diff options
| author | Maoyi Xie <maoyixie.tju@gmail.com> | 2026-05-27 21:33:58 +0800 |
|---|---|---|
| committer | Johannes Berg <johannes.berg@intel.com> | 2026-05-28 09:49:48 +0200 |
| commit | f681502c79173a79c3de16be274eca83e8fd8d3f (patch) | |
| tree | 8f7f4b6d499dfed962ec85798b2f1d5a43404123 /net/wireless | |
| parent | db215112182e3eddf2a551e25a7dfd23aca3a7fc (diff) | |
wifi: nl80211: re-check wiphy netns in testmode and vendor dump continuations
Commit 79240f3f6d76 ("wifi: nl80211: re-check wiphy netns in
nl80211_prepare_wdev_dump() continuation") fixed one dumpit path that
looked the wiphy up by index on a later call without confirming it was
still in the caller's netns. Two more dumpit paths have the same gap.
nl80211_testmode_dump() and nl80211_prepare_vendor_dump() both keep the
wiphy index in cb->args[] and look it up again on later calls, through
cfg80211_rdev_by_wiphy_idx() and wiphy_idx_to_wiphy(). The first call
binds to the caller's netns. A later call does not check it again. In
between, the wiphy can move to another netns via
NL80211_CMD_SET_WIPHY_NETNS.
Add the same net_eq() check to both. On a mismatch, return -ENODEV and
the dump ends.
No mainline driver registers .testmode_dump or
wiphy_vendor_command.dumpit, so these paths are not reachable today.
Drivers outside the tree can register either.
Signed-off-by: Maoyi Xie <maoyixie.tju@gmail.com>
Link: https://patch.msgid.link/20260527133358.2853238-1-maoyixie.tju@gmail.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Diffstat (limited to 'net/wireless')
| -rw-r--r-- | net/wireless/nl80211.c | 19 |
1 files changed, 19 insertions, 0 deletions
diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c index c272a2fbad03..cdb5e9b77143 100644 --- a/net/wireless/nl80211.c +++ b/net/wireless/nl80211.c @@ -13730,6 +13730,16 @@ static int nl80211_testmode_dump(struct sk_buff *skb, err = -ENOENT; goto out_err; } + + /* + * The wiphy may have moved netns between dumpit + * invocations (via NL80211_CMD_SET_WIPHY_NETNS), so + * re-check that it still matches the caller's netns. + */ + if (!net_eq(wiphy_net(&rdev->wiphy), sock_net(skb->sk))) { + err = -ENODEV; + goto out_err; + } } else { attrbuf = kzalloc_objs(*attrbuf, NUM_NL80211_ATTR); if (!attrbuf) { @@ -17771,6 +17781,15 @@ static int nl80211_prepare_vendor_dump(struct sk_buff *skb, if (!wiphy) return -ENODEV; + + /* + * The wiphy may have moved netns between dumpit + * invocations (via NL80211_CMD_SET_WIPHY_NETNS), so + * re-check that it still matches the caller's netns. + */ + if (!net_eq(wiphy_net(wiphy), sock_net(skb->sk))) + return -ENODEV; + *rdev = wiphy_to_rdev(wiphy); *wdev = NULL; |