Skip to contentSkip to conversation

netfilter: nf_tables: fix infinite loop when expr is not available

PATCH Applied Florian Westphal 2 participants 2 messages Mar 5, 2020

Subsystems (2)
lore
Share & tools
lore

Florian Westphal

author maintainer
#0 lore raw

PATCH 1/1 netfilter: nf_tables: fix infinite loop when expr is not available

nft will loop forever if the kernel doesn't support an expression:

1. nft_expr_type_get() appends the family specific name to the module list.
2. -EAGAIN is returned to nfnetlink, nfnetlink calls abort path.
3. abort path sets ->done to true and calls request_module for the expression.
4. nfnetlink replays the batch, we end up in nft_expr_type_get() again.
5. nft_expr_type_get attempts to append family-specific name. This one already exists on the list, so we continue
6. nft_expr_type_get adds the generic expression name to the module list. -EAGAIN is returned, nfnetlink calls abort path.
7. abort path encounters the family-specific expression which has 'done' set, so it gets removed.
8. abort path requests the generic expression name, sets done to true.
9. batch is replayed.

If the expression could not be loaded, then we will end up back at 1), because the family-specific name got removed and the cycle starts again.

Note that userspace can SIGKILL the nft process to stop the cycle, but the desired behaviour is to return an error after the generic expr name fails to load the expression.

Fixes eb014de4fd418 ("netfilter: nf_tables: autoload modules from the abort path")
---
 net/netfilter/nf_tables_api.c | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)
net/netfilter/nf_tables_api.c +3 -7
index bb064aa4154b..798a20648c2f 100644
@@ -7383,13 +7383,8 @@ static void nf_tables_module_autoload(struct net *net)
 	list_splice_init(&net->nft.module_list, &module_list);
 	mutex_unlock(&net->nft.commit_mutex);
 	list_for_each_entry_safe(req, next, &module_list, list) {
-		if (req->done) {
-			list_del(&req->list);
-			kfree(req);
-		} else {
-			request_module("%s", req->module);
-			req->done = true;
-		}
+		request_module("%s", req->module);
+		req->done = true;
 	}
 	mutex_lock(&net->nft.commit_mutex);
 	list_splice(&module_list, &net->nft.module_list);
@@ -8172,6 +8167,7 @@ static void __net_exit nf_tables_exit_net(struct net *net)
 	__nft_release_tables(net);
 	mutex_unlock(&net->nft.commit_mutex);
 	WARN_ON_ONCE(!list_empty(&net->nft.tables));
+	WARN_ON_ONCE(!list_empty(&net->nft.module_list));
 }
 
 static struct pernet_operations nf_tables_net_ops = {
Signature · 1 line
2.24.1
maintainer Applied
#1 lore raw
Quoted 16 lines · Florian Westphal

nft will loop forever if the kernel doesn't support an expression:

1. nft_expr_type_get() appends the family specific name to the module list.
2. -EAGAIN is returned to nfnetlink, nfnetlink calls abort path.
3. abort path sets ->done to true and calls request_module for the expression.
4. nfnetlink replays the batch, we end up in nft_expr_type_get() again.
5. nft_expr_type_get attempts to append family-specific name. This one already exists on the list, so we continue
6. nft_expr_type_get adds the generic expression name to the module list. -EAGAIN is returned, nfnetlink calls abort path.
7. abort path encounters the family-specific expression which has 'done' set, so it gets removed.
8. abort path requests the generic expression name, sets done to true.
9. batch is replayed.

Quoted 6 lines

If the expression could not be loaded, then we will end up back at 1), because the family-specific name got removed and the cycle starts again.

Note that userspace can SIGKILL the nft process to stop the cycle, but the desired behaviour is to return an error after the generic expr name fails to load the expression.

Applied, thanks Florian.