[PATCH nft v2] netfilter: nf_tables: Fix pertential data-race in __nft_flowtable_type_get()

[PATCH nft v2] netfilter: nf_tables: Fix pertential data-race in __nft_flowtable_type_get()

[Ziyang Xuan]

nft_unregister_flowtable_type() within nf_flow_inet_module_exit() can
concurrent with __nft_flowtable_type_get() within nf_tables_newflowtable().
And thhere is not any protection when iterate over nf_tables_flowtables
list in __nft_flowtable_type_get(). Therefore, there is pertential
data-race of nf_tables_flowtables list entry.

Use list_for_each_entry_rcu() to iterate over nf_tables_flowtables list
in __nft_flowtable_type_get(), and use rcu_read_lock() in the caller
nft_flowtable_type_get() to protect the entire type query process.

Fixes: 3b49e2e94e6e ("netfilter: nf_tables: add flow table netlink frontend")
Signed-off-by: Ziyang Xuan <william.xuanziyang@huawei.com>
---
v2:
  - Move rcu_read_lock() outside to caller. [Florian Westphal]
  - Add "Fixes" tag.
---
 net/netfilter/nf_tables_api.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index fd86f2720c9e..47e1a22e8fb1 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -8293,11 +8293,12 @@ static int nft_flowtable_parse_hook(const struct nft_ctx *ctx,
 	return err;
 }
 
+/* call under rcu_read_lock */
 static const struct nf_flowtable_type *__nft_flowtable_type_get(u8 family)
 {
 	const struct nf_flowtable_type *type;
 
-	list_for_each_entry(type, &nf_tables_flowtables, list) {
+	list_for_each_entry_rcu(type, &nf_tables_flowtables, list) {
 		if (family == type->family)
 			return type;
 	}
@@ -8309,9 +8310,13 @@ nft_flowtable_type_get(struct net *net, u8 family)
 {
 	const struct nf_flowtable_type *type;
 
+	rcu_read_lock();
 	type = __nft_flowtable_type_get(family);
-	if (type != NULL && try_module_get(type->owner))
+	if (type != NULL && try_module_get(type->owner)) {
+		rcu_read_unlock();
 		return type;
+	}
+	rcu_read_unlock();
 
 	lockdep_nfnl_nft_mutex_not_held();
 #ifdef CONFIG_MODULES
-- 
2.25.1

Re: [PATCH nft v2] netfilter: nf_tables: Fix pertential data-race in __nft_flowtable_type_get()

[Florian Westphal]

Ziyang Xuan <william.xuanziyang@huawei.com> wrote:
> nft_unregister_flowtable_type() within nf_flow_inet_module_exit() can
> concurrent with __nft_flowtable_type_get() within nf_tables_newflowtable().
> And thhere is not any protection when iterate over nf_tables_flowtables
> list in __nft_flowtable_type_get(). Therefore, there is pertential
> data-race of nf_tables_flowtables list entry.
> 
> Use list_for_each_entry_rcu() to iterate over nf_tables_flowtables list
> in __nft_flowtable_type_get(), and use rcu_read_lock() in the caller
> nft_flowtable_type_get() to protect the entire type query process.

Reviewed-by: Florian Westphal <fw@strlen.de>

Would you be so kind to send followup patches for the other two types
Pablo pointed out?

static LIST_HEAD(nf_tables_expressions);
static LIST_HEAD(nf_tables_objects);

It looks like they have same issue.

Thanks!

Re: [PATCH nft v2] netfilter: nf_tables: Fix pertential data-race in __nft_flowtable_type_get()

[Ziyang Xuan (William)]

> Ziyang Xuan <william.xuanziyang@huawei.com> wrote:
>> nft_unregister_flowtable_type() within nf_flow_inet_module_exit() can
>> concurrent with __nft_flowtable_type_get() within nf_tables_newflowtable().
>> And thhere is not any protection when iterate over nf_tables_flowtables
>> list in __nft_flowtable_type_get(). Therefore, there is pertential
>> data-race of nf_tables_flowtables list entry.
>>
>> Use list_for_each_entry_rcu() to iterate over nf_tables_flowtables list
>> in __nft_flowtable_type_get(), and use rcu_read_lock() in the caller
>> nft_flowtable_type_get() to protect the entire type query process.
> 
> Reviewed-by: Florian Westphal <fw@strlen.de>
> 
> Would you be so kind to send followup patches for the other two types
> Pablo pointed out?
> 
> static LIST_HEAD(nf_tables_expressions);
> static LIST_HEAD(nf_tables_objects);
> 
> It looks like they have same issue.

Yes, I am doing and testing.

Best regards.
> 
> Thanks!
> 
> 
> .
>