[PATCH] wifi: zd1211rw: unwind failed interface additions

[PATCH] wifi: zd1211rw: unwind failed interface additions

[Jiale Yao]

zd1211rw stores the vif and interface type before programming the MAC
and receive configuration. If that setup fails, mac80211 does not call
remove_interface and the stale type prevents another interface from
being added.

Clear the vif and type and restore the hardware MAC state before
returning the setup error.

Fixes: 459c51ad6e1f ("zd1211rw: port to mac80211")
Signed-off-by: Jiale Yao <yaojiale02@163.com>
---
 drivers/net/wireless/zydas/zd1211rw/zd_mac.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/net/wireless/zydas/zd1211rw/zd_mac.c b/drivers/net/wireless/zydas/zd1211rw/zd_mac.c
index 03948f0b4628..abbe25623b44 100644
--- a/drivers/net/wireless/zydas/zd1211rw/zd_mac.c
+++ b/drivers/net/wireless/zydas/zd1211rw/zd_mac.c
@@ -1102,6 +1102,7 @@ static int zd_op_add_interface(struct ieee80211_hw *hw,
 				struct ieee80211_vif *vif)
 {
 	struct zd_mac *mac = zd_hw_mac(hw);
+	int ret;
 
 	/* using NL80211_IFTYPE_UNSPECIFIED to indicate no mode selected */
 	if (mac->type != NL80211_IFTYPE_UNSPECIFIED)
@@ -1120,8 +1121,14 @@ static int zd_op_add_interface(struct ieee80211_hw *hw,
 	}
 
 	mac->vif = vif;
+	ret = set_mac_and_bssid(mac);
+	if (ret) {
+		mac->type = NL80211_IFTYPE_UNSPECIFIED;
+		mac->vif = NULL;
+		zd_write_mac_addr(&mac->chip, NULL);
+	}
 
-	return set_mac_and_bssid(mac);
+	return ret;
 }
 
 static void zd_op_remove_interface(struct ieee80211_hw *hw,
-- 
2.34.1

Re: [PATCH] wifi: zd1211rw: unwind failed interface additions

[Johannes Berg]

On Mon, 2026-10-19 at 09:32 +0800, Jiale Yao wrote:
> zd1211rw stores the vif and interface type before programming the MAC
> and receive configuration. If that setup fails, mac80211 does not call
> remove_interface and the stale type prevents another interface from
> being added.
> 
> Clear the vif and type and restore the hardware MAC state before
> returning the setup error.

This is completely crazy. "I have an LLM and it found something that
doesn't make sense to it" is not a reason to send a patch.

Does this actually matter at all? This driver only supports a single
interface anyway - who cares about the hardware state when it's going to
be shut down right away?!

johannes