Doug / smarc-fsl-linux-kernel

Download as
Browse Code ยป

Commit 1252ecf63f77ea147bd40f5462c7d9e3d3ae2815

Authored by Arjan van de Ven
Committed by David S. Miller
1 parent 00181fc946
Exists in master and in 7 other branches imx_3.0.35_4.1.0, imx_3.10.17_1.0.1_ga, imx_3.10.53_1.1.0_ga, imx_3.14.28_1.0.0_ga, smarc-imx_3.10.53_1.1.0_ga, smarc-imx_3.14.28_1.0.0_ga, smarc-l5.0.0_1.0.0-ga

[ATM]: fix possible recursive locking in skb_migrate() 

ok this is a real potential deadlock in a way, it takes two locks of 2
skbuffs without doing any kind of lock ordering; I think the following
patch should fix it. Just sort the lock taking order by address of the
skb.. it's not pretty but it's the best this can do in a minimally
invasive way.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Chas Williams <chas@cmf.nrl.navy.mil>
Signed-off-by: David S. Miller <davem@davemloft.net>

Showing 1 changed file with 11 additions and 6 deletions Side-by-side Diff

... ... @@ -25,22 +25,27 @@
25 25 /*
26 26 * skb_migrate appends the list at "from" to "to", emptying "from" in the
27 27 * process. skb_migrate is atomic with respect to all other skb operations on
28   - * "from" and "to". Note that it locks both lists at the same time, so beware
29   - * of potential deadlocks.
  28 + * "from" and "to". Note that it locks both lists at the same time, so to deal
  29 + * with the lock ordering, the locks are taken in address order.
30 30 *
31 31 * This function should live in skbuff.c or skbuff.h.
32 32 */
33 33  
34 34  
35   -void skb_migrate(struct sk_buff_head *from,struct sk_buff_head *to)
  35 +void skb_migrate(struct sk_buff_head *from, struct sk_buff_head *to)
36 36 {
37 37 unsigned long flags;
38 38 struct sk_buff *skb_from = (struct sk_buff *) from;
39 39 struct sk_buff *skb_to = (struct sk_buff *) to;
40 40 struct sk_buff *prev;
41 41  
42   - spin_lock_irqsave(&from->lock,flags);
43   - spin_lock(&to->lock);
  42 + if ((unsigned long) from < (unsigned long) to) {
  43 + spin_lock_irqsave(&from->lock, flags);
  44 + spin_lock_nested(&to->lock, SINGLE_DEPTH_NESTING);
  45 + } else {
  46 + spin_lock_irqsave(&to->lock, flags);
  47 + spin_lock_nested(&from->lock, SINGLE_DEPTH_NESTING);
  48 + }
44 49 prev = from->prev;
45 50 from->next->prev = to->prev;
46 51 prev->next = skb_to;
... ... @@ -51,7 +56,7 @@
51 56 from->prev = skb_from;
52 57 from->next = skb_from;
53 58 from->qlen = 0;
54   - spin_unlock_irqrestore(&from->lock,flags);
  59 + spin_unlock_irqrestore(&from->lock, flags);
55 60 }
56 61  
57 62