nf_conntrack_all() contains an smp_rmb() after spin_unlock_wait().
This was done to allow safe backporting.
Commit 2c6100227116 ("locking/qspinlock: Fix spin_unlock_wait() some more")
(and the commits for the other archs) ensure that spin_unlock_wait()
is an ACQUIRE.
Therefore the smp_rmb() after spin_unlock_wait() can be removed.
Not for stable!
Signed-off-by: Manfred Spraul <[email protected]>
Cc: Pablo Neira Ayuso <[email protected]>
Cc: [email protected]
---
net/netfilter/nf_conntrack_core.c | 5 -----
1 file changed, 5 deletions(-)
diff --git a/net/netfilter/nf_conntrack_core.c
b/net/netfilter/nf_conntrack_core.c
index 3847f09..c8a2d26 100644
--- a/net/netfilter/nf_conntrack_core.c
+++ b/net/netfilter/nf_conntrack_core.c
@@ -153,11 +153,6 @@ static void nf_conntrack_all_lock(void)
for (i = 0; i < CONNTRACK_LOCKS; i++) {
spin_unlock_wait(&nf_conntrack_locks[i]);
}
- /* spin_unlock_wait() is at least a control barrier.
- * Add smp_rmb() to upgrade it to an ACQUIRE, it must
- * pair with the spin_unlock(&nf_conntrack_locks[])
- */
- smp_rmb();
}
static void nf_conntrack_all_unlock(void)
--
2.7.4
