On Wed, Feb 19, 2020 at 03:46:27PM +0530, Amol Grover wrote:
quoted hunk ↗ jump to hunk
cn->configs is traversed using list_for_each_entry_rcu
outside an RCU read-side critical section but under the protection
of cn->lock.
Hence, add corresponding lockdep expression to silence false-positive
warnings, and harden RCU lists.
Signed-off-by: Amol Grover <redacted>
---
net/ipv4/netfilter/ipt_CLUSTERIP.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/net/ipv4/netfilter/ipt_CLUSTERIP.c b/net/ipv4/netfilter/ipt_CLUSTERIP.c
index 6bdb1ab8af61..df856ff835b7 100644
--- a/net/ipv4/netfilter/ipt_CLUSTERIP.c
+++ b/net/ipv4/netfilter/ipt_CLUSTERIP.c
@@ -139,7 +139,8 @@ __clusterip_config_find(struct net *net, __be32 clusterip)
struct clusterip_config *c;
struct clusterip_net *cn = clusterip_pernet(net);
- list_for_each_entry_rcu(c, &cn->configs, list) {
+ list_for_each_entry_rcu(c, &cn->configs, list,
+ lockdep_is_held(&cn->lock)) {
bh is disabled before calling __clusterip_config_find(), then
rcu_read_lock_any_held() evaluates true.
Are you sure this really results in a WARN_ON splat?