There are many, many, many severe SMP locking bugs in the ipv6
advanced routing selection code. I wish the author had thought about
these issues more carefully when writing that code because I'm now
stuck here fixing all of this :-/
The fix for the most serious of them is below, and I'd appreciate
any feedback if people spot any problems or holes in that approach.
This will fix bugs that look like an OOPS in fib6_add(), essentially
rt6_select() corrupts the list by making fn->leaf NULL, this is an
illegal state and will cause a crash because the fib6_add() code
(correctly) assumes fn->leaf can never be NULL.
The other bug is that rt6_probe() executes with the per-routing-table
lock held as a reader, but when NDISC is protected by tunneled IPSEC,
this can call right back into the routing code and if that tunneled
IPSEC route lookup COWs a route it will try to take the rwlock as a
writer and deadlock.
There are probably other bugs of this nature lurking in this new code.
commit 4c68db63b8314df3cf30b7fe595a1b8935bb2cb0
Author: David S. Miller <[EMAIL PROTECTED]>
Date: Sat Mar 24 12:06:32 2007 -0700
[IPV6]: Fix routing round-robin locking.
As per RFC2461, section 6.3.6, item #2, when no routers on the
matching list are known to be reachable or probably reachable we
do round robin on those available routes so that we make sure
to probe as many of them as possible to detect when one becomes
reachable faster.
Each routing table has a rwlock protecting the tree and the linked
list of routes at each leaf. The round robin code executes during
lookup and thus with the rwlock taken as a reader. A small local
spinlock tries to provide protection but this does not work at all
for two reasons:
1) The round-robin list manipulation, as coded, goes like this (with
read lock held):
walk routes finding head and tail
spin_lock();
rotate list using head and tail
spin_unlock();
While one thread is rotating the list, another thread can
end up with stale values of head and tail and then proceed
to corrupt the list when it gets the lock. This ends up causing
the OOPS in fib6_add() later onthat many people have been hitting.
2) All the other code paths that run with the rwlock held as
a reader do not expect the list to change on them, they
expect it to remain completely fixed while they hold the
lock in that way.
So, simply stated, it is impossible to implement this correctly using
a manipulation of the list without violating the rwlock locking
semantics.
Reimplement using a per-fib6_node round-robin pointer. This way we
don't need to manipulate the list at all, and since the round-robin
pointer can only ever point to real existing entries we don't need
to perform any locking on the changing of the round-robin pointer
itself. We only need to reset the round-robin pointer to NULL when
the entry it is pointing to is removed.
The idea is from Thomas Graf and it is very similar to how this
was implemented before the advanced router selection code when in.
Signed-off-by: David S. Miller <[EMAIL PROTECTED]>
diff --git a/include/net/ip6_fib.h b/include/net/ip6_fib.h
index 9eda572..cf355a3 100644
--- a/include/net/ip6_fib.h
+++ b/include/net/ip6_fib.h
@@ -58,6 +58,7 @@ struct fib6_node
__u16 fn_bit; /* bit key */
__u16 fn_flags;
__u32 fn_sernum;
+ struct rt6_info *rr_ptr;
};
#ifndef CONFIG_IPV6_SUBTREES
diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c
index f4d7be7..c46f909 100644
--- a/net/ipv6/ip6_fib.c
+++ b/net/ipv6/ip6_fib.c
@@ -1109,6 +1109,10 @@ static void fib6_del_route(struct fib6_node *fn, struct
rt6_info **rtp,
rt6_stats.fib_rt_entries--;
rt6_stats.fib_discarded_routes++;
+ /* Reset round-robin state, if necessary */
+ if (fn->rr_ptr == rt)
+ fn->rr_ptr = NULL;
+
/* Adjust walkers */
read_lock(&fib6_walker_lock);
FOR_WALKERS(w) {
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index a6b3117..3931b33 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -363,55 +363,76 @@ static int rt6_score_route(struct rt6_info *rt, int oif,
return m;
}
-static struct rt6_info *rt6_select(struct rt6_info **head, int oif,
- int strict)
+static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
+ int *mpri, struct rt6_info *match)
{
- struct rt6_info *match = NULL, *last = NULL;
- struct rt6_info *rt, *rt0 = *head;
- u32 metric;
+ int m;
+
+ if (rt6_check_expired(rt))
+ goto out;
+
+ m = rt6_score_route(rt, oif, strict);
+ if (m < 0)
+ goto out;
+
+
