When find_table_lock() returns NULL to t, no error return code of
do_update_counters() is assigned.
To fix this bug, ret is assigned with -ENOENT in this case.
Fixes: 49facff9f925 ("netfilter: ebtables: split update_counters into two functions")
Reported-by: TOTE Robot <[email protected]>
Signed-off-by: Jia-Ju Bai <[email protected]>
---
net/bridge/netfilter/ebtables.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/net/bridge/netfilter/ebtables.c b/net/bridge/netfilter/ebtables.c
index ebe33b60efd6..66c9e4077985 100644
--- a/net/bridge/netfilter/ebtables.c
+++ b/net/bridge/netfilter/ebtables.c
@@ -1256,8 +1256,10 @@ static int do_update_counters(struct net *net, const char *name,
return -ENOMEM;
t = find_table_lock(net, name, &ret, &ebt_mutex);
- if (!t)
+ if (!t) {
+ ret = -ENOENT;
goto free_tmp;
+ }
if (num_counters != t->private->nentries) {
ret = -EINVAL;
--
2.17.1
Jia-Ju Bai <[email protected]> wrote:
> When find_table_lock() returns NULL to t, no error return code of
> do_update_counters() is assigned.
Its -ENOENT.
> t = find_table_lock(net, name, &ret, &ebt_mutex);
^^^^^
ret is passed to find_table_lock, which passes it to
find_inlist_lock_noload() which will set *ret = -ENOENT
for NULL case.
On 2021/3/9 19:01, Florian Westphal wrote:
> Jia-Ju Bai <[email protected]> wrote:
>> When find_table_lock() returns NULL to t, no error return code of
>> do_update_counters() is assigned.
> Its -ENOENT.
>
>> t = find_table_lock(net, name, &ret, &ebt_mutex);
> ^^^^^
>
> ret is passed to find_table_lock, which passes it to
> find_inlist_lock_noload() which will set *ret = -ENOENT
> for NULL case.
Thanks for the reply!
I did not notice "&ret" in find_table_lock()...
I am sorry for the false positive.
Best wishes,
Jia-Ju Bai