214 218 */
215 219 local_bh_enable();
216   - return 1;
  220 + return NULL;
217 221 }
218   - return 0;
  222 + return sk;
219 223 }
220 224  
221 225 static inline void icmp_xmit_unlock(struct sock *sk)
222 226  
223 227  
... ... @@ -354,15 +358,17 @@
354 358 struct ipcm_cookie ipc;
355 359 struct rtable *rt = skb->rtable;
356 360 struct net *net = dev_net(rt->u.dst.dev);
357   - struct sock *sk = icmp_sk(net);
358   - struct inet_sock *inet = inet_sk(sk);
  361 + struct sock *sk;
  362 + struct inet_sock *inet;
359 363 __be32 daddr;
360 364  
361 365 if (ip_options_echo(&icmp_param->replyopts, skb))
362 366 return;
363 367  
364   - if (icmp_xmit_lock(sk))
  368 + sk = icmp_xmit_lock(net);
  369 + if (sk == NULL)
365 370 return;
  371 + inet = inet_sk(sk);
366 372  
367 373 icmp_param->data.icmph.checksum = 0;
368 374  
... ... @@ -419,7 +425,6 @@
419 425 if (!rt)
420 426 goto out;
421 427 net = dev_net(rt->u.dst.dev);
422   - sk = icmp_sk(net);
423 428  
424 429 /*
425 430 * Find the original header. It is expected to be valid, of course.
... ... @@ -483,7 +488,8 @@
483 488 }
484 489 }
485 490  
486   - if (icmp_xmit_lock(sk))
  491 + sk = icmp_xmit_lock(net);
  492 + if (sk == NULL)
487 493 return;
488 494  
489 495 /*
... ... @@ -91,19 +91,22 @@
91 91 .flags = INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL,
92 92 };
93 93  
94   -static __inline__ int icmpv6_xmit_lock(struct sock *sk)
  94 +static __inline__ struct sock *icmpv6_xmit_lock(struct net *net)
95 95 {
  96 + struct sock *sk;
  97 +
96 98 local_bh_disable();
97 99  
  100 + sk = icmpv6_sk(net);
98 101 if (unlikely(!spin_trylock(&sk->sk_lock.slock))) {
99 102 /* This can happen if the output path (f.e. SIT or
100 103 * ip6ip6 tunnel) signals dst_link_failure() for an
101 104 * outgoing ICMP6 packet.
102 105 */
103 106 local_bh_enable();
104   - return 1;
  107 + return NULL;
105 108 }
106   - return 0;
  109 + return sk;
107 110 }
108 111  
109 112 static __inline__ void icmpv6_xmit_unlock(struct sock *sk)
110 113  
... ... @@ -392,12 +395,11 @@
392 395 fl.fl_icmp_code = code;
393 396 security_skb_classify_flow(skb, &fl);
394 397  
395   - sk = icmpv6_sk(net);
  398 + sk = icmpv6_xmit_lock(net);
  399 + if (sk == NULL)
  400 + return;
396 401 np = inet6_sk(sk);
397 402  
398   - if (icmpv6_xmit_lock(sk))
399   - return;
400   -
401 403 if (!icmpv6_xrlim_allow(sk, type, &fl))
402 404 goto out;
403 405  
404 406  
... ... @@ -539,11 +541,10 @@
539 541 fl.fl_icmp_type = ICMPV6_ECHO_REPLY;
540 542 security_skb_classify_flow(skb, &fl);
541 543  
542   - sk = icmpv6_sk(net);
543   - np = inet6_sk(sk);
544   -
545   - if (icmpv6_xmit_lock(sk))
  544 + sk = icmpv6_xmit_lock(net);
  545 + if (sk == NULL)
546 546 return;
  547 + np = inet6_sk(sk);
547 548  
548 549 if (!fl.oif && ipv6_addr_is_multicast(&fl.fl6_dst))
549 550 fl.oif = np->mcast_oif;