Commit 55c0022e authored by David S. Miller's avatar David S. Miller

[IPV4] ip_fragment: Always compute hash with ipfrag_lock held.

Otherwise we could compute an inaccurate hash due to the
random seed changing.

Noticed by Zach Brown and patch is based upon some feedback
from Herbert Xu.
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parent 19910d1a
...@@ -304,13 +304,17 @@ out: ...@@ -304,13 +304,17 @@ out:
/* Creation primitives. */ /* Creation primitives. */
static struct ipq *ip_frag_intern(unsigned int hash, struct ipq *qp_in) static struct ipq *ip_frag_intern(struct ipq *qp_in)
{ {
struct ipq *qp; struct ipq *qp;
#ifdef CONFIG_SMP #ifdef CONFIG_SMP
struct hlist_node *n; struct hlist_node *n;
#endif #endif
unsigned int hash;
write_lock(&ipfrag_lock); write_lock(&ipfrag_lock);
hash = ipqhashfn(qp_in->id, qp_in->saddr, qp_in->daddr,
qp_in->protocol);
#ifdef CONFIG_SMP #ifdef CONFIG_SMP
/* With SMP race we have to recheck hash table, because /* With SMP race we have to recheck hash table, because
* such entry could be created on other cpu, while we * such entry could be created on other cpu, while we
...@@ -345,7 +349,7 @@ static struct ipq *ip_frag_intern(unsigned int hash, struct ipq *qp_in) ...@@ -345,7 +349,7 @@ static struct ipq *ip_frag_intern(unsigned int hash, struct ipq *qp_in)
} }
/* Add an entry to the 'ipq' queue for a newly received IP datagram. */ /* Add an entry to the 'ipq' queue for a newly received IP datagram. */
static struct ipq *ip_frag_create(unsigned hash, struct iphdr *iph, u32 user) static struct ipq *ip_frag_create(struct iphdr *iph, u32 user)
{ {
struct ipq *qp; struct ipq *qp;
...@@ -371,7 +375,7 @@ static struct ipq *ip_frag_create(unsigned hash, struct iphdr *iph, u32 user) ...@@ -371,7 +375,7 @@ static struct ipq *ip_frag_create(unsigned hash, struct iphdr *iph, u32 user)
spin_lock_init(&qp->lock); spin_lock_init(&qp->lock);
atomic_set(&qp->refcnt, 1); atomic_set(&qp->refcnt, 1);
return ip_frag_intern(hash, qp); return ip_frag_intern(qp);
out_nomem: out_nomem:
LIMIT_NETDEBUG(KERN_ERR "ip_frag_create: no memory left !\n"); LIMIT_NETDEBUG(KERN_ERR "ip_frag_create: no memory left !\n");
...@@ -387,11 +391,12 @@ static inline struct ipq *ip_find(struct iphdr *iph, u32 user) ...@@ -387,11 +391,12 @@ static inline struct ipq *ip_find(struct iphdr *iph, u32 user)
__u32 saddr = iph->saddr; __u32 saddr = iph->saddr;
__u32 daddr = iph->daddr; __u32 daddr = iph->daddr;
__u8 protocol = iph->protocol; __u8 protocol = iph->protocol;
unsigned int hash = ipqhashfn(id, saddr, daddr, protocol); unsigned int hash;
struct ipq *qp; struct ipq *qp;
struct hlist_node *n; struct hlist_node *n;
read_lock(&ipfrag_lock); read_lock(&ipfrag_lock);
hash = ipqhashfn(id, saddr, daddr, protocol);
hlist_for_each_entry(qp, n, &ipq_hash[hash], list) { hlist_for_each_entry(qp, n, &ipq_hash[hash], list) {
if(qp->id == id && if(qp->id == id &&
qp->saddr == saddr && qp->saddr == saddr &&
...@@ -405,7 +410,7 @@ static inline struct ipq *ip_find(struct iphdr *iph, u32 user) ...@@ -405,7 +410,7 @@ static inline struct ipq *ip_find(struct iphdr *iph, u32 user)
} }
read_unlock(&ipfrag_lock); read_unlock(&ipfrag_lock);
return ip_frag_create(hash, iph, user); return ip_frag_create(iph, user);
} }
/* Is the fragment too far ahead to be part of ipq? */ /* Is the fragment too far ahead to be part of ipq? */
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment