On Mon, 2010-08-16 at 12:44 -0700, David Miller wrote:
quoted text
> He's only accessing the per-cpu counter locks of other cpus.
> 
> The per-cpu lock is only locally accessed by a cpu in software
> interrupt context.
> 
> That is why his transformation is legal.
> 
> Lockdep simply hasn't been informed of this invariant and has
> to assume the worst. 

Something like the below will keep lockdep coverage, still going back to
RCU sounds like the best option.

---
 include/linux/netfilter/x_tables.h |    9 +++++++++
 1 files changed, 9 insertions(+), 0 deletions(-)

diff --git a/include/linux/netfilter/x_tables.h b/include/linux/netfilter/x_tables.h
index 24e5d01..a195feb 100644
--- a/include/linux/netfilter/x_tables.h
+++ b/include/linux/netfilter/x_tables.h
@@ -511,12 +511,21 @@ static inline void xt_info_rdunlock_bh(void)
  */
 static inline void xt_info_wrlock(unsigned int cpu)
 {
+#ifdef CONFIG_PROVE_LOCKING
+	/*
+	 * XXX foo
+	 */
+	local_bh_disable();
+#endif
 	spin_lock(&per_cpu(xt_info_locks, cpu).lock);
 }
 
 static inline void xt_info_wrunlock(unsigned int cpu)
 {
 	spin_unlock(&per_cpu(xt_info_locks, cpu).lock);
+#ifdef CONFIG_PROVE_LOCKING
+	local_bh_enable();
+#endif
 }
 
 /*

--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/